The Acquisition of Verb + Noun Collocations by Japanese Learners of English

Home , English collocations, English verbs

INVESTIGATING ESL LEARNERS’ LEXICAL COLLOCATIONS: THE ACQUISITION OF VERB + NOUN COLLOCATIONS BY JAPANESE LEARNERS OF ENGLISH

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

LINGUISTICS

MAY 2009

By Tomoko Miyakoshi

Dissertation Committee:

Ann M. Peters, Chairperson Robert Bley-Vroman Patricia A. Lee Robert E. Gibson Richard J. Guillory

iii

ACKNOWLEDGMENTS

I would like to express my gratitude to my teachers and friends who inspired and

encouraged me throughout the period of my doctoral study. First and foremost, I would

like to thank my chair Ann Peters. Ann has been my advisor and a wonderful mentor

since I started my doctorate study at the University of Hawai‘i at Mānoa. Her pioneering views on the units of language acquisition have inspired me to investigate the development of collocational knowledge by second language learners. She is open- minded, energetic, and extremely patient. She carefully read all my drafts and conference abstracts, spent a lot of time analyzing and discussing my data and encouraged me to consider in various view points. Thanks to her guidance and support, my journey to the doctorate was very rewarding and successful.

Next my thanks go to Robert Bley-Vroman. I am grateful to Robert for teaching me second language acquisition theories and corpus linguistics. He always helped my progres by giving me valuable comments on my experimental designs, research materials and instruction methods. His comments and advice often led me to reconsider my methodology and encouraged me to improve my teaching methods. I am very thankful to

Patricia Lee for introducing me to the study of semantics and pragmatics. Pat is very enthusiastic, friendly, has a great sense of humor. It was a great joy for me to broaden my knowledge by taking her stimulating courses and I truly enjoyed having discussions with

her. I would like to thank Robert Gibson for giving me thoughtful comments and

constant encouragement. I was very happy to receive feedback from Robert who is

proficient in Japanese and is very open to both practical and theoretical points of views. I

am very thankful to Richard Guillory for being supportive and for sharing his insights

from a more outside point of view.

I would like to extend my special thanks to John Haig, Dina Yoshimi, Richard

Schmidt, James Brown, Zoya Proshina, Liubov’ Konstanovna, Yasuko Ito and Mari

Miyao for discussing the ideas in my research at various stages. I am grateful to Laurie

Durand for editing my drafts. I would like to express my gratitude to all the participants

in my experiments of this study. I am very thankful to Hiroko Sato, Mai Takemoto and

Satoko Kimura for their kind cooperation in recruiting subjects for my experiments.

I thank the Department of Linguistics of the University of Hawai‘i at Mānoa for generous financial support during my years at UH. Additional support was provided by the East-West Center, Arts and Science Advisory Council and Graduate Student

Organization (both at UH).

I would like to thank all my friends, inside and outside the University of Hawai‘i, for their support and friendship: Emily Bartelson, Jin-Sun Choe, Yumiko Enyo, Toshiaki

Furukawa, Yukie Hara, Ryoko Hattori, Elena Indjieva, Tatjana Ilic, Aya Inoue, Kyuseek

Hwang, Jason Jackson, Sang-Gu Kang, On-Soon Lee, Tsai-Hsiu Liu (Dorinda), Jun

Nomura, Manami Sato, Yoongsang Song, Diana Stojanovic, Jennie Tran, Kaori Ueki, and

Qinglian Zhao. And I am very thankful to Seiichi Nakada, my undergraduate advisor, my first teacher of linguistics and an alumnus of UH linguistics, for his strong encouragement that made me decide to come to Hawai‘i.

Finally, my greatest thanks are reserved for my family who has given me the opportunity to pursue my academic career. Especially, I would like to thank my parents

Kazuhiro and Yoko Miyakoshi for their unconditional and understanding support and my sister Kyoko Momiyama for her warm-hearted encouragement.

ABSTRACT

Although it is widely acknowledged that collocations play an important part in

second language learning, especially at intermediate-advanced levels, learners’

difficulties with collocations have not been investigated in much detail so far. The

present study examines ESL learners’ use of verb-noun collocations, such as take notes,

place an order, cut corners and make a discovery, and the effects of instructions which

direct learners’ attention to input and to restrictions of combinations.

Sixty Japanese students (30 intermediate, 30 advanced) took fill-in-the-blank tests

followed by one session of instruction, involving a brief introduction to collocations and

a discussion of common mistakes with collocations and differences in the collocational restrictions between English and Japanese. At home the subjects studied collocations using paper-based exercises provided after the pretest and/or online flashcards. Within two weeks after the pretest and instruction, the subjects came back to take a second fill- in-the blank test which served as the posttest.

Statistical analyses show significant influences of various attributes, including overall frequency, literal vs. abstract meaning, the existence of L1 equivalents, and the presence of light vs. content verbs.

The following eleven error types were identified: (1) Inappropriate paraphrases;

(2) Misuse of light verbs; (3) Interference of the native language Japanese; (4) Blending two collocations with similar meaning; (5) Mistakes by using morphological synonymy;

(6) Use of words other than verbs; (7) Inserting unnecessary articles, particles and prepositions between verbs and nouns; (8) Mistake in distinguishing intransitive and

vi transitive verbs; (9) Creating collocations from compound nouns; (10) Misunderstanding actor-patient relations of verbs; and (11) Phonological errors. Among these error types, it was found that paraphrases, misuse of light verbs and interference of the native language are the strongest indicators of difficulty of collocations for the learners.

Significant improvements were observed in the learners’ performance in the posttest. These findings highlight the efficacy of improving second language learners’ collocational knowledge to enhance their proficiency in the target language, and show that explicit instruction using learners’ selective attention to input indeed improves their collocational competence in the target language.

vii

TABLE OF CONTENTS

Acknowledgments………………………………………………………………………..iv Abstract…………………………………………………………………………………...vi List of tables……………………………………………………………………………....xi List of figures…………………………………………………………………………….xv

Chapter1:Introduction……………………....………………………...……...……1 1.1. Collocations…………………………………………………….………….1 1.2. Studies on acquisition of collocations…………………………………...7 1.2.1. Research on collocations in first language acquisition...…...... …7 1.2.2. Research on collocations in second language learning……….....13 1.2.3. Functions of collocations in second language acquisition…..….18 1.3. Purpose of this study: What factors may affect the acquisition of lexical collocations?…………………………………………………...…………..22

Chapter 2: Preliminary survey: Categorization of L2 learners’ problems in acquiring collocations………………………………………………………………………….….28 2.1. Purpose of the survey and research questions……………………………28 2.2. Method.……………………………………………………………….….29 2.2.1. Subjects……………...……………………………….……...….29 2.2.2. Stimuli and procedure………………………………….………..29 2.3. Results……………………………………………………….…………...30 2.4. Summary: Implication for the experiments……………………………..….38

Chapter 3: The roles of frequency and meaning in acquiring Verb + Noun collocations.41 3.1. Purpose of frequency and meaning studies: research questions and hypotheses……...………………………………………………………….….41 3.2. General method………………………………….………………..….….44 3.2.1. Subjects……………………………………………………...... ….44 3.2.2. Stimuli and procedure…………………………………..….……45

Chapter 4: Experiment 1: The role of frequency in acquiring Verb + Noun collocations.49 4.1. Stimuli.………………………..……………………………………….49 4.2. Overall results and single variable analyses of results for the number of collocations produced by Japanese learners………………………….…….50 4.3. Native speakers’ Judgments of Grammaticality…………...….….…....57 4.3.1. Average scores of native speakers’ Grammaticality Judgments……..57 4.3.2. Three-scale analysis of Grammaticality Judgments……………....62 4.3.3. Two-scale analysis of Grammaticality Judgments…………...... …67 4.3.4. Hierarchy of difficulty based on Grammaticality Judgments….…….73 4.4. Adequacy of meaning conveyed……………………………………....…....75 4.4.1. Average scores of native speakers’ Meaning Conveyed….………….75 4.4.2. Three-scale analysis of Meaning Conveyed……...... …81

viii

4.4.3. Two-scale analysis of Meaning Conveyed………….……...... …..…85 4.4.4. Hierarchy of difficulty based on Meaning Conveyed……………….91 4.5. Learners’ confidence in their own productions………..………….………...92 4.6. Learners’ familiarity with their own productions……...…....….…………...97 4.7. Multi-variable analyses of results……………………...…....….……...…..103 4.7.1. Grammaticality Judgments vs. learners’ confidence………...... ….103 4.7.2. Meaning Conveyed vs. Learners’ confidence…………..…...... ….106 4.7.3. Learners’ confidence vs. Heard or Read……………………....….109 4.8. Discussion of Experiment 1…………………………………………..113 4.8.1. Number of collocations produced…………….………….………114 4.8.2. Observed frequency………………...…………………………..…..115 4.8.3. With vs. without L1 equivalents…….………………………..….…116 4.8.4. Light verbs vs. content verbs………….……………………...... …..117 4.8.5. Same vs. different light verbs………………..…………….…….…118 4.8.6.Learners’ confidence and familiarity……………………………..…118

Chapter 5: Experiment 2: The role of meaning in acquiring Verb + Noun collocations.121 5.1. Stimuli…………….…………………………………………………..…...121 5.2. Overall results and single variable analyses of results for the number of collocations produced by Japanese learners………..…………………...... ….122 5.3. Native speakers’ Judgments of Grammaticality…….….……………….…128 5.3.1. The average scores of native’s Grammaticality Judgments……...... 128 5.3.2. Three-scale analysis of Grammaticality Judgments……………..….134 5.3.3. Two-scale analysis of Grammaticality Judgments……………….…139 5.3.4. Hierarchy of difficulty based on Grammaticality Judgments……....145 5.4. Adequacy of meaning conveyed…………………………………………...146 5.4.1. Average scores of native speakers’ Meaning Conveyed…………....146 5.4.2. Three-scale analysis of Meaning Conveyed………………………..152 5.4.3. Two-scale analysis of Meaning Conveyed………………………....157 5.4.4. Hierarchy of difficulty based on Meaning Conveyed…………...…162 5.5. Learners’ confidence in their own productions…………………………....163 5.6. Learners’ familiarity with their own productions……………………….....169 5.7. Multi-variable analyses of results……………………………………….…175 5.7.1. Grammaticality Judgments vs. learners’ confidence…………….…175 5.7.2. Meaning Conveyed vs. Learners’ confidence……………………....177 5.7.3. Learners’ confidence vs. Heard or Read…………………….……...180 5.8. Discussion of Experiment 2…………………………………………….….184 5.8.1. Number of collocations produced…………………………………..185 5.8.2. Literal vs. abstract meaning………………………………………...187 5.8.3. With vs. without L1 equivalents…………………………………....188 5.8.4. Light verbs vs. content verbs……………………………………….189 5.8.5. Same vs. different light verbs…………………………………...….190 5.8.6. Learners’ confidence and familiarity……………………………….191 5.9. General discussion of Experiments 1 and 2………………………………..192 5.9.1. Support for predictions……………………………………………..192 5.9.2. Other remarks……………………………………………………….198

5.9.2.1. Types of errors…………………………………………………198 5.9.2.2. L1 transfer……………………………………………………..201

Chapter 6: The effects of Instruction in acquiring Verb + Noun collocations……….…205 6.1. Purpose of the experiments, research questions and hypotheses………….205 6.2. Method…………………………………………………………………….207 6.2.1. Subjects……………………….…………………………………….207 6.2.2. Stimuli and procedure………………………………………………207 6.3. Experiment 3 (The post test for Experiment 1): The role of frequency in acquiring Verb + Noun collocations……………………….…………210 6.3.1. Results for Experiment 3…………………………………………211 6.4. Experiment 4 (The post test for Experiment 2: The role of meaning in acquiring Verb + Noun collocations…………………………………...…221 6.4.1. Results for Experiment 4…………………………………………...222 6.5. Studying for the posttests………………………………………………….233 6.5.1. Materials learners used, amount of time they studied and their improvement………………………………………………………..233 6.6. Other factors……………………………………………………….………237 6.7. General discussion of Experiments 3 and 4……………………………….241

Chapter 7: Summary and conclusion: findings, implications and further directions…..251 7.1. Summary…………………………………………………………………..251 7.2. Suggestions for future research……………………………………………252

Appendix I. Preliminary survey questionnaires……………………………………...257 Appendix II. Data from the preliminary survey……………………………………....260 Appendix III. Test sentences for Experiment 1………………………………………..267 Appendix IV. Questionnaire before the pretest (Experiments 1 and 2)…………..……270 Appendix V. Data from Experiment 1………………………………………………...271 Appendix VI. Tables from Experiment 1………………………………………………295 Appendix VII. Test sentences for Experiment 2………………………………………..302 Appendix VIII. Data from Experiment 2………………………………………….……305 Appendix IX. Tables from Experiment 2………………………………………..……..322 Appendix X. Instruction handout……………………………………………………..329 Appendix XI. Take-home exercises…………………………………………………....361 Appendix XII. Questionnaire before the posttest (Experiments 3and 4)……………….372 Appendix XIII. Test sentences for Experiment 3……………………………………….373 Appendix XIV. Data from Experiment 3……………………………………………….376 Appendix XV. Tables from Experiment 3……………………………………………....383 Appendix XVI. Test sentences for Experiment 4……………………………………....384 Appendix XVII. Data from Experiment 4…………………………………………...…387 Appendix XVIII. Tables from Experiment 4…………………………………………...393

References………………………………………………………………………………394

LIST OF TABLES

Table 1-1. Examples of collocations. Source: Benson (1985)……………….…….…7 Table 1-2. The developmental stages of interrogative chunk comment t’appelles-tu? (Myles et al., 1999: 67)…………………………………………………………………..19 Table 1-3. The developmental stages of the formula I don’t know (Ellis, 1984)…....20 Table 1-4. The development of a chunk involving I wanna by an Estonian child acquiring English (The data from Vihman [1982: 277-278])……………………...... 21

Table 2-A. Types of collocations………………….……………………………...….31

Table 3-1. Eight light verbs investigated in this study……………………………....41 Table 3-2. Three collocational types…………………………………….…….…….42

Table 4-1. Six collocational types for Experiment 1………………………..…….…50 Table 4-A. Number of responses produced by the subjects in Experiment 1……..…51 Table 4-B. Number of responses produced by the subjects for frequent vs. infrequent collocations of Types A1, A2 and A3………….………………………………..….…….52 Table 4-C. Number of responses produced by the subjects for collocations with vs. without equivalent collocations in Japanese for Types A1, A2, and A3…………….…...53 Table 4-D. Number of responses produced by the subjects for collocations with light verbs vs. content verbs of Types A1, A2, and A3…………………………..…………....54 Table 4-E. Number of responses produced by the subjects for collocations with same light verbs vs. different verbs…………………………………………………………….55 Table 4-F. Native speakers’ Grammaticality Judgments. Experiment 1 (6-point scale)……………………………………………………………………………………..58 Table 4-G. Native speakers’ Grammaticality Judgments for frequent vs. infrequent collocations. (6-point scale)……………………………………………………………...59 Table 4-H. Native speakers’ Grammaticality Judgments for collocations with vs. without L1 equivalents. (6-point scale)……………………………………………….…59 Table 4-I. Native speakers’ Grammaticality Judgments for collocations with light verbs vs. content verbs. (6-point scale)……………………………………………...…61 Table 4-J. Native speakers’ Grammaticality Judgments for collocations with same vs. different light verbs. (6-point scale)……………………………………………………...61 Table 4-K. Grammaticality Judgments. Frequent vs. infrequent collocations……….68 Table 4-L. Grammaticality Judgments. Frequent vs. infrequent collocations……….69 Table 4-M. Grammaticality Judgments. Light verb + noun vs. content verb + noun……………………………………………………………………………………...71 Table 4-N. Grammaticality Judgments. Same light verb + noun vs. different verb + noun……………………………………………………………………………………...72 Table 4-2. Difficulty hierarchy of collocations…………………………………...…74 Table 4-O. Native speakers’ judgments of Meaning Conveyed. Experiment 1 (6-point- scale)…………………………………………………………………………………..…76

Table 4-P. Native speakers’ judgments of Meaning Conveyed: Frequent vs. infrequent collocations. (6-point-scale)…………………………………………..…77 Table 4-Q. Native speakers’ judgments of Meaning Conveyed: Collocations with vs. without Japanese equivalents. (6-point-scale)…………………………………………...78 Table 4-R. Native speakers’ judgments of Meaning Conveyed for ‘Light verb + noun’ vs. ‘Content verb + noun’ collocations (6-point-scale)………………………………….79 Table 4-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs. (6-point-scale)…………………………………………...79 Table 4-T. Meaning Conveyed. Frequent vs. infrequent collocations……………..86 Table 4-U. Meaning Conveyed. Collocations with vs. without L1 equivalents……..87 Table 4-V. Meaning Conveyed. Light verb + noun vs. content verb + noun……..88 Table 4-W. Meaning Conveyed. Same light verb + noun vs. different light verb + noun…………………………………………………………………………………..…89 Table 4-3. Difficulty hierarchy of collocations……………………………………..91 Table 4-X. Average confidence scores for each collocation type: A1, A2, A3 (10- point-scale)……………………………………………………………………………….93 Table 4-Y. Average confidence scores for frequent vs. infrequent collocations. (10- point-scale)……………………………………………………………………………….94 Table 4-Z. Average confidence scores: with vs. without L1 equivalents. (10-point- scale)…………………………………………………………………………………..…95 Table 4-AA. Average confidence scores: Light verb + noun vs. content verb + noun. (10-point-scale)…………………………………………………………………………..95 Table 4-BB. Average confidence scores: same vs. different light verb. (10-point- scale)……………………………………………………………………………………..96 Table 4-CC. Average familiarity scores for combinations A1, A2 and A3. (10-point scale)……………………………………………………………………………………..98 Table 4-DD. Average familiarity for frequent vs. infrequent collocations. (10-point- scale)……………………………………………………………………………………..99 Table 4-EE. Average familiarity scores for collocations with vs. without L1 equivalents. (10-point-scale)…………………………………………………………………………100 Table 4-FF. Average familiarity scores for combinations using Light vs. Content verbs (10-point-scale)………………………………………………………………………....100 Table 4-GG. Average familiarity scores for Same vs. Different light verb collocations. (10-point-scale)…………………………………………………………………………101 Table 4-4. Easy vs. difficult types of collocations…………………………………113

Table 5-1. The six collocational types for Experiment 2………………………..…122 Table 5-A. Number of responses produced by the subjects for Experiment 2……..122 Table 5-B. Number of responses produced by the subjects for literal vs. abstract collocations……………………………………………………………………………..124 Table 5-C. Number of responses produced by the subjects for collocations with vs. without L1 equivalents. A1, A2, A3………………………………………………….…125 Table 5-D. Number of responses produced by the subjects for collocations with light verbs vs. content verbs………………………………………………………………….126 Table 5-E. Number of responses produced by the subjects for collocations with same vs. different light verbs…………………………………………………………………127

xii

Table 5-F. Native speakers’ Grammaticality Judgment in Experiment 2. (6-point- scale)…………………………………………………………………………………....129 Table 5-G. Native speakers’ Grammaticality Judgments: Literal vs. abstract collocations. (6-point-scale)…………………………………………………………….130 Table 5-H. Native speakers’ Grammatical Judgments: collocations with vs. without L1 equivalents. (6-point-scale)…………………………………………………………131 Table 5-I. Native speakers’ Grammatical Judgments for collocations with light verbs vs. content verbs. (6-point-scale)……………………………………………………….132 Table 5-J. Native speakers’ Grammatical Judgment for collocations with same vs. different light verbs. (6-point-scale)…………………………………………………....140 Table 5-K. Grammaticality Judgments: Collocations with literal vs. abstract meanings……………………………………………………………………………….142 Table 5-L. Grammaticality Judgments: Collocations with vs. without L1 equivalents……………………………………………………………………………..142 Table 5-M. Grammaticality Judgments. Light verb + Noun vs. Content verb + Noun……………………………………………………………………………………143 Table 5-N. Grammaticality Judgments: Same vs. Different light verb + Noun…...144 Table 5-2. Difficulty hierarchy of collocations…………………………………….146 Table 5-O. Native speakers’ judgments of Meaning Conveyed in Experiment 2. (6- point-scale)……………………………………………………………………………...147 Table 5-P. Native speakers’ judgments of Meaning Conveyed in Experiment 2. Literal vs. abstract collocations. (6-point-scale)………………………………………..148 Table 5-Q. Native speakers’ grading of Meaning Conveyed: Collocations with vs. without translation equivalent in Japanese and English. (6-point-scale)……………….149 Table 5-R. Native speakers’ judgments of Meaning Conveyed: Collocations with light verbs vs. content verbs. (6-point-scale)………………………………………………...150 Table 5-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs. (6-point-scale)……………………………………….…150 Table 5-T. Meaning Conveyed. Literal vs. abstract collocations………………...157 Table 5-U. Meaning Conveyed. Collocations with vs. without L1 equivalents….159 Table 5-V. Meaning Conveyed. Light verbs vs. content verbs…………………...160 Table 5-W. Meaning Conveyed. Same vs. different light verb…………………...161 Table 5-3. Difficulty hierarchy of collocations………………………………….…163 Table 5-X. Average confidence scores for collocation types: A1, A2, A3 (10-point- scale). Experiment 2…………………………………………………………………….164 Table 5-Y. Average confidence for literal vs. abstract collocations. (10-point scale)……………………………………………………………………………………165 Table 5-Z. Average confidence scores: collocations with vs. without L1 equivalents. (10-point-scale)………………………………………………………………………....166 Table 5-AA. Average confidence scores: light verb vs. content verb. (10-point- scale)…………………………………………………………………………………....166 Table 5-BB. Average confidence scores: same vs. different light verb. (10-point- scale)……………………………………………………………………………………167 Table 5-CC. Average familiarity scores for each collocation type: A1, A2, A3 (10- point-scale). Experiment 2……………………………………………………………..169 Table 5-DD. Average familiarity ratings: Literal vs. abstract collocations. (10-point-

xiii scale)……………………………………………………………………………………170 Table 5-EE. Average familiarity scores: collocations with vs. without L1 equivalent. (10-point-scale)………………………………………………………………………....171 Table 5-FF. Average familiarity scores: light vs. content verb. (10-point-scale)…172 Table 5-GG. Average familiarity scores: same vs. different light verb. (10-point- scale)…………………………………………………………………………..173 Table 5-4. Easy vs. difficult types of collocations…………………………..….184 Table 5-5. Number of transfer errors in each subtype………………..…..202-203

Table 6-A. Summary of tokens and percentages of accurate collocations in Experiments 1 and 3………………………………………………………………….…213 Table 6-B. Tokens and percentages of accurate collocations in Experiments 1 and 3. Frequent vs. infrequent types…………………………………………………………...214 Table 6-C. Tokens and percentages of accurate collocations in Experiments 1 and 3. With vs. without L1 equivalents………………………………………………………..216 Table 6-D. Tokens and percentages of accurate collocations in Experiments 1 and 3. Light verbs vs. content verbs…………………………………………………………...217 Table 6-E. Tokens and percentages of accurate collocations in Experiments 1 and 3. Same light verbs vs. different light verbs………………………………………………219 Table 6-1. Subtypes where significantly greater improvement was observed for more difficult types: Experiment 3………………………………………………………...…221 Table 6-F. The summary of tokens and percentages of accurate collocations in Experiments 2 and 4…………………………………………………………………….224 Table 6-G. Tokens and percentages of accurate productions in Experiments 2 and 4. Literal vs. abstract……………………………………………………………………....226 Table 6-H. Tokens and percentages of accurate collocations in Experiments 2 and 4: with vs. without L1 equivalents……………………………………………………...…227 Table 6-I. Tokens and percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs………………………………………………………...…229 Table 6-J. Tokens and percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs………………………………………………………...…231 Table 6-2. Subtypes where significantly greater improvement was observed for more difficult types: Experiment 4………………………………………………………...…233 Table 6-3. Numbers of learners who did each exercise and average amount of time spent for each exercise……………………………………………………………….....234

xiv

LIST OF FIGURES

Figure 1-1. Lewis’ model of continuum (Lewis, 2000: 74)……………………...…...3 Figure 1-2. Howarth’s model of continuum (Howarth 1998: 28)………………….....6 Figure 1-3. The balance of holistic and analytic processing from birth to adulthood..9

Figure 2-A. Types of L2 collocations……………………………………………..31 Figure 2-1. Grammaticality Judgment (5-point-scale): Paraphrasing…….……....35 Figure 2-2. Meaning Conveyed (5-point-scale): Paraphrasing……………………....36 Figure 2-3. Grammaticality Judgment (5-point-scale): Misuse of light verbs……….37 Figure 2-4. Meaning Conveyed (5-point-scale): Misuse of light verbs……………...37

Figure 4-A. Number of combinations produced by subjects………………………….51 Figure 4-B. Average number of combinations of frequent vs. infrequent collocations of Types A1, A2, and A3……………………………………………………………………52 Figure 4-C. Average number of combinations with vs. without equivalent collocations in Japanese for Types A1, A2 and A3……………………………………………………53 Figure 4-D. Average number of combinations of light verb + noun collocations and Content verb + Noun collocations of Types A1, A2, and A3…………………………….55 Figure 4-E. Average number of combinations of same vs. different light verb + noun for Types A1, A2 and A3………………………………………………………………....56 Figure 4-F. Native speakers’ Grammaticality Judgments (6-point scale)…………….58 Figure 4-G. Native speakers’ Grammaticality Judgments for frequent vs. infrequent collocations (6-point scale)…………………………………………………………...….59 Figure 4-H. Native speakers’ Grammaticality Judgments for collocations with vs. without L1 equivalents. (6-point scale)…………………………………………………60 Figure 4-I. Native speakers’ Grammaticality Judgments for light verb + noun vs. content verb + noun collocations (6-point scale)……………………………………….61 Figure 4-J. Native speakers’ Grammaticality Judgments for collocations with same vs. different light verbs. (6-point scale)……………………………………………………...62 Figure 4-1. Native speakers’ Grammatical Judgment A1 (6-point-scale)…………....63 Figure 4-2. Native speakers’ Grammatical Judgment A2 (6-point-scale)………...….64 Figure 4-3. Native speakers’ Grammatical Judgment A3 (6-point-scale)…………....64 Figure 4-4. Native speakers’ Grammatical Judgment. Total. A1, A2, A3…………...65 Figure 4-5. Grammaticality Judgments. Frequent vs. infrequent collocations………66 Figure 4-6. Grammaticality Judgments. With vs. without equivalent collocations in L1……………………………………………………………………………………...…66 Figure 4-7. Grammaticality Judgments. Light verb + noun vs. content verb + noun……………………………………………………………………………………...66 Figure 4-8. Grammaticality Judgment. Same light verb + noun vs. different light verb + noun……………………………………………………………………………...67 Figure 4-K. Grammaticality Judgment. Frequent vs. infrequent collocations………..68 Figure 4-L. Grammaticality Judgment. Collocations with vs. without equivalent

collocations in L1……………………………………………………………………..…70 Figure 4-M. Grammaticality Judgment. Light verb + noun collocations vs. content verb + Noun collocations……………………………………………………………...…71 Figure 4-N. Grammaticality Judgments. Same light verb + noun vs. different light verb + noun…………………………………………………………………………...…72 Figure 4-O. Native speakers’ judgments of Meaning Conveyed (6-point-scale)….….76 Figure 4-P. Native speakers’ judgments of Meaning Conveyed: Frequent vs. infrequent collocations. (6-point-scale)………………………….….…………………..77 Figure 4-Q. Native speakers’ judgments of Meaning Conveyed: Collocations with vs. without Japanese equivalents. (6-point-scale)…………………….……………………..78 Figure 4-R. Native speakers’ judgments of Meaning Conveyed for ‘light verb + noun’ vs. ‘content verb + noun’ collocations (6-point-scale)………………………………….79 Figure 4-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs (6-point-scale)…………………………………………....80 Figure 4-9. A1 Native speakers’ grading of Meaning Conveyed…………………....81 Figure 4-10. A2 Native speakers’ grading of Meaning Conveyed…………………....82 Figure 4-11. A3 Native speakers’ grading of Meaning Conveyed…………………….82 Figure 4-12. Meaning Conveyed. Total. A1, A2, A3…………………………………..83 Figure 4-13. Meaning Conveyed. Frequent vs. infrequent collocations……………….84 Figure 4-14. Meaning Conveyed. With vs. without equivalent collocations in L1...…84 Figure 4-15. Meaning Conveyed. Light verb + noun vs. content verb + noun……...84 Figure 4-16. Meaning Conveyed. Same light verb + noun vs. different light verb + noun……………………………………………………………………………………...85 Figure 4-T. Meaning Conveyed. Frequent vs. infrequent collocations……………….86 Figure 4-U. Meaning Conveyed. Collocations with vs. without L1 equivalents……..87 Figure 4-V. Meaning Conveyed. Light verb + noun vs. content verb + Noun………88 Figure 4-W. Meaning Conveyed. Same light verb + noun vs. different light verb + noun…………………………………………………………………………………...... 90 Figure 4-X. Average confidence scores for each collocation type (10-point-scale)..…93 Figure 4-Y. Average scores of learners’ confidence: frequent vs. infrequent collocations. (10-point-scale)…………………………………………………………….94 Figure 4-Z. Average confidence scores: with vs. without L1 equivalents. (10-point- scale)……………………………………………………………………………………..95 Figure 4-AA. Average confidence scores: light verb + noun vs. content verb + noun. (10- point-scale)…………………………………………………………………………..96 Figure 4-BB. Average confidence scores: same vs. different light verb. (10-point- scale)……………………………………………………………………………………..97 Figure 4-CC. Average familiarity scores for combinations: A1, A2 and A3. (10-point- scale)……………………………………………………………………………………..98 Figure 4-DD. Average familiarity scores for frequent vs. infrequent collocations. (10- point-scale)……………………………………………………………………………….99 Figure 4-EE. Average scores of learners’ familiarity with collocations with vs. without L1 equivalents. (10-point-scale)………………………………………………………..100 Figure 4-FF. Average familiarity scores for combinations using light vs. content verbs (10-point-scale)…………………………………………………………………………101 Figure 4-GG. Average familiarity scores for same vs. different light verb collocations.

xvi

(10-point-scale)…………………………………………………………………………102 Figure 4-17. Grammaticality when learners were vs. were not confident in their productions……………………………………………………………………………..104 Figure 4-18. Learners’ confidence when their productions were judged grammatical vs. ungrammatical…………………………………………………………………………..105 Figure 4-19. Ratings of Meaning Conveyed when learners were confident vs. not confident in their productions…………………………………………………………..107 Figure 4-20. Learners’ confidence when their productions did vs. did not convey intended meaning……………………………………………………………………….108 Figure 4-21. Learners’ confidence and familiarity (10-point scale): Frequent vs. infrequent……………………………………………………………………………..110 Figure 4-22. Learners’ confidence and familiarity (10-point scale): With vs. without L1 equivalents……………………………………………………………………………...110 Figure 4-23. Learners’ confidence and familiarity (10-point scale): Light vs. content verbs…………………………………………………………………………………….111 Figure 4-24. Learners’ confidence and familiarity (10-point scale): Same vs. different light verbs…………………………………………………………………………….…111 Figure 4-25. Learners’ confidence and familiarity (10-point scale): Easy vs. difficult types…………………………………………………………………………………….112 Figure 4-26. Learners’ confidence and familiarity (10-point scale): Total…………...112 Figure 4-27. The average numbers of different tokens produced per item of each subtype…………………………………………………………………………………115

Figure 5-A. Average numbers of literal and abstract combinations produced………123 Figure 5-B. Average number of responses produced by each subject for literal vs. abstract collocations…………………………………………………………………….124 Figure 5-C. Number of responses produced by the subjects for collocations with vs. without L1 equivalents. A1, A2, A3………………………………………………….…125 Figure 5-D. Number of responses produced by the subjects for collocations with light verbs vs. content verbs………………………………………………………………….126 Figure 5-E. Number of responses produced by the subjects for collocations with same vs. different light verbs…………………………………………………………………127 Figure 5-F. Native speakers’ Grammaticality Judgment (6-point-scale). Experiment 2...... 129 Figure 5-G. Native speakers’ Grammaticality Judgments: Literal vs. abstract collocations. (6-point-scale)…………………………………………………………….130 Figure 5-H. Native speakers’ Grammatical Judgments: collocations with vs. without L1 equivalents. (6-point-scale)………………………………………………………....131 Figure 5-I. Native speakers’ Grammatical Judgments for collocations with light verbs vs. content verbs. (6-point-scale)……………………………………………………….132 Figure 5-J. Native speakers’ Grammatical Judgment for collocations with same vs. different light verbs. (6-point-scale)…………………………………………………....133 Figure 5-1. Native speakers’ Grammaticality Judgment. A 1……………………….135 Figure 5-2. Native speakers’ Grammaticality Judgment. A 2……………………….135 Figure 5-3. Native speakers’ Grammaticality Judgment. A 3……………………….135 Figure 5-4. Native speakers’ Grammaticality Judgments. Total A1, A2, A3………..136

xvii

Figure 5-5. Grammaticality Judgments. Literal vs. abstract collocations………….137 Figure 5-6. Grammaticality Judgments. Collocations with vs. without L1 equivalents……………………………………………………………………………...138 Figure 5-7. Grammaticality Judgments. Light vs. content verbs……………………138 Figure 5-8. Grammaticality Judgments. Same vs. different light verb + noun……..138 Figure 5-K. Grammaticality Judgments: Collocations with literal vs. abstract meanings………………………………………………………………………………..141 Figure 5-L. Grammaticality Judgments: Collocations with vs. without L1 equivalents……………………………………………………………………………...142 Figure 5-M. Grammaticality Judgments. Light verb + noun vs. content verb + noun……………………………………………………………………………………..143 Figure 5-N. Grammaticality Judgments: Same vs. different light verb + noun……..144 Figure 5-O. Native speakers’ judgments of Meaning Conveyed in Experiment 2. (6- point-scale)……………………………………………………………………………...147 Figure 5-P. Native speakers’ judgments of Meaning Conveyed in Experiment 2. Literal vs. abstract collocations. (6-point-scale)………………………………………..148 Figure 5-Q. Native speakers’ grading of Meaning Conveyed: Collocations with vs. without translation equivalent in Japanese and English. (6-point-scale)……………….149 Figure 5-R. Native speakers’ judgments of Meaning Conveyed: Collocations with light verbs vs. content verbs. (6-point-scale)………………………….………………...150 Figure 5-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs. (6-point-scale)………………………………………….151 Figure 5-9. Native speakers’ judgments of Meaning Conveyed. A1…………….…153 Figure 5-10. Native speakers’ judgments of Meaning Conveyed. A2…………….…153 Figure 5-11. Native speakers’ judgments of Meaning Conveyed. A3…………….…153 Figure 5-12. Meaning Conveyed. Total A1, A2, A3………………………………….154 Figure 5-13. Meaning Conveyed. Literal vs. abstract collocations…………………..155 Figure 5-14. Meaning Conveyed. With vs. without L1 equivalents…………………155 Figure 5-15. Meaning Conveyed. Light verb vs. Content verb………………………156 Figure 5-16. Meaning Conveyed. Same vs. Different light verbs…………………...156 Figure 5-T. Meaning Conveyed. Literal vs. abstract collocations…………………..158 Figure 5-U. Meaning Conveyed. Collocations with vs. without L1 equivalents……159 Figure 5-V. Meaning Conveyed. Light verb vs. content verb……………………..160 Figure 5-W. Meaning Conveyed. Same vs. different light verb…………………...161 Figure 5-X. Average confidence scores for each collocation type (10-point-scale)…164 Figure 5-Y. Average confidence scores for literal vs. abstract collocations. (10-point- scale)…………………………………………………………………………………....165 Figure 5-Z. Average confidence scores for collocations with vs. without L1 equivalent. (10-point-scale)…………………………………………………………………………166 Figure 5-AA. Average confidence scores: light verb vs. content verb. (10-point- scale)……………………………………………………………………………………167 Figure 5-BB. Average confidence scores: same vs. different light verb. (10-point- scale)……………………………………………………………………………………168 Figure 5-CC. Average familiarity scores for each collocation type: A1, A2, A3 (10- point-scale). Experiment 2……………………………………………………………...170 Figure 5-DD. Average familiarity ratings: Literal vs. Abstract collocations. (10-point-

xviii scale)………………………………………………………………………………...….171 Figure 5-EE. Average familiarity scores: Collocations with vs. without L1 equivalent. (10-point-scale)………………………………………………………………………....172 Figure 5-FF. Average familiarity scores: light vs. content verb. (10-point-scale)……174 Figure 5-GG. Average familiarity scores: same vs. different light verb. (10-point- scale)……………………………………………………………………………………176 Figure 5-17. Grammaticality when learners were vs. were not confident in their productions……………………………………………………………………………..176 Figure 5-18. Learners’ confidence when their productions were grammatical vs. ungrammatical…………………………………………………………………………..177 Figure 5-19. Meaning conveyed or not conveyed when learners were confident vs. not confident in their productions…………………………………………………………..178 Figure 5-20. Learners’ confidence when their combinations did vs. did not convey the intended meaning……………………………………………………………………….179 Figure 5-21. Learners’ confidence and familiarity (10-point scale): literal vs. abstract...... 180 Figure 5-22. Learners’ confidence and familiarity (10-point scale): With vs. without L1 equivalent……………………………………………………………………………….181 Figure 5-23. Learners’ confidence and familiarity (10-point scale): light vs. content verbs…………………………………………………………………………………….181 Figure 5-24. Learners’ confidence and familiarity (10-point scale): same vs. different light verbs……………………………………………………………………………….182 Figure 5-25. Learners’ confidence and familiarity (10-point scale): easy vs. difficult types…………………………………………………………………………………….182 Figure 5-26. Learners’ confidence and familiarity (10-point scale): Total…………...183 Figure 5-27. High, mid low scores of the Grammaticality Judgments. Easy vs. difficult types………………………………………………………………………………….…185 Figure 5-28. The average numbers of different tokens produced per item of each subtype………………………………………………………………………………….187 Figure 5-29. Acceptable collocations. Frequent, neither frequent nor infrequent and infrequent types……………………………………………………………………...….193 Figure 5-30. Acceptable collocations. Frequent, neither frequent nor infrequent and infrequent types………………………………………………………………………....194 Figure 5-31. Percentages of negative transfer for each Japanese light verb………….196 Figure 5-32. Misuse of light verbs……………………………………………………197 Figure 5-33. Percentages of transfer errors………………………………………..….202 Figure 5-34. Percentages of transfer errors…………………………………………...204

Figure 6-1. Percentages of accurate collocations of types A1………………………211 Figure 6-2. Percentages of accurate collocations of types A2………………………212 Figure 6-3. Percentages of accurate collocations of types A3……………………....212 Figure 6-A. Summary of percentages of accurate collocations in Experiments 1 and 3…………………………………………………………………………………………213 Figure 6-B. Percentages of accurate collocations in Experiments 1 and 3. Frequent vs. infrequent types…………………………………………………………………………214 Figure 6-C. Percentages of accurate collocations in Experiments 1 and 3. With vs.

xix

without L1 equivalents………………………………………………………………….216 Figure 6-D. Percentages of accurate collocations in Experiments 1 and 3. Light verbs vs. content verbs………………………………………………………………………...218 Figure 6-E. Percentages of accurate collocations in Experiments 1 and 3. Same light verbs vs. different light verbs…………………………………………………………..219 Figure 6-4. Percentages of accurate collocations of types A1……………………...223 Figure 6-5. Percentages of accurate collocations of types A2………………………223 Figure 6-6. Percentages of accurate collocations of types A3………………….…...224 Figure 6-F. Summary of percentages of accurate collocations in Experiments 1 and 3…………………………………………………………………………………………225 Figure 6-G. Percentages of accurate productions in Experiments 2 and 4. Literal vs. abstract………………………………………………………………………………….226 Figure 6-H. Percentages of accurate collocations in Experiments 2 and 4: with vs. without L1 equivalents………………………………………………………………….228 Figure 6-I. Percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs………………………………………………………………………...229 Figure 6-J. Percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs………………………………………………………………………...231 Figure 6-7. Percentages of accurate collocations in the pretest and the posttest…....235 Figure 6-8. Learners’ improvement vs. amount of time they studied……………….236 Figure 6-9. Percentages of accurate collocations by learners who used 1, 2 or 3 materials………………………………………………………………………………...236 Figure 6-10. Learners’ improvement vs. age…………………………………………238 Figure 6-11. Evaluations of materials (4-point scale) vs. improvements…………….239 Figure 6-12. Subjects’ age and average ratings of the materials……………………..240 Figure 6-13. Improvement vs. the percentages of English use………………………241 Figure 6-14. Acceptable collocations. Frequent, neither frequent nor infrequent and infrequent types………………………………………………………………………...243 Figure 6-15. Percentages of improvement…………………………………………....244 Figure 6-16. Acceptable collocations. Literal, relatively literal and abstract types.…245 Figure 6-17. Increase of acceptable collocations in the posttest. Literal, relatively literal and abstract types……………………………………………………………………….246 Figure 6-18. Percentages of negative transfer for each Japanese light verb………….247 Figure 6-19. Misuse of light verbs pretest Experiments 3 and 4……………………..249 Figure 6-20. Increase of accurate percentages in the posttest. Easy vs. difficult types…………………………………………………………………………………….250

CHAPTER 1

Introduction

1.1. Collocations

The term collocation has been defined in various ways and the terminology is not yet fixed. Overall, there are three main kinds of views in the work on collocations: (1) corpus research views, (2) sentence- and discourse-building views, and (3) linguistic views.

In his corpus research on collocations, Kjellmer (1994) defines collocations as such sequences of items as are grammatically well formed. This definition adopts a broad interpretation of the term, giving the name to all greater-than-chance groups of words which are grammatically well-formed. Benson et al (1986) and Kjellmer (1995) adopt a practical operational approach which reflects the procedure they use to extract collocations from language corpora data and study them. Corpus-generated dictionaries of English collocations include “The BBI Combinatory Dictionary of English Word

Combinations” (Benson et al., 1986), “A Dictionary of English Collocations: Based on the Brown corpus” (Kjellmer, 1995), “Collins Cobuild English words in use: A dictionary of collocations” (Collins Cobuild, 1999), and “Oxford collocations: Dictionary for students of English” (Oxford, 2002) based on the British National Corpus. These dictionaries serve as tools to help learners use English in a native-like way, by providing crucial information on how English words combine with other words.

Taking a sentence- and discourse-building view, Nattinger and DeCarrico (1992) investigate both the psychological and the distributional phenomena of language. They

1 provide extensive categorizations of statistically recurring combinations of words, and investigate various psychological as well as distributional phenomena of language. They use the term lexical phrases as their general designation for multiword linguistic phenomena, including polywords (hold your horses), institutional expressions (how are you?), phrasal constraints (…ago), and sentence builders (either … or …). They suggest that lexical phrases are extremely common in fluent speech and writing, and that they are an important source of linguistic material for language learners to analyze and from which to derive syntactic and lexical information.

In the linguistic view, the term collocation is reserved for a quite specific linguistic phenomenon. Wood (1981), for example, considering the native speakers’ entire repertoire of utterances, proposes a cline of language from productivity to frozenness. He considers semantic compositionality and productivity of form for phrases as factors which constitute a continuum from complete frozenness to full freedom of combination.

In my study, I adopt Lewis’s simple concept that: “A collocation is two or more words that tend to occur together” (Lewis, 2000: 73) since his definition best describes the phenomenona I will be studying. Collocations are frequent co-occurrences of lexical items or of particular constructions. They are a type of syntagmatic relation that is predictable to a greater or lesser extent (e.g., the bond between spick and span is stronger than that between letter and pill-box) (Crystal 1997).

The terminology is not yet fixed. A number of labels have been given to formulaic language, including ‘phraseological units’ (Ginzburg et al. 1979, Gläser 1986),

‘word-combinations’ (Akhmanova 1974, Cowie 1994), and ‘phrasal lexemes’ (Lipka,

1991, Moon, 1998). Despite differing labels, researchers are, more or less, investigating

the same phenomenon: they are pursuing the role of patterning of words and phrases in

communication, language processing, first/second language acquisition, language loss, etc. In this sense, I will use the term collocations and other terms which refer to formulaic language interchangeably.

A range of criteria has been proposed for formula identification (Peters, 1983,

1985; Weinert, 1995). They include: A formula is 1) at least two morphemes in length; 2) phonologically coherent; fluently articulated, non-hesitant, encoded without break in an intonation contour; 3) unrelated to production patterns in a learners’ speech; 4) of greater complexity than a learner’s other output; 5) used repeatedly and always in the same form;

6) possibly inappropriate (syntactically, semantically, or pragmatically; for example, I carry you = I want you to carry me (Weinert, 1995), or otherwise idiosyncratic; 7) situationally dependent (I’m finished = I want to go, used only in the context of a child’s painting class (Hatch, Peck, and Wagner-Gough, 1979); and 8) often community-wide in use.

Some phrases are entirely fixed (How are you?) while other sequences have open slots (can you …) (Hakuta, 1974). The connection between words in a formula lies on a continuum between strong—the presence of one word means you strongly expect the other, too—and weak, as illustrated in the following cline (Lewis, 2000: 74):

friendly dog strong coffee sibling rivalry throw in the towel Weaker ------Stronger old car heavy smoker mitigating circumstances Stars and Stripes

Figure 1-1. Lewis’ model of continuum (Lewis, 2000: 74)

Sometimes collocability can be predicted with reference to semantics. For example, the verb cause almost always co-occurs with unpleasant collocates (problems, death, damage, trouble, disease, etc.), the verb provide occurs with nouns which denote things that are desirable or necessary (information, service(s), support, help, money, protection, food, care, etc. (Stubbs, 2001: 45). But collocability cannot be solely determined by universal semantic constraints (McCarthy, 1990). There are many combinations that semantics cannot predict. For instance, the word beige may modify the noun car but not hair (She has a beige car vs.*She has beige hair), and the word blond modifies hair but not car (She has blond hair vs. *She has a blond car) (McCarthy

(1990: 12-3). The ability to predict such collocability is part of a native speaker’s collocational competence. Native speakers develop knowledge of which words are more likely to occur together through repeated exposure to the language. Indeed there is a close connection between frequency and collocations; frequent combinations are more likely to be institutionalized and conventionalized than less frequent ones.

The frequent co-occurrence of certain words results from their institutionalization as expressions (Moon, 1998). For example, native speakers of English say strong wind and heavy rain, but it is not normal to say *heavy wind or *strong rain. Although these last two combinations of words follow syntactic rules, native speakers simply do not produce such combinations in speech or writing (Pawley & Syder, 1983, Oxford, 2002).

The ability to combine words to produce natural-sounding speech is part of a native speaker’s collocational knowledge.

However, we cannot rely solely on frequency as a criterion to distinguish

collocations from syntactic constructions. Many syntactic constructions recur frequently as a result of how common their referents are rather than as a result of their institutionalization as expressions. For example, the phrase a white house may occur

frequently, not because the phrase has been institutionalized, but simply because the

referent is common, so this phrase is not considered to be a collocation.

It is sometimes not easy to distinguish between collocations, idioms, and free

combinations. An idiom is characterized as a relatively frozen expression whose

meaning cannot be built compositionally from the meanings of its component words, and

the component words cannot be replaced with synonyms (kick the bucket but not *boot

the bucket or *kick the pail). In contrast, collocations are sequences of words which habitually co-occur and whose meanings can be derived compositionally; they often permit a limited degree of substitution of their component words (such as do one’s best, try one’s best, but not *make one’s best). (Bentivogli and Pianta, 2003). And a free combination is a combination of words following only the general rules of syntax: the elements are not bound specifically to each other and they can be substituted with other lexical items freely (read a book, read a newspaper, read a magazine, write a book, borrow a book, etc.) (Benson et al., 1986). While idioms and collocations are lexical units, free combinations do not belong to the lexicon; rather, the individual elements of a free combination, which are stored separately in the lexicon, are combined following syntactic rules.

However, the boundaries between idioms, collocations, and free combinations are not clear-cut. It is sometimes difficult to distinguish an idiom from a collocation or a free combination from a collocation. Howarth (1998) assumed that lexical items in these

categories do not have definite boundaries, rather they are ranged on a cline from pure

idioms to free combinations.

------ Pure idioms figurative idioms restricted collocations free combinations [blow the gaff] [blow your own trumpet] [blow a fuse] [blow a trumpet] [under the weather] [under the microscope] [under attack] [under the table]

Figure 1-2. Howarth’s model of continuum (Howarth 1998: 28)

The leftmost extreme of the continuum contains pure idioms, and the rightmost extreme

contains free combinations (pure syntactic constructions). Howarth explains that

collocations are sets of words which lie between figurative idioms and free combinations

in the cline. This approach allows for fuzzy areas between collocations and figurative

idioms/free combinations. That is, these categories do not have definite boundaries, but

rather reflect shifting proportions of semantic and syntactic characteristics, such as

number and type of arguments for verbs, or animacy for nouns.

Collocations can be divided into two subtypes: grammatical collocations and

lexical collocations (Benson 1985). Grammatical collocations consist either of a

dominant word (usually a verb, a noun, or an adjective) and a dependent word such as a

preposition; or of a particular structural pattern, such as the dative-movement

transformation, that-clause, or to + infinitival + gerund. Lexical collocations, in contrast,

consist of two “equal” components, such as verb + noun or adjective + noun. Examples of grammatical and lexical collocations are given in Table 1-1. Verb + Noun collocations, the topic of my research are classified as lexical collocations. So I have only considered lexical collocations in my study.

Type Examples Grammatical Collocations: ● verb + preposition ● (to) get at, (to) go for ● adjective + preposition ● different from, curious about, full of ● adjective + preposition + ● fed up with preposition ● preposition + noun ● for sale, on time ● dative movement ● She sent the book to him/ She sent him the book. transformation ● He described the book to me/ *He described me the book. Lexical Collocations: ●verb + noun (pronoun, ● (to) reach a verdict, (to) launch a missile, (to) prepositional phrase lift a ) blockade, (to) revoke a license ●adjective + noun ● reckless abandon, sweeping generalization ●noun + verb ● adjectives modify, alarms go off ●noun + of + noun ● a bunch of flowers, a piece of advice ●adverb + adjective ● deeply religious, fiercely independent ●verb + adverb ● (to) apologize humbly, (to) affect deeply TABLE 1-1. Examples of collocations. Source: Benson (1985).

1.2. Studies on acquisition of collocations

1.2.1. Research on collocations in first language acquisition

Young children often produce what seem to be unanalyzed chunks which an adult would perceive as multi-morphemic, such as lemme-see, I-wanna-do-it, what’s-this?, gimme-that, I-know-how-to-do-it, that’s-mine. This phenomenon raises a question about the validity of the well-known traditional assumption that most children begin producing

just one word at a time (Bloom, 1973) and of the near-universal dependence on mean

length of utterance (MLU) as a standard against which to judge the comparability of

children in different studies. Peters (1983) proposed that the first units young children

extract from adult speech do not coincide exactly with the words and morphemes of the

language system. Moreover, Locke (1997) claims that young children have little

7 awareness of words as units, nor of the morphemes and phonemes within them. From this perspective, it is not surprising that the first chunks young children extract from adult speech do not coincide exactly with the words and morphemes of the language system.

Collocations are very important in learning one’s first language. Wray (2000,

2002) describes several important roles for collocations in language acquisition. Young children’s use of collocations may supplement their gestures and other nonlinguistic behaviors in conveying manipulative messages. Young children store and use complex strings because they need to use language to get things long before their grammar has developed. Using collocations also makes it possible to produce utterances which contain materials well in advance of a child’s grammatical knowledge. For example, children may produce the chunk what’s-that? before they learn the internal structure of wh-questions. This phenomenon is what Clark (1974) called ‘performance without competence’. Another role is that collocations may reduce children’s processing load, allowing them to retain fluency while gaining control of processing. In addition, collocations help children interact with caregivers, and support their entry into the social group. Young children may use particular collocations out of a desire to be part of a group, interacting with their caregivers and peers.

In addition to their roles in communication and language production, collocations may be important for children’s grammar development. Although some scholars (Brown,

1973; Krashen and Scarcella, 1978; Bohn, 1986) have claimed that collocations play only a minor role in learning grammar, evidence has been accumulating which suggests that they are central to the language acquisition process, especially at the early stages. The idea is that memorized collocations yield templates for further analysis; children segment

previously unanalyzed sequences by taking note of their internal patterns of rhythm,

intonation and stress, as well as the boundaries which are revealed through a speaker’s

repetition of fragments of an utterance (Bloom, Hood and Lightbown, Locke, 1975, Snow

1981). For example, children frequently use the phrase I-want-to-go as though it were a single, unsegmented unit in social situations in which the function of request is clearly associated. Gradually, in the process of acquiring other chunks with similar syntactic patterning (I-want-to-get-up, I want-my-ball, I-want-a-cookie), the child detaches the syntactic pattern from its connection to the context, and analyzes and generalizes it into regular syntactic rules (Peters, 1983).

Wray (2002) developed a model of formulaicity in language acquisition, to account for how unanalyzed collocations are later unpacked and analyzed by children, with concomitant shift in the proportion of holistic versus analytic language processing.

Her model consists of the following four phases (Wray, 2002: 133).

Proportion of processing ALL

HOLISTIC ANALYTIC

NONE Age 0------2------8------18------(yrs) |______Phase1______|_____Phase2______|_____Phase3______|___Phase4______

Figure 1-3. The balance of holistic and analytic processing from birth to adulthood.

Phase 1 (from birth to around 20 months) is dominated by holistic processing, beginning

with the imitation, interpretation, and use of facial and body gestures and intonation, and moving to the production of gestalt utterances in specific social contexts. Phase 2 (from

20 months to 8 years) starts with a vocabulary spurt and the development of grammatical awareness. During this phase, children’s processing becomes predominantly analytic; their interest is in identifying words and morphemes and in analyzing how constituents are combined. Phase 3 (from about 8 to 18 years) is the stage during which formulaic language and collocations again begin to make up a greater proportion of the individual’s output, resulting in increasingly holistic language processing. During phase 4 (from the late teens) a balance is reached between holistic and analytic processing. This model suggests that children first memorize many collocations holophrastically, analyzing them into smaller pieces by the process of segmentation as their grammatical competence develops.

The transition from holistic to analytic processing in child language has been observed in various studies. Tomasello (2003: 262) observes that children first learn those lexical items and item-based constructions (I wanna, I gotta, I think, look at…, see, it’s …) which they hear most frequently; before three years of age, they seldom embed one proposition into another in any of these complements. Karniol (1990) found that children become disfluent around age 2-3 years. He suggests that this is because children at this age are in the transition of changing their sentence-production strategies from holistic to analytic. Cruttenden (1981) claims and provides evidence that a transition from holistic to analytic processing happens at all levels of language, including intonation, semantics, and syntax. Taking a cognitive perspective, Bolander (1989: 85) proposes that

“when the number of prefabs stored in memory is large enough, syntactic rules are

derived as help for the memory to economize and rationalize processing”. That is, the

pressure to memorize forces a reorganization of the language data into sub-units suitable

for fast and flexible manipulation. This claim provides an account of why the onset of

grammar development immediately follows the vocabulary spurt around 18 months of

age (Plunkett, 1993).

The underlying assumption here is different from the traditional generative view

that children are born with an innate endowment which makes language acquisition

possible: language acquisition is specifically a process of parameter-setting, and the primary interest is to investigate the abstract internalized system of grammar (language competence). In this framework, in producing multi-word speech, children must be manipulating syntactic categories such as subject, direct object, verb and auxiliary to produce rule-governed grammatical utterances. However, many findings in language development research support more constructivist approaches, especially emergentist or usage-based models which claim that children’s learning is gradual, piecemeal, and lexically dependent. This approach is concerned with language use as the basis for understanding the nature of language, positing that a language is acquired through usage, by extracting patterns and regularities from the input, and building ever-stronger associations in the brain. Language develops as children move away from learning exemplars (words, formulae) that are first memorized holistically. From these, regularities emerge, giving rise to slot-and-frame patterns or constructions, such as all- gone + referent or I can’t + verb. As more and more such formulae develop, they can be compared and analyzed, regularities can be extracted and applied elsewhere (Ellis, 2003:

84). This approach assumes that children’s knowledge is lexically-specific, with their

knowledge of, for example, verb-argument structure developing initially around

individual verbs (Tomasello, 1992; Allen, 1998) and individual lexical frames (Braine,

1976; Lieven et al., 1997). This claim suggests that children do not initially have a

grammatical category of verb when they first begin to produce them, nor do they operate

with abstract categories such as Subject or Direct Object (Theakston et al., 2001: 131)

In general, the importance of collocations has been proposed for the early stages

of acquisition. In view of the dynamic transition from holistic to analytic processing seen in Wray’s model, it is possible that collocations have more important roles in the earlier stages. However, there are a number of claims that collocations may also contribute to later stages of language acquisition. From a pragmatic and sociolinguistic point of view,

House (1996) argues that pragmatic competence involves memorizing large numbers of

routines together with the social knowledge that members of a given speech community

share; these include many formulaic expressions. Learning collocations may thus

continue as more and more cultural bonds are acquired.

In reality, we don’t yet know to what extent learning of collocations is intertwined

with learning of traditional grammar. Elbers (1995) claims that much of children’s

syntactic development derives from their analysis of their own linguistic production.

This claim suggests that collocations are intertwined with learning syntax to quite a large extent. However, to date little research has been conducted as to how very abstract grammatical rules, such as the concept of c-command, can be learned through pattern- extraction. We must devise more empirical studies which focus on the interconnection between learning collocations and grammar development.

1.2.2. Research on collocations in second language learning

Krashen and Scarcella (1978) claim that formulaic language is a means of outperforming competence but does not serve a primary role in language acquisition and performance. Bohn (1986) also argues that learners internalize L2 rules independently by attending to input, and that routines, although certainly an integral part of language use, are definitely not numerous or crucial.

However, Fillmore (1979), Pawley and Syder (1983), and more recently,

Nattinger and DeCarrico (1992) stress that linguistic behavior is ritualized to a large extent and that routines as memorized stretches do indeed form a high proportion of the fluent stretches of adult native speakers’ everyday conversation. Pawley and Syder

(1983) point out that only a small proportion of the total set of potential ‘grammatical’ sentences is ‘nativelike’, in the sense of being readily acceptable to natives as ordinary, natural forms of expression, as opposed to being grammatical but unidiomatic, odd, or foreign-sounding. They claim that full control of a language must entail knowledge of something more than a generative grammar. For example, when an English speaker expresses his or her desire to marry someone, a natural way of saying this would be I want to marry you. Other sentences such as I wish to be wedded to you, Your marrying me is desired by me, or My becoming your spouse is what I want, although grammatical, would not normally be used by native speakers (Pawley and Syder, 1983). Pawley and

Syder propose that the additional knowledge that underlies nativelike language abilities is

‘memorized sentences’ and ‘lexicalized sentence stems’. According to them, formulaic language makes up a large percentage of the language we use, and the stock of

prefabricated units enables natives to select appropriate forms of expression from a

smallish range of grammatically acceptable alternatives, and to produce fluent speech or

writing without apparent planning, effort, or hesitation.

In contrast, when L2 learners make productions in a target language, their

performance may be perceived as non-nativelike due to their limited command of acceptable lexical phrases. They often confine themselves to a limited range of familiar vocabulary, or produce expressions that sound odd, unidiomatic, or unintentionally amusing; and they are likely to be hesitant, to speak more slowly and write less volubly

than natives do. In the next section, I will describe some research on L2 formulaic

language and discuss some problems learners have in acquiring formulaic language.

A number of research projects support Pawley and Syder’s claim that L2 learners

are not aware of the importance of formulae in a second language, and do not have

sufficient collocational competence. Indeed, the acquisition of collocational competence

is regarded as one of the most difficult parts of language learning. Bahns and Eldaw

(1990) conducted an experiment consisting of a translation task and a gap-filling task

with advanced learners of English who had German as a native language. They found

that the students’ knowledge of collocations had not developed at the same rate as their

knowledge of vocabulary in general. Their findings suggest that adult L2 learners rely

heavily on holistic processing in early stages when formulaic sequences predominate,

such as ‘Good morning’, ‘How are you?’ and ‘Thank you’. But when they outstrip the

limitations of such fixed forms, they rely on individual words rather than on formulaic

sequences as stretched units for building up phrases and sentences.

Corpus studies have revealed different proportions of formulaic language use

14 between native and non-native speakers. According to Howarth (1996), in the whole native speakers’ corpus, formulaic language was 38%, while the non-native speakers’ proportion was 25%, suggesting that non-native speakers produce fewer collocations, suggesting a lower level of knowledge of collocations or the lack of awareness of how to deploy them appropriately, or both.

Schmidt and Frota (1986) state that errors made by L2 learners are often not an indication of creative hypothesis testing, but rather result from re- or misuse of stored chunks and from not yet fully developed and automatized scripts at the learners’ disposal.

For example, Granger (1998) analyzed the written performance of advanced French students of English and found that the learners overused very frequent collocations but underused creative constructions, while in addition some of their creative constructions were non-native like. In a judgment task, the non-native speakers accepted a greater number of types of adverb-adjective combinations than did natives. He proposed that a learner’s sense of salience is not only weak but partly misguided.

A number of research projects have attempted to categorize L2 learners’ problems with acquiring collocational competence. Transfer is one of the common factors which causes problems. Biskup (1992) collected interference errors made by Polish and

German learners of English (Polish interference errors: *to state a record [set]) and

German: *to lead a bookshop [run]). The learners translated word-for-word from their

L1. But Granger (1998) argues that the learners’ L1 may affect learning collocations either positively or negatively. He found evidence of successful collocational transfer from L1. For example, French learners of English produced severely punished, which corresponds to French: sévèrement puni. Irujo (1986) also found evidence of positive

15 transfer in learning idiomatic expressions by German learners of English, and claimed that it is unnecessary to teach idiomatic expressions if literal translation equivalents exist in the learners’ native language.

Overextending or underextending collocational restrictions is another common problem. Howarth (1998) collected informal recordings of non-native speech, and illustrated learners’ misguided competence in collocational restrictions. According to his findings, L2 learners have a misguided range of collocabilities; for example, a learner produced ?Give lip service to instead of the conventional collocate pay lip service to.

The learner had mastered the idiomatic component of the phrase but allowed it more freedom to collocate with verbs than it has for native speakers (overextending the collocational restriction).

Analogical errors are also common. L2 learners sometimes mix up overlapping collocational restrictions of other semantically similar words (e.g., give vs. make; give a reaction, give a comment, give a response, and make a comment, make a response, but not *make a reaction) (Howarth, 1998).

Blending is not very common but is one of the factors which induce collocational errors. Howarth observed blending errors among restricted collocations such as *pay care, by blending pay attention and take care. But Dechert and Lennon (1989) consider that blendings happen because the learners’ attention is focused on syntactic structure rather than procedural automaticity; they even comment that blending can be a sign of competence rather than of its absence, because one can only produce blends of collocations that are already known. According to their analysis, for example, if an advanced learner produces *He has got financially broke, it means that he or she has

heard both he is broke and he has got into financial difficulties.

Phonological similarity may also lead to errors in collocations. For example, a

learner may produce the erroneous form The general Synod took the monumental decision, because of the phonological similarity between the word monumental and momentous (Howarth, 1998).

The learners’ use of collocations may reflect not only the structural aspects of language, but also other aspects, including pragmatics, instructional settings, and socio- cultural and psychological issues. From a series of contrastive (German-English) pragmatic and discourse analyses, House (1996) found a pattern of cross-linguistic and cross-cultural differences in the use of formulaic language: German speakers interact in ways that tend to be more direct, more explicit, and more self- and content-focused and— most important in the context of this study—are less prone to use verbal routines than

English speakers. The German-English differences in reliance on routines led to many pragmatic failures on the part of German learners of English because of their under-use and/or inappropriate use of routines in English.

Instructional settings also affect the production of formulaic language. For example, when Biskup (1992) collected data on formulaic language use from Polish and

German learners of English, he found that Polish students were more inclined not to answer than German students and less likely than the latter to produce a descriptive paraphrase. He explained that this is because of the differences between instructional settings in these two countries: Germans pay more attention to fluency and communication while the Polish educational system insists on accuracy.

Wray (2002: 175) suggests that there is a link between the use of formulae and

17 learners’ psychological factors, such as motivation, need, and desire to interact with speakers of the target language. She found that these together correlate with the overall achievement of communicative competence. Schmidt’s (1983) subject Wes, a 33-year- old Japanese immigrant to Hawai‘i who was highly motivated to make himself understood, achieved a high level of fluency and used many formulaic sequences. On the other hand, Schumann’s (1978) subject and Shapira’s (1978) Guatemalan subject, who each maintained a social and psychological distance from the new language, used fewer formulaic sequences and achieved less success in learning. L2 learners who strive to be accepted as members of the target language community are more motivated in learning, interact more with native speakers, receive more input and feedback, and have more opportunity for output.

1.2.3. The functions of collocations in second language acquisition

Weinert (1995) outlines three functions for the use of formulaic language: communicative, productive, and learning strategies. First, the use of formulaic language reflects a communicative strategy allowing learners to communicate. Learners at early stages rely almost entirely on formulaic sequences, such as ‘Good morning’, ‘How are you?’ and ‘Thank you’. They acquire a limited set of formulas to compensate for their lack of ability to construct novel sentences (Myles et al., 1999).

Second, the use of chunks may be viewed as a production strategy, whereby use of formulaic sequences allows fluency in production, faster processing, and more time for language planning (Raupach, 1984).

The third function is as a learning strategy. That is, it is possible that unanalyzed

utterances contribute to the development of an emerging grammatical competence. The

assumption is that learners may use formulae to derive rules from them which they then

use more productively. Fillmore (1976) states that formulaic utterances are eventually

analyzed into their component parts and thereby contribute to the learner’s creative rule

system. More recently, Ellis (1996) argues that language sequences serve as the database

for the acquisition of language grammar and that much of language learning is the acquisition of memorized sequences of language.

Various empirical studies support the view that learners unpack their early chunks

and use parts of them productively in the generation of new utterances. Myles et al.

(1999) track the development of three French chunks for two years to chart their

breakdown and explore their contribution to grammar development. They found that

most of the learners gradually broke down their early chunks, and used parts of them to

produce new utterances. The following table shows the developmental stages of

interrogative chunk comment t’appelles-tu?:

The developmental stages of interrogative chunk Examples: comment t’appelles-tu?:

1. Chunk inappropriately used, over extended. *comment t’appelles-tu? ↓ 2. Chunk over extended, but with lexical NP tagged on. *comment t’appelles-tu le ↓ garcon? 3. Chunk starting to break down: subject pronoun *comment t’appelles (la fille)? omitted or replaced by a NP. ↓ 4. Further breaking down: reflexive pronoun changes *comment s’appelle?; to s’: subject pronoun is omitted or replaced by a comment s’appelle… lexical NP. garç-un garcon? ↓ 5. Third-person pronoun is used. comment s’appelle-t-il?

TABLE 1-2. The developmental stages of interrogative chunk comment t’appelles-tu? (Myles et al., 1999: 67)

These developmental stages show that the learners were struggling to analyze and restructure the previously learned chunk comment t’appelles-tu?. Though the learners produced erroneous forms in the developmental stages, grammatical competence gradually emerged regarding wh-question construction as well as reflexive pronouns and verbal agreement. Myles et al. concluded that rote-learning of formulae and the construction of rules are not independent processes, but interact and actively feed into one another (Myles et al., 1999: 323).

Ellis (1984) analyzed formulaic speech in early classroom second language development. The formula I don’t know went through the following developmental stages:

The developmental stages of the formula I don’t Examples know. 1. Don’t was used in similar but different I don’t understand, I don’t like expressions. holiday. ↓ 2. The alternation of the subject. You don’t know where it is. ↓ 3. know was released for use without don’t. I know this. ↓ 4. An additional constituent occurred. I don’t know that big one.

TABLE 1-3. The developmental stages of the formula I don’t know (Ellis, 1984)

Again, Ellis’s findings show that chunks contribute to the development of an emerging grammatical competence.

Vihman (1982) analyzed the development of a chunk involving I wanna by an

Estonian child acquiring English over an eight-month period.

The child used wannna followed by an adverb, a verb, or a noun: 2;1;28 I wanna down. 2;2;2 I wanna here. 2;2;22 I wanna home. 2;2;3 I wanna go back. 2;2;14 I wanna eat-it yoghurt. 2;2;22 I wanna see three-jump? (Posing a toy bear on the edge of a couch) 2;4;16 You wanna this do. (This word order, which persisted throughout the period covered here, reflects a common Estonian syntactic choice.)

2;2;4 I wanna elutoas book. (elutoas ‘in the living room’) 2;2;14 I wanna sandals on. 2;2;14 I wanna clothes on. (on hearing Virve will have to put her clothes on) 2;2;22 I don’t wannu potty. 2;4;6 I wanna my lap, mama. (asking for a picnic lunch, to eat on her lap) 2;4;6 I wanna candy. 2;4;16 You wanna your lunch? 2;4;16 I wanna crayon.

Then, she used want followed by a noun or pronoun complement: 2;4;18 You want your mama? 2;2;20 You want another puzzle? 2;4;20 This puzzle you want? 2;4;25 I don’t want it.

Then, she overgeneralized the want-construction. She used want followed by a verb. 2;5;0 You want come my house?

The approximation to the correct English use; wanna appeared followed by a verb or a noun, but not followed by an adverb.

2;7;11 I wanna do puzzle. 2;7;11 I wanna go. 2;7;11 I wanna read a book. 2;7;13 I wanna come inside. (responding to No one can come inside.) 2;7;20 I wanna go back. 2;7;29 I wanna play with Linda. 2;8;15 I wanna play my room. 2;8;22 I wanna this do, you, Jeff. (‘I want to do this with you, Jeff’)

TABLE 1-4. The development of a chunk involving I wanna by an Estonian child acquiring English (The data from Vihman [1982: 277-278])

Vihman’s data suggest a dynamic interplay between grammar and lexicon, in which a ready-made formula I wanna was subject to the child’s emergent syntactic rules or overgeneralizations, with a gradual reorganization in the direction of conventional

English grammar.

These findings show that chunks are very common in the early stages of L2 learning, and learners gradually analyze chunks and derive abstract generalizations out of unanalyzed chunks that they holistically memorized into smaller constituents. These developmental patterns support the constructivist or emergentist views of language learning which claim that a language is acquired through usage, by extracting pattern and regularities from the input, and building ever-stronger associations in the brain (Mitchell and Myles, 2004: 97). Barlow (2000: 319) adopted a usage-based grammar and argued that repeated exposure to collocations leads to the entrenchment of collocational patterns and their associated meanings in the grammar. Ellis (2003) argued that these processes of chunking are an important part of L2 learning.

1.3. Purpose of this study: What factors may affect the acquisition of lexical collocations?

We have seen in the previous section that multiword complexes of language play an important role in the development of the first language, and that adult native speakers’ production is made up of prefabricated chunks of language to a large extent, rather than consisting solely of creatively constructed utterances (Peters, 1983). Research in psycholinguistics (Aitchison, 1987) also provides evidence that the native speakers’ lexicon is not exclusively composed of single-word units, but also includes phrases, sentence stems, and entire sentences.

In second language acquisition, it has been suggested that beginning and even

advanced L2 learners often have considerable difficulty in acquiring the native-like

collocational knowledge needed for the productive skills of speaking and writing. This issue of collocation is a crucial one, for as Pawley and Syder (1983) noted, only a small

proportion of the total set of ‘grammatical’ sentences of a language is nativelike—in the sense of being readily acceptable to natives as ordinary, natural forms of expression. This contrasts with being grammatical but unidiomatic, odd or ‘foreign’. L2 learners’ performance may be perceived as non-nativelike due to their limited command of acceptable lexicalized phrases. They often confine themselves to a limited range of familiar vocabulary, produce expressions that sound unidiomatic, and they are likely to be hesitant, speak more slowly and write less volubly than natives do (Read 2000).

Kjellmer (1991: 124) studied several L2 corpora and concluded that learners were acquiring individual bricks rather than prefabricated chunks. Bley-Vroman (1996) claimed that L2 learners’ competence is in large part characterized by a large collection of instances and lexicalized patterns rather than a full-fledged grammar derived by the applications of rules and resetting parameters.

In order to examine Kjellmer’s claim, DeCock et al. (1998) conducted a corpus- based study of French learners of English, and found that learners did use what Sinclair

(1987) called an ‘open-choice principle’, using a much smaller variety of collocations than native speakers. Granger (1998) investigated French learners’ collocational

combinations of intensifying adverb plus adjective in English, and found that the learners

underuse nativelike collocations and overuse atypical adjectives, such as ‘completely’ and

‘totally’ because there are French cognate equivalents. She concluded that learners’ sense

23 of collocational salience is generally weak and their use of collocations is influenced by knowledge of their first language. Gabry-Biskup (1992) investigated Polish and German speaking learners’ performance on various lexical collocations in English, and found that there were more interference errors when a given L1 lexical item had a wider semantic field than the corresponding lexical item in L2. In addition, the learners were, in some cases, reluctant to use their L1 knowledge even though the L2 collocation was a word- for-word equivalent of an L1 one. He referred to Kellerman and Jordens’ term psychological markedness (1977) which states that: “Psychologically marked features tend to be perceptually more complex, and are thus less likely to serve as models of transfer than unmarked ones.” The learners were more ready to transfer core meanings, while marked meanings were often rendered by means of other strategies, such as synonyms and paraphrases.

Based on the L2 studies discussed above, we would expect that Japanese-speaking learners of English would have the same or similar kinds of problems when they produce collocations in English. Japanese learners’ collocational knowledge is expected to be weaker than that of native speakers’, they may underuse and/or overuse some collocations, they may be willing or unwilling to use their L1 knowledge, and so forth.

My dissertation deals with English Verb+Noun lexical collocations, focusing on a number of factors which may affect L2 learners’ performance. These are: L1 transfer, frequency and semantic transparency of collocations, the proficiency level of learners, and the role of instruction. Each of these factors will be briefly discussed in the next section.

The purposes of this study are to investigate factors that affect L2 learners’

24 performance with lexical collocations and the effect of explicit instructions which focus learners’ attention on input and encourage learners to become aware of collocational restrictions. My goals are to deepen understanding of learners’ performance and to improve instructional methods, and eventually to help learners acquire a native-like command of second language collocations and to facilitate mutual understanding across cultures.

The following five factors were considered in my study.

(1) L1 transfer (word-for-word translations of equivalent collocations in L1). A number of researchers have shown that the learners’ L1 plays a role in creating both positive and negative transfer of forms from L1 to L2 (Bahns and Eldaw, 1993, Hussein,

1991, Marton, 1977). Risk-taking by learners may result in interference errors (Biskup,

1992). However Bahns (1993) found evidence of successful collocational transfer from

L1. I investigate Japanese-speaking learners’ positive and negative transfer and how L1 transfer interacts with other factors, such as the frequency and meaning of collocations, and the proficiency level of learners.

(2) Frequency of collocations. There has been little empirical research which has investigated the relationship between the frequency of collocations and learners’ performance. But Hatch (1974) states that the frequency of input may have an accuracy relationship to second language acquisition. Miyakoshi (2004) investigated collocations involving predicate adjectives and found that learners’ performance was positively related to the frequency of the target collocations. In contrast, other researchers such as Howarth

(1994) found no correlation between learners’ knowledge of collocations and the frequency of target collocations. But Howarth tested an extremely narrow range of ability, so it was highly unlikely that any significant correlation would have emerged from such limited data. Schmitt and Dunham (1999) suggest that advanced learners have intuitions about frequency which are as good or better than those of native speakers. In their study of making judgments of frequency for lexical sets, some advanced learners scored higher for highly frequent words than native speakers. It is possible that the learners’ intuitions of frequency may reflect their knowledge of collocations, and we may observe a correlation between frequency and learners’ performance.

(3) Semantic transparency of collocations. Irujo (1986) proposed non-literalness of meaning as one of several explanations for the fact that figurative expressions are difficult to learn in a second language. Semantic transparency is also related to Kellerman and Jordens’ notion of psychological markedness (1977). It is expected that core meanings will be acquired easily while non-literal meanings will be difficult for L2 learners. I investigate how transparency of meaning affects learners’ performance.

(4) Proficiency level. An overall correlation has been reported between learners’ general proficiency in English (e.g., TOEFL scores) and their productive knowledge of target language collocations (Zhang 1993, as cited by Gitsaki, 1996; Bonk 1996; Miyakoshi

2004). These studies suggest some sort of developmental pattern, whereby learners at higher levels of proficiency tend to use certain types of collocations more often and more accurately than others, as compared with the performance of lower-intermediate learners.

(5) The effect of instruction. Collocational knowledge has been thought to be acquired through years of exposure to the target language. But simple repeated exposure is not sufficient for non-native learners to attain native-like collocational knowledge. Even highly advanced non-natives who have lived in English-speaking countries for many years and use English as their primary communication have mostly not attained native- like collocational competence. Riopel (1984) investigated the acquisition of

Verb+Particle collocations and found an effect of instruction in the second language classroom. So far, very little research has been done regarding the effect of explicit instruction on the acquisition of collocations. I investigate the effects of explicit instruction, which is designed to focus learners’ attention on input and encourage learners to become aware of how collocational restrictions do and do not overlap between two languages, with the hypothesis that such focused attention and awareness are essential to improve their performance in second language collocations.

CHAPTER 2

Preliminary survey: Categorization of L2 learners’ problems in acquiring collocations

2.1. Purpose of the survey and research questions

A preliminary survey was conducted in order to formulate a rudimentary

categorization of L2 learners’ problems in acquiring Verb + Noun collocations. The

following three questions and hypotheses were used in the survey:

Question 1: What kind of errors do L2 learners make? Do they make errors because of

L1 transfer? Do they use such strategies as translating word-for-word from

L1, paraphrasing, avoidance, etc.?

Question 2: Does the learners’ performance in lexical collocations correlate with their

general proficiency (i.e., TOEFL scores) in English?

Question 3: How well is the intended semantic content conveyed when L2 learners

produce erroneous collocations?

Hypothesis for Question 1: L2 learners will produce both intra-lingual and inter-lingual

errors. They may translate word-for-word from L1 into L2, have difficulty distinguishing

words which have similar meanings, lack sufficient knowledge about collocational

restrictions, have difficulty with light verbs which carry relatively small semantic content.

Hypothesis for Question 2: Learners with greater L2 proficiency will demonstrate a

broader knowledge of collocations. Performance will improve along with proficiency.

Hypothesis for Question 3: L2 learners’ collocational errors will not successfully convey

the intended meaning.

2.2. Method

2.2.1. Subjects

The participants in this study were thirty-two Japanese learners of English, including graduate and undergraduate students at the University of Hawai‘i at Mānoa and students studying English in the Hawai‘i English Language Program (HELP) or the New

Intensive Course in English (NICE) program. Non-native speakers were divided into two proficiency groups based on their TOEFL scores: the low-intermediate group had TOEFL scores lower than 550, while the advanced group had TOEFL scores of 575 or higher.

Twenty-five learners (6 males and 19 females) were in the low-intermediate group and

seven learners (2 males and 5 females) were in the advanced group. The mean number of

years of residence in English-speaking countries was one year and six months in the low-

intermediate group and nine years and six months in the advanced group.

2.2.2. Stimuli and procedure

A total of forty Verb+Noun collocations was selected from Kizuka’s (1995)

Compact Dictionary of Verb-Noun Collocations. All the collocations were either

restricted or fixed collocations. That is, a given NP can occur with only a limited number

of verbs for the Verb+Noun phrase to convey the same semantic content. For example,

the NP your best occurs with try and do (i.e., try your best and do your best) to refer to similar semantic content but it does not occur with other verbs such as make, get, or accomplish (i.e., *make your best, *get your best, *accomplish your best). None of the collocations had Japanese literal equivalents. See Appendix I for a list of all the test

29 sentences.

The task in this study was a fill-in-the blank test in which 40 collocations were inserted in sentences of semantically natural context, with the verb of the V + NP collocation being replaced with a blank. The subjects were instructed to fill in the missing words on their own without using a dictionary or any other ESL materials. They were allowed to spend as much time as they wanted to finish the test. They took fifteen to twenty minutes to complete the test. The material used in the experiment is in

Appendix I.

The collocations produced by the L2 learners were graded by 10 native speakers of English using a 5-point scale on two scales: (1) Grammaticality Judgment: whether or not native speakers of English produce the same collocation in the same context; and (2)

Meaning Conveyed: how well the intended meaning is conveyed. The collocations which were rated 4.5 or higher as the average of both of these scales were regarded as native- like collocations. L2 learners’ (particularly beginning learners) occasional mistakes in verb-inflection, such as dropping the third-person-singular –s, and overgeneralization of the past-tense morpheme –ed, e.g., speaked, were ignored, since the focus of this study was to examine the learners’ choice of words in producing V + NP collocations rather than small inflectional errors.

2.3. Results

A total of 1195 collocations were collected (915 collocations produced by low- intermediate learners and 280 collocations produced by advanced learners). Among them,

35.0% of low-intermediate learners’ collocations and 71.8% of advanced learners’

collocations were judged to be native-like. This difference is statistically significant (χ2 =

116.212*, df = 1, p <.05). See Appendix II for all the data from the preliminary survey.

The following six error types were identified: (1) Paraphrasing; (2) Misuse of

light verbs; (3) L1 transfer; (4) Use of words other than verbs; (5) Mixing collocations; and (6) Synonymy of nouns. Table 2-A and Figure 2-A summarize the types and percentages of erroneous collocations produced by L2 learners.

Para- Misuse of L1 other than Mixing Synony- TOTAL phrasing light verbs transfer verbs collo- my of cations nouns Low- 262 219 45 46 19 4 595 intermediate 33.03% 36.81% 7.56% 7.73% 3.19% 0.67% advanced 45 30 2 0 1 1 79 56.96% 37.97% 2.53% 0.00% 1.27% 1.27% p > .05 *4.193 0.006 2.001 *5.396 0.014 0.354 *38.078 TABLE 2-A. Types of collocations.

60% 56.96% Low-intermediate 50% Advanced

37.97% 40% 36.81% 33.03%

30%

20%

10% 7.56% 7.73%

2.53% 3.19% 1.27% 0.00% 1.27% 0.67% 0% Paraphrasing Misuse of light L1 transfer Other than V Mixing Synonymy of V collocations N

Figure 2-A. Types of L2 collocations.

Table 2-A shows that paraphrasing and misuse of light verbs are predominant strategies

employed by L2 learners. It was also found that the advanced learners produced

significantly higher percentages of paraphrasing errors than low-intermediate learners,

while low-intermediate learners produced significantly higher percentages of errors

involving words other than verbs than did advanced learners. Examples of each error

type with the average scores of Grammaticality Judgment and Meaning Conveyed by

native speakers are provided below:

(1) PARAPHRASING: The commonest strategy was paraphrasing. L2 learners used semantically related verbs, such as …if it matches/ fits your convenience and It means no difference to me…, or used verbs whose meaning was roughly the meaning of the whole target collocation, such as Please watch/ look an eye on children… and Donna suicided her life…. There were a few instances in which L2 learners made up creative expressions, such as Mary’s sweater froze my eyes and Maria killed her voice…. Some of the paraphrases were not very successful in achieving the intended meaning, such as Please lose an eye on the children… and Maria taught her voice to tell me the secret.

(1) a. Please copy this photo right away. (develop this photo)

b. Please screw the eggs. (beat the eggs)

c. You can judge a man’s character by the company he contacts. (by the company he keeps)

d. Mail me a line when you have time. (drop me a line)

(2) MISUSE OF LIGHT VERBS: L2 learners often used light verbs, extremely frequent verbs, such as take, make, do, be, get, and have, which can freely combine with a variety

32 of nouns. L2 learners hoped that their collocations would sound native-like by using such frequent verbs, although their attempts were not really successful.

(2) a. The former champion had an attempt to come back after a year’s retirement but failed. (made an attempt)

b. You sometimes have to get a risk to succeed in an undertaking. (take/run a risk)

c. To take her justice, Julie is a better cook than Helen. (to do her justice)

d. Shall we do the carpet here? (lay the carpet)

(3) L1 TRANSFER: L2 learners produced some inter-lingual errors in which they created

collocations that were obviously literal translations of Japanese collocations involving

verbs with the same or quite similar meanings. Most of the collocations of this type did

not get high scores on either Meaning or Naturalness. In particular, the collocations that

originated from Japanese figurative expressions such as me-ni tomaru (literally ‘to stop

one’s eyes’ which means to catch one’s eye) and iki-o korosu (literally ‘to kill one’s

breath’ which means to hold one’s breath) were nearly unintelligible by native speakers.

(3) a. She met an accident on her way home from school. (had an accident < jiko-ni au, literal translation in English: ‘meet an accident’)

b. All the students sent Prof. Sumitomo a big hand when his final lecture ended. (gave a big hand < hakushu-o okuru ‘send applause’)

c. Seeing the horrible scene of the accident, he blued color. (changed/lost color < aozameru ‘turn blue’)

d. Please foam the eggs. (beat the eggs < tamago-o awadateru ‘make-foam the eggs’)

(4) USE OF WORDS OTHER THAN VERBS: Low-intermediate learners sometimes used words other than verbs. They commonly used prepositions but also used adjectives,

adverbs, conjunctions, and auxiliary verbs.

(4) a. I often up a kite when I was a child. (flew a kite)

b. She often off class. (cuts/skips class)

c. I’ll come to your place if it good your convenience. (met your convenience)

d. To average her justice, Julie is a better cook than Helen. (to do her justice)

(5) MIXING WITH OTHER COLLOCATIONS: There were a few collocations in which

L2 learners mixed the target collocation with another collocation which involves one or

more common word with the target collocation and usually has quite similar meaning.

(5) a. None your own business. (‘None of your business’ + ‘Mind your own business’)

b. May I help a favor of you? (‘May I help you?’ + ‘May I ask a favor of you?’)

c. May I do a favor of you? (‘Would you do me a favor?’ + ‘May I ask a favor of you?’)

d. Donna committed her life when she was 20 years old. (‘commit suicide’ + ‘take her life’)

(6) SYNONYMY OF NOUNS: There were a few instances in which L2 learners produced

morphological synonymy of nouns. Most of the collocations of this type were rejected

by native speakers in both Grammaticality and Meaning Conveyed.

(6) a. We need to gather more information before we act action. (take action)

b. To judge her justice, Julie is a better cook than Helen. (to do her justice)

c. The fish negotiations progressed rapid progress last week. (made rapid progress)

d. You can judge a man’s company by the company he accompanies. (by the

company he keeps)

As shown in Table 2-A and Figure 2-A, paraphrasing and misuse of light verbs are predominant strategies employed by L2 learners. These two types of errors amount to

82.49% of the total errors committed by L2 learners. Since paraphrasing and misuse of light verbs are the most commonly used strategies, I further examined native speakers’ reactions to their erroneous forms.

Figures 2-1 and 2-2 display the percentages of scores on Grammaticality Judgment and Meaning Conveyed by native speakers when learners produced paraphrasing. Native speakers rated the collocations using a 5-point-scale: 5=highest (native-like), 1=lowest

(not acceptable).

60% 4- 53.33%

50% 3- 2-

1- 38.94% 40%

30% 26.33%

19.08% 20% 17.56% 15.65% 15.55% 13.56%

10%

0% Low-intermediate Advanced

Figure 2-1. Grammaticality Judgment (5-point-scale): Paraphrasing.

60.01% 60% 4-

50% 3- 2- 1- 40%

31.30% 28.89% 30% 24.43% 25.19%

19.08% 20%

10% 8.88%

2.22% 0% Low-intermediate Advanced

Figure 2-2. Meaning Conveyed (5-point-scale): Paraphrasing.

It is evident that advanced learners’ paraphrasing had much higher scores in both

Grammaticality Judgment and Meaning Conveyed (mean scores of Grammaticality

Judgment: low-intermediate = 2.54, advanced = 3.24, and mean scores of Meaning

Conveyed: low-intermediate = 2.89, advanced = 3.83). These differences were

statistically significant (Grammaticality Judgment: χ2 = 32.677, df = 1, p < .000; Meaning

Conveyed: χ2 = 29.124, df = 1, p < .000). Low-intermediate learners’ paraphrasing was

often regarded as unnatural and did not convey intended meanings while many advanced

learners’ paraphrasing was regarded as more native-like and successfully conveyed the

intended semantic content.

Figures 2-3 and 2-4 show the percentages of scores on Grammaticality Judgment

and Meaning Conveyed when light verbs were misused by low-intermediate learners and

advanced learners.

60% 4- 3- 50% 2- 1- 41.56% 40% 38.47%

28.77% 30% 26.64% 25.11% 21.83% 20%

9.99% 10% 4.56%

0% Low-intermediate Advanced

Figure 2-3. Grammaticality Judgment (5-point-scale): Misuse of light verbs.

60% 4- 3- 50% 2- 1- 38.96% 40% 35.16% 33.30% 30.60% 30% 26.48%

20% 15.46%

9.99% 10% 7.76%

0% Low-intermediate Advanced

Figure 2-4. Meaning Conveyed (5-point-scale): Misuse of light verbs.

Here, the results were quite different from paraphrasing. Advanced learners’ paraphrasing was quite successful while low-intermediate learners still have problems. In contrast, misuse of light verbs in collocations by both advanced and low-intermediate learners is generally regarded as unnatural or non-native-like and does not convey the intended meaning. The mean scores of Grammaticality Judgment of low-intermediate and advanced learners were 2.31 and 2.71 respectively. This difference is not statistically

significant. Figure 2-4 shows that advanced learners were somewhat more successful in

conveying the intended meanings when they misused light verbs (mean scores of

Meaning Conveyed: low-intermediate = 2.61, advanced = 3.03), but the difference was

not statistically significant.

2.4. Summary: Implication for the experiments

These results show trends in L2 learners’ performance in V + NP collocations

which are, on the whole, consistent with my hypotheses.

Hypothesis 1, that L2 learners will produce various types of intra-lingual and

inter-lingual errors, was supported. Six types of errors were found in this survey.

Hypothesis 2: improvement in performance was observed at higher levels;

35.04 % of low-intermediate learners’ collocations were judged as acceptable while

71.79% of advanced learners’ collocations were acceptable.

Hypothesis 3, that many of the L2 learners’ erroneous collocations do not convey

the intended semantic content was also supported, particularly when learners produced

collocations which involved misuse of light verbs.

This survey suggested some implications for further research in collocations.

First of all, the learners indeed use their L1 knowledge when they produce collocations.

In this survey, I only tested collocations which did not have literal Japanese equivalents;

therefore, positive transfer of L1 knowledge was not tested. It is possible that my survey

underestimated amounts of L1 transfer. Mohmoud (2005) collected essays written by

Arabic learners of English. He found that sixty-one percent of the learners’ collocational

errors were due to negative transfer from Arabic. As Bahns’ study (1993) reveals, it is

38 worth pursuing both positive and negative L1 transfer.

Second, the frequent use of paraphrases indicates that learners are not aware of the fact that words which have similar meanings may have different collocational restrictions and that they can only occur in specific combinations. Such results motivate further investigation of collocation, involving instruction which would help learners understand overlapping collocational restrictions between words which have similar meanings and correct their mistaken over- or under-extensions.

Third, the high percentage of misuse of light verbs suggests that learners have difficulty with light verbs. In productions such as take a decision, have a chat, give a scream, and make a note, all the meaning seems to be in the noun, and the verb can be considered semantically bleached. It has been claimed that L2 learners often find it difficult to learn grammatical, lexical, or phonological features which do not have single clear meanings. I think the same is true for collocations, and I believe that this is an area where much work needs to be done. Therefore, in my study I decided to examine how L2 learners deal with collocations that involve light verbs.

Fourth, my results showed a correlation between general proficiency in English and productive knowledge of collocations. An improvement in performance was observed in the advanced learners’ performance. However, it was found that even advanced learners still have problems with collocations. So it will be informative to continue collecting data from advanced learners as well as from low-intermediate learners.

In the next section, I will describe the design of my experiments which are presented in two parts. In Part I, I investigate the roles of frequency and meaning in

39 acquiring Verb+Noun collocations, and in part II, I examine the effects of instruction.

Each part contains two experiments.

CHAPTER 3

Experiments: Part I. The roles of frequency and meaning in acquiring Verb + Noun

collocations

3.1. Purpose of frequency and meaning studies: research questions and hypotheses

The purpose of these two experiments is to investigate the relationship between

ESL learners’ proficiency level and their collocational competence in frequent vs. infrequent collocations, semantically transparent vs. abstract collocations, and the effect of interference of the learner’s L1 in learning L2 collocations. This study explores the following eight Japanese light verbs and corresponding English Verb + Noun collocations.

Japanese verb Corresponding English verb

naru to be/ to become suru to do aru to have toru to take eru to get ataeru to give ukeru to receive tsukuru to make

TABLE 3-1. Eight light verbs investigated in this study.

These eight light verbs were chosen because they are extremely frequent in both English and Japanese. These verbs have very similar semantic contents, but different collocational restrictions are sometimes imposed on the type of dependent noun the verb can take in English and Japanese. Various English textbooks, phrase books, and

collocation dictionaries, including Kizuka (1995, 2002), Konishi (1989), Terasawa et al.

(1993), Paul and Nishimura (1994, 1996), Okitsu (2000) and Seki (2001) suggest that

Japanese learners often make errors when they translate Japanese Light Verb + Noun collocations into English if they involve similar but different semantic fields than the corresponding English verbs.

When Japanese Verb + Noun collocations are translated into English Verb + Noun

collocations the translation types can be divided into the following three types.

Types of Does direct Verbs in English Verbs in Japanese collocations translational equivalent collocations are: collocations are: exist in L1? (Positive transfer is possible) A1 Same light verbs (be, Yes become, do, have, Always light verbs (With L1 equivalents) take, get, give, receive, (naru, suru, aru, toru,

make) eru, ataeru, ukeru, A2 tsukuru) No Different light verbs (Without L1 A3 equivalents) Content verbs

TABLE 3-2. Three collocational types.

Examples:

A1: anata-no miyaku-o toru (literal) you-gen pulse-acc take to take your pulse

A2: shaken-o ukeru (literal) exam-acc receive to take an exam

A3: hikaku-ni naru (literal) comparison-loc become to bear/stand comparison

In this study, I am not going to deal with Japanese Verb + Noun collocations which are translated into English using other types of constructions, such as Verb + Particle + Noun

Phrase, Verb + Prepositional Phrase, Verb + Noun Phrase + Prepositional Phrase and so forth, in order to focus on Verb + Noun collocations in both languages.

The research questions are:

Question 1: Does the learners’ performance correlate with the frequency of the target

collocations?

Question 2: Is the learners’ performance directly influenced by the literalness of the

target collocations?

Question 3: Do learners perform better if there is a translational equivalent in L1?

Question 4: Are English collocations involving light verbs more difficult that those

involving content verbs?

Question 5: Is there a significant difference between the performance of low-

intermediate and advanced learners?

Specific hypotheses are:

Hypothesis for Question 1: Learners will demonstrate better productive knowledge of

frequent collocations as compared to less frequent ones.

Hypothesis for Question 2: Learners will demonstrate better productive knowledge of

collocations which have literal meanings (e.g., take time, have an idea, make a

mistake) as compared to those which have metaphorical meanings (e.g., take wing,

have the brains, make history).

Hypothesis for Question 3: Learners will perform better if there is a translational

equivalent in L1.

Hypothesis for Question 4: Light verbs are more difficult than content verbs.

Hypothesis for Question 5: Advanced learners will perform significantly better than

low-intermediate learners.

3.2. General method

3.2.1. Subjects

The participants in this study were sixty-six Japanese learners of English, who

included graduate and undergraduate students at the University of Hawai‘i at Mānoa

(UH), and students studying English in the NICE (New Intensive Course in English)

program. All the participants were recruited on the UH campus. The subjects were selected from two proficiency groups based on their TOEFL scores: a low-intermediate group with TOEFL scores lower than PBT 550, CBT 213, or iBT 79, and an advanced group with TOEFL scores of PBT 580, CBT 237, or iBT 92 or higher. Thirty-five learners (eight males and twenty-seven females) were in the low-intermediate group, and thirty-one learners (six males and twenty-five females) were in the advanced group.

Their ages varied from 19 to 48 years, with the mean age being 23 years 7 months (23 years 6 months for low-intermediate subjects, and 23 years 8 months for advanced subjects). The subjects have lived in English-speaking countries for periods of from 1 month to 22 years, with the mean length of residence being 4 years 3 months (2 years 1 month for the low-intermediate group, and 6 years 8 months for the advanced group).

The data collection was done individually or in small groups of two to four people

44 at several places on the university campus, including group-study rooms in the library, cafeterias, the student lounge at the Campus Center, and the lanai of the NICE program building.

3.2.2. Stimuli and procedure

Sixty-six Verb+Noun collocations of Types A1, A2 and A3 were tested in this study. Thirty-six collocations were used for Experiment 1 which tested the role of frequency in acquiring Verb + Noun collocations. Thirty collocations were used for

Experiment 2 which investigated the role of meaning in acquiring Verb + Noun collocations.

The procedure was a fill-in-the-blank test where the object noun of the collocation was presented and the verb had to be produced by the subject. The subjects were asked to think of combinations which would match the Japanese translation given above the

English sentence, and produce as many combinations as they could think of. Five blank spaces were given for each question. Here is an example:

Example: 「試験を受けた」 I an examination yesterday.

自信小 ------ 自信大聞いたり読んだりした事ない----ある [ ] 1—2—3—4—5—6—7—8—9—10 1—2—3—4—5—6—7—8—9—10

[ ] 1—2—3—4—5—6—7—8—9—10 1—2—3—4—5—6—7—8—9—10

Then the subjects answered the following two questions using a 10-point scale:

Questions 1: Are you confident in your answer? / Do you think your answer is correct?

(10 = They are very sure that their answer is right; 1= They are very unsure whether

their answer is an appropriate combination).

Question 2: Have you ever heard or read this combination? (10 = They are very sure

they have heard or read this combination; 1= They are very unsure that they have heard

or read this combination.)

The first question was intended to elicit the level of the subjects’ confidence in their combinations, and the second question asked about the subjects’ familiarity with combinations. For the second question, the subjects were instructed that tense and aspect did not have to match the one they gave in their answer in the underlined form (e.g., take an examination, takes an examination, took an examination, will take an examination, have taken an examination, is taking an examination, etc.). Here are sample answers for these questions:

Example: 「試験を受けた」 I an examination yesterday. 自信小 ------ 自信大聞いたり読んだりした事ない---ある

[ took ] 1—2—3—4—5—6—7—8—9—⑩ 1—2—3—4—5—6—7—8—9—⑩

[ had ] 1—2—3—4—5—6—7—8—9— ⑩ 1—2—3—4—5—6—7—8—9—⑩

[ received ] 1—2—3—④—5—6—7—8—9—10 1—②—3—4—5—6—7—8—9—10

[ did ] 1—2—3—4—5—6—⑦—8—9—10 1—2—3— ④ —5—6—7—8—

9—10

[ ] 1—2—3—4—5—6—7—8—9—10 1—2—3—4—5—6—7—8—9—10

The subjects were allowed to spend as much time as they wanted to complete these tasks, and they spent between fifty minutes and two hours. Many of the low-intermediate learners took more time to complete the test than the advanced learners.

After this data had been collected, all the combinations produced by the subjects were given to six native speakers of English, who graded each combination using two 6- point scales: (1) Meaning: How well was the intended meaning conveyed? and (2)

Acceptability: Would you produce this phrase in the same context? Here is an example:

Example. Joshua got a passport in 1997, hoping to travel to the UK for a year.

Questions: (1) How well is the intended meaning conveyed when the underlined phrase is replaced with this phrase?

(2) Would you produce this phrase in the same context?

(1) Not conveyed --- Conveyed (2) Do not produce ----- Produce accepted a passport 1 — ② — 3 — 4 — 5 — 6 ①— 2 — 3 — 4 — 5 — 6 acquired a passport 1 — 2 — 3 — 4 — 5 — ⑥ 1 — 2 — 3 — 4 — 5 — ⑥

bought a passport 1 — 2 — ③ — 4 — 5 — 6 1 — 2 — ③ — 4 — 5 — 6

created a passport ① — 2 — 3 — 4 — 5 — 6 ① — 2 — 3 — 4 — 5 — 6 gained a passport 1 — 2 — 3 — ④ — 5 — 6 1 — 2 — ③ — 4 — 5 — 6

issued a passport 1 — ② — 3 — 4 — 5 — 6 ① — 2 — 3 — 4 — 5 — 6 made a passport ① — 2 — 3 — 4 — 5 — 6 ① — 2 — 3 — 4 — 5 — 6 obtained a passport 1 — 2 — 3 — 4 — 5 — ⑥ 1 — 2 — 3 — 4 — ⑤ — 6

picked a passport 1 — ② — 3 — 4 — 5 — 6 1 — ② — 3 — 4 — 5 — 6

prepared a passport 1 — ② — 3 — 4 — 5 — 6 ① — 2 — 3 — 4 — 5 — 6 received a passport 1 — 2 — 3 — 4 — ⑤ — 6 1 — 2 — 3 — 4 — ⑤ — 6

reserved a passport 1 — ② — 3 — 4 — 5 — 6 ① — 2 — 3 — 4 — 5 — 6 retrieved a passport 1 — 2 — ③ — 4 — 5 — 6 1 — ② — 3 — 4 — 5 — 6

took a passport 1 — ② — 3 — 4 — 5 — 6 ① — 2 — 3 — 4 — 5 — 6

All the native speakers were recruited on the UH campus. There were five females and

one male. There were four undergraduate students, one graduate student, and one post

doctorate researcher. Their ages varied from 19 to 45 years, with the mean age being 25

years 7 months. None of the native speakers had studied Japanese, nor were there

Japanese native speakers in their family. The native speakers worked on grading at home

and returned the completed survey within two weeks.

CHAPTER 4

Experiment 1: The role of frequency in acquiring Verb+Noun collocations

4.1. Stimuli

Thirty-six Verb+Noun collocations were selected from the British National

Corpus (BNC) on the basis of their frequency. The BNC is a 100 million word collection of samples of written and spoken language from a wide range of sources, designed to represent a wide cross-section of current British English, both spoken and written. In this experiment, a collocation was considered to be highly frequent when the BNC had more than

150 instances that involved the collocation, while a collocation was regarded as infrequent when there were fewer than 14 instances.

The transparency of meaning of the collocations was rated by four native speakers of

English using a 10-point scale (10 = transparent, 1= not transparent). All the collocations

used in this experiment had average scores over 6.0, i.e., their meanings are relatively

transparent or literal.

The thirty-six collocations were divided into the following six types: (1) Highly

frequent phrases of Type A1; (2) Infrequent phrases of Type A1; (3) Highly frequent phrases of Type A2; (4) Infrequent phrases of Type A2; (5) Highly frequent phrases of

Type A3; and (6) Infrequent phrases of Type A3. Six tokens were prepared for each type.

VERB TYPE FREQUENCY MEANING EXAMPLES A1 same light verb in Frequent Get a job English and Japanese Infrequent Have a cut Relatively A2 different light verb Frequent literal Have lunch (Take lunch in in English and Japanese) Japanese Infrequent Make a discovery (Do a discovery in Japanese) A3 content verb in Frequent Fall victim (Become victim in English, and light Japanese) verb in Japanese Infrequent Cast a ballot (Do a ballot in Japanese) TABLE 4-1. Six collocational types for Experiment 1.

All the test sentences are listed in Appendix III. Before Experiment 1 all the subjects

filled out a questionnaire regarding their background information (age, gender, school,

length of stay in English-speaking countries, languages (English, Japanese, etc) they use

in their daily life, etc). See Appendix IV for a copy of the questionnaire.

4.2. Overall results and single variable analyses of results for the numbers of collocations produced by Japanese learners

A total of 5056 responses was collected from sixty-six subjects in Experiment I.

All the combinations collected from Experiment 1 are given in Appendix V. Table 4-A

and Figure 4-A show the numbers of responses the subjects of each proficiency group

gave for the six collocation types.

Low-intermediate Advanced learners Total (# = 66) Types of collocations learners (# = 35) (# = 31) Token avg/psn Token Avg/psn token avg/psn A1 Frequent 419 2.00 410 2.20 829 2.09 same light V Infrequent 374 1.78 398 2.14 772 1.95 A2 Frequent 452 2.15 405 2.18 857 2.16 different light V Infrequent 471 2.24 418 2.25 889 2.24 A3 Frequent 419 2.00 376 2.02 795 2.01 content V Infrequent 453 2.16 461 2.48 914 2.31 Total (# = 66) 2588 2.05 2468 2.21 5056 2.13 TABLE 4-A. Number of responses produced by the subjects in Experiment 1.

low-intermediate

s advanced 2.48 2.5 2.25 total 2.31 2.18 2.2 2.24 2.24 2.14 2.15 2.16 2.16 2.09 2.02 2 2 2.01 2 1.95 1.78

1.5 b e r o f b i c m o n at i o n 1

0.5 v e r a g e n u m u r v e e n a g A 0 frequent Infrequent frequent infrequent frequent infrequent A 1 A 2 A 3 Figure 4-A. Number of combinations produced by subjects.

The subjects produced an average of 2.13 combinations per question; low-

intermediate learners produced an average of 2.05 combinations, while advanced learners

averaged 2.21 combinations. A one-way Chi-square test shows this difference is

statistically significant (χ2 = 6.82, df = 1, p = 0.009). For Types A1 and A2 there was no statistically significant difference in the number of combinations produced, either

between the two proficiency groups or between frequent vs. infrequent collocations.

For Type A3 collocations, some statistical differences were observed. All subjects

averaged 2.01 combinations for frequent collocations but 2.31 combinations for

infrequent collocations. This difference is significant (χ2 = 8.14, df = 1, p = 0.0043). In addition, advanced learners produced significantly more responses for infrequent Type

A3 collocations (2.48) than for frequent ones (2.02) (χ2 = 8.44, df = 1, p = 0.0037).

Advanced learners produced more combinations for Infrequent Type A3 collocations

(2.48) than did low-intermediate learners (2.16) (χ2 = 4.28, df = 1, p = 0.0386).

Table 4-B and Figure 4-B compare the numbers of frequent vs. infrequent

combinations of Types A1, A2, and A3 produced by low-intermediate and advanced

learners’ groups.

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token avg/psn token avg/psn token avg/psn Frequent collocations 1290 2.05 1191 2.13 2481 2.09 (freq A1, A2, A3) Infrequent collocations 1298 2.06 1277 2.29 2575 2.17 (infreq A1, A2, A3) TABLE 4-B. Number of responses produced by the subjects for frequent vs. infrequent collocations of Types A1, A2 and A3.

2.3 2.29

2.2 2.17 2.13 2.09 2.1 2.05 2.06

1.9

1.8

1.7 low-intermediate advanced total 1.6 A v e rA b e r a g e b i o f n u m c o m n a t i o n s

1.5 Frequent collocations Infrequent collocations

Figure 4-B. Average number of combinations of frequent vs. infrequent collocations of Types A1, A2, and A3.

In total, the subjects produced an average of 2.09 combinations for frequent collocations

(low-intermediate: 2.05, advanced: 2.13) and 2.17 combinations for infrequent collocations (low-intermediate: 2.06, advanced: 2.29). A significant difference was found for infrequent collocations between low-intermediate (2.06) and advanced learners (2.29)

(χ2 = 7, df = 1, p = 0.0082). There was no statistically significant difference between low-intermediate and advanced learners in frequent collocations.

Table 4-C and Figure 4-C below show the numbers of combinations for collocations with vs. without equivalent collocations in the learners’ L1 Japanese in

Types A1, A2 and A3.

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token avg/psn token avg/psn Token avg/psn With L1 equivalents (A1) 793 1.89 808 2.17 1601 2.03 Without L1 equivalents (A2, A3) 1795 2.14 1660 2.23 3455 2.19 TABLE 4-C. Number of responses produced by the subjects for collocations with vs. without equivalent collocations in Japanese for Types A1, A2, and A3.

2.3 2.23 2.19 2.2 2.17 2.14

2.1 2.03

1.89 1.9

1.8

1.7 low-intermediate advanced total 1.6 A v A e r a g e b e n u rm o f b i c o m n a t i o n s

1.5 With equivalent collocation in Without equivalent collocation in Japanese Japanese

Figure 4-C. Average number of combinations with vs. without equivalent collocations in

Japanese for Types A1, A2 and A3.

Collocations with L1 equivalents are Type A1 and collocations without L1

equivalents are Types A2 and A3. The learners produced on average 2.03 combinations

for collocations which have equivalent collocations in Japanese (low-intermediate: 1.89,

advanced: 2.17) and 2.19 combinations for collocations with Japanese equivalents (low-

intermediate: 1.41, advanced: 2.23). They produced significantly more combinations for

collocations without L1 equivalents (χ2 = 4.01, df = 1, p = 0.0452). Low-intermediate learners produced more combinations for collocations without equivalents than for those

with L1 equivalents (χ2 = 6.4, df = 1, p = 0.0114). Advanced learners produced

significantly more combinations for collocations with L1 equivalents than did low-

intermediate learners (χ2 = 7.59, df = 1, p = 0.0059). No significant difference was

observed between the low-intermediate and advanced groups in the numbers of

collocations produced when there were no equivalents in Japanese, or in the advanced learners’ production of collocations with vs. without L1 equivalents.

Table 4-D and Figure 4-D display the numbers of combinations produced for

Light verb + Noun collocations and Content Verb + Noun collocations.

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token avg/psn token avg/psn token avg/psn Light Verb + Noun (A1, A2) 1716 2.04 1631 2.19 3347 2.12 Content Verb + Noun (A3) 872 2.08 837 2.25 1709 2.17 TABLE 4-D. Number of responses produced by the subjects for collocations with light verbs vs. content verbs of Types A1, A2, and A3.

2.3 2.25 2.19 2.2 2.17 2.12 2.1 2.08 2.04

1.9

1.8

1.7 low-intermediate advanced total 1.6 A v e r a g e n u m b e r o f c o m b i e r a t o f b m n i c o u m n s v e r a g e n o A

1.5 Light verb + Noun Content verb + Noun

Figure 4-D. Average number of combinations of Light verb + Noun collocations and Content verb + Noun collocations of Types A1, A2, and A3.

Light verb collocations are Types A1 and A2, and content verb collocations are Type A3.

The learners produced on average 2.12 Light Verb + Noun collocations (low- intermediate: 2.04, advanced: 2.19) and 2.17 Content Verb + Noun collocations (low- intermediate: 2.08, advanced: 2.25). Advanced learners produced significantly more combinations for Light Verb + Noun collocations than did low-intermediate learners (χ2 =

3.96, df = 1, p = 0.0466). No significant difference was observed between low- intermediate and advanced learners’ performance or for low-intermediate learners’ performance in Light Verb vs. Content Verb collocations.

Finally, Table 4-E and Figure 4-E present the number of responses produced by the subjects for collocations with same vs. different light verbs.

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token avg/psn token avg/psn token avg/psn Same Light Verb + Noun (A1) 793 1.89 808 2.17 1601 2.02 Different Light Verb + Noun (A2) 923 2.20 823 2.21 1746 2.20 TABLE 4-E. Number of responses produced by the subjects for collocations with same light verbs vs. different verbs.

2.3

2.2 2.21 2.2 2.2 2.17

2.1 2.02 2

1.89 1.9

1.8

1.7 low-intermediate advanced total 1.6 A v eb rA ao f u m b i g e n c a o m t n i o n s

1.5 Same light verb + Noun Different light verb + Noun

Figure 4-E. Average number of combinations of same vs. different Light Verb + Noun for Types A1, A2 and A3.

Collocations of Type A1 are grouped as Same Light Verb + Noun, while collocations of

Type A2 are grouped as Different Light Verb + Noun. The average numbers of Same

Light Verb + Noun collocations were 2.02 (low-intermediate: 1.89, advanced: 2.17) and the average numbers of Different Light Verb + Noun were 2.20 (low-intermediate: 2.20, advanced: 2.21). Overall, the learners produced significantly more collocations of the

Different Light Verb + Noun type as compared to the Same Light Verb + Noun type (χ2 =

6.2, df = 1, p = 0.0128). Advanced learners produced significantly more Same Light Verb

+ Noun collocations than low-intermediate learners did (χ2 = 9.7, df = 1, p = 0.0018). No

significant difference was observed between low-intermediate and advanced learners in

the numbers of Different Light Verb + Noun collocations.

In summary, advanced learners generally tended to produce slightly more combinations than low-intermediate learners. The difference was significant for

infrequent collocations, for collocations with equivalents in L1, for Light Verb + Noun collocations and for Same Light Verb + Noun collocations. These results suggested that advanced learners are more able to challenge themselves to produce a variety of

56 combinations while low-intermediate learners are more careful or less confident, and hesitate to produce many combinations.

On the whole, the learners produced more attempts for infrequent collocations, collocations without equivalent collocations in Japanese, Content Verb + Noun collocations and Different Light Verb + Noun collocations than for frequent collocations, collocations with equivalent collocations in Japanese, Light Verb + Noun collocations and Same Light Verb + Noun collocations. Significant differences were observed in the learners’ overall performance in collocations with vs. without L1 equivalents, low- intermediate learners’ performance in collocations with vs. without L1 equivalents, and advanced learners’ performance in Light Verb + Noun collocations vs. Content Verb +

Noun collocations.

In the next section, we consider the native speakers’ Grammaticality Judgments of the collocations produced by the learners. First, we examine combined average scores of native speakers’ judgments of the grammaticality of each collocation type, followed by analyses using three- and then two-scales to better understand learners’ performance.

4.3. Native speakers’ Grammaticality Judgments

4.3.1. Average scores of native speakers’ Grammaticality Judgments

Table 4-F and Figure 4-F show the average scores of native speakers’

Grammatical Judgments for each type of collocation using a 6-point scale.

Type of collocations Low-intermediate Advanced learners Total learners (# = 35) (# = 31) (# = 66) A1 Frequent 4.20 3.79 4.00 same light V Infrequent 3.49 3.48 3.49 A2 Frequent 3.71 3.91 3.81 different light V Infrequent 2.89 3.38 3.12 A3 Frequent 2.87 3.38 3.11 content V Infrequent 2.65 3.62 3.14 Total (# = 66) 3.29 3.60 3.44 TABLE 4-F. Native speakers’ Grammaticality Judgments. Experiment 1 (6-point scale).

5 low-intermediate 4.5 advanced 4.2 total 4 3.91 4 3.79 3.81 3.71 3.49 3.62 3.48 3.49 3.5 3.38 3.38 3.12 3.11 3.14 3 2.89 2.87 2.65

2.5

1.5 a t i c a l t i e n ( t m 6 g - d y J u i o p t n - s c a l e )

m 1 frequent infrequent frequent infrequent frequent infrequent

G ra m A 1 A 2 A 3

Figure 4-F. Native speakers’ Grammaticality Judgments (6-point scale).

The average of judgments of the performance by advanced learners was higher (3.60) than for low-intermediate learners (3.29). The averages for frequent collocations of types

A1 and A2 were higher (A1: 4.00 and A2: 3.81) than for infrequent collocations (A1:

3.49 and A2: 3.12). The average for infrequent combinations of Type A3 was slightly higher (3.14) than for frequent combinations (3.11).

Table 4-G and Figure 4-G compare judgments of frequent vs. infrequent collocations.

Types of Low-intermediate Advanced Total Collocation learners (# = 35) learners (# = 31) (# = 66) Frequent collocations (Freq A1, A2, A3) 3.60 3.70 3.65 Infrequent collocations (Infreq A1, A2, A3) 2.97 3.50 3.24 TABLE 4-G. Native speakers’ Grammaticality Judgments for frequent vs. infrequent collocations. (6-point scale)

5 low-intermediate 4.5 advanced total

4 3.67 3.7 3.65 3.5 3.5 3.24 2.97 3

2.5

1.5

1 G r a m m a t I c a l i t y J u d g m e n t ( 6 - p o I n t - s c a l e ) s c a l - n t p o I y J u d 6 - g m t ( i e n t a m c a l I m r G a t Frequent collocations Infrequent collocations

Figure 4-G. Native speakers’ Grammaticality Judgments for frequent vs. infrequent collocations (6-point scale).

The average for frequent collocations was higher (3.65) than that for infrequent collocations (3.24). Scores for advanced learners were higher than for low-intermediate learners for both frequent and infrequent collocations.

Next, Table 4-H and Figure 4-H show the average Grammaticality Judgment scores for all types of collocations which do vs. do not have Japanese equivalent collocations.

Types of Low-intermediate Advanced Total Collocation learners (# = 35) learners (# = 31) (# = 66) With L1 equivalent (A1) 3.85 3.64 3.75 Without L1 equivalent (A2, A3) 3.03 3.57 3.30 TABLE 4-H. Native speakers’ Grammaticality Judgments for collocations with vs. without L1 equivalents. (6-point scale)

5 low-intermediate 4.5 advanced total 4 3.85 3.75 3.64 3.57 3.5 3.3 3.03 3

2.5

1.5

G r a m m a t I c a l i t y J u d g m e n t ( 6 - p o I n t - s c a l e ) c a t - s I n o p ( 6 - t e n m g d J u a l t I i t y c a m a m r G With equivalent collocation in Without equivalent collocation in Japanese Japanese

Figure 4-H. Native speakers’ Grammaticality Judgments for collocations with vs. without L1 equivalents. (6-point scale).

Collocations with equivalent translation in Japanese are Type A1, and collocations without equivalent translation are Types A2 and A3. Both low-intermediate and advanced learners did better with collocations which have equivalent collocations in

Japanese than with those which do not have equivalent collocations in Japanese. The low-intermediate learners’ score for collocations with equivalent collocations in Japanese was higher than the advanced learners’ score, while the advanced learners’ score was higher than the low-intermediate learners’ score in collocations without equivalent collocations in Japanese.

Next, Table 4-I and Figure 4-I show the average scores of Light Verb + Noun collocations and Content Verb + Noun collocations.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Light Verb + Noun (A1, A2) 3.57 3.64 3.61 Content Verb + Noun (A3) 2.76 3.50 3.13 TABLE 4-I. Native speakers’ Grammaticality Judgments for collocations with light verbs vs. content verbs. (6-point scale)

5 low-intermediate 4.5 advanced total

4 3.64 3.61 3.57 3.5 3.5 3.13 3 2.76

2.5

1.5

1 G r a m m a t I c a l i t y J u d g m e n t ( 6 - p o I n t - s c a l e ) l a s c t - n I o p - ( 6 t n e m g d u J y i t l c a I a t m a m r G Light verb + Noun Content verb + Noun

Figure 4-I. Native speakers’ Grammaticality Judgments for light verb + noun vs. content verb + noun collocations (6-point scale).

Both low-intermediate and advanced learners attained higher grammaticality scores for

Light Verb + Noun collocations (3.61) than for Content Verb + Noun collocations (3.13).

Scores for advanced learners were higher than those of low-intermediate learners for both

Light Verb + Noun and Content Verb + Noun collocations.

Finally, Table 4-J and Figure 4-J present average grammaticality scores for Same

Light Verb + Noun and Different Light Verb + Noun.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Same Light Verb + Noun (A1) 3.85 3.64 3.75 Different Light Verb + Noun (A2) 3.3 3.65 3.47 TABLE 4-J. Native speakers’ Grammaticality Judgments for collocations with same vs. different light verbs. (6-point scale)

5 low-intermediate 4.5 advanced total

4 3.85 3.75 3.64 3.65 3.47 3.5 3.3

2.5

1.5

1 G r a m m a t t I e n ( Ia l m 6 - c o t p n - i m l g rm s c a a t d e ) J u y G Same light verb + Noun Different light verb + Noun

Figure 4-J. Average scores of Native speakers’ Grammaticality Judgments for same vs. different light verb collocations (6-point-scales).

Low-intermediate learners attained higher scores for Same light verb + Noun collocations

(3.85) than for Different light verb + Noun collocations (3.30), while advanced learners

did slightly better on different light verb + Noun collocations (3.65) than Same light verb

+ Noun collocations (3.64). The total score was higher for Same light verb + Noun

collocations (3.75) than that for Different light verb + Noun collocations (3.47).

In summary, the learners attained higher scores for the following collocation

types: (1) Frequent collocations, (2) Collocations with Japanese equivalents, (3) Light verb + Noun collocations, and (4) Same light verb + Noun collocations. These results are in reverse proportion to the numbers of collocations produced, which we reviewed in the

section 4.2. The fewer combinations learners produce, the higher their score for

Grammaticality, and vice versa. We will come back to this discussion in section 4.4.

4.3.2. Three-scale analysis of Grammaticality Judgments.

In this section, we examine the results of native speakers’ Grammaticality

Judgments more closely. The criteria used in this analysis were three scales; a

collocation was judged acceptable when the native speakers rated it as 4.5 or higher on

the six-point scale; it was regarded neither acceptable or unacceptable when it was rated

as between 2.67 and 4.33; and a combination was judged unacceptable when it was rated

as 2.5 or lower.

This three-scale analysis helps better understand learners’ performance because,

as can be seen in Table VI-1 in Appendix VI and Figures 4-1, 4-2 and 4-3 below which

display the results from the grammaticality judgments of the low-intermediate and

advanced learners for each collocation type. The average scores of native speakers’

Grammaticality Judgments do not correspond to the most typical performance by

Japanese learners. Most commonly, many learners’ combinations were rated as either

‘acceptable’ or ‘not acceptable’ and there were relatively few combinations rated as

‘neither acceptable nor unacceptable’ although most of the average scores fall in the

‘neither acceptable nor unacceptable’ slot in the three-scale analysis.

70% low-intermediate 60.62 59.76 60% advanced

48.39 50% 43.72

38.44 36.93 40% 37.23 31.22

30%

19.35 20% 13.17 9.02 10% 2.15

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 A1 frequent A1 infrequent

Figure 4-1. Native speakers’ Grammatical Judgment A1 (6-point-scale).

70% low-intermediate advanced 60.61 60%

49.28 50% 45.27 42.92

37.83 37.32 40% 32.59

30% 24.68 22.14 19.25 20% 14.72 13.4

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 A2 frequent A2 infrequent

Figure 4-2. Native speakers’ Grammatical Judgment A2 (6-point-scale).

70% low-intermediate advanced 60% 54.89

47.81 49.45 50%

40% 33.06 33.61 31.61 22.31 26.73 28.7 30% 21.85 18.58 18.38 20%

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 A3 frequent A3 infrequent Figure 4-3. Native speakers’ Grammatical Judgment A3 (6-point-scale).

Almost all the graphs in Figures 4-1, 4-2 and 4-3 above have the shape of a hollow between two hills. This is because the majority of learners’ collocations were rated as either acceptable (4.5~6) or not acceptable (1~2.5), with very few being rated as neither acceptable nor unacceptable (2.67~4.33). However this trend was not observed for Type

A3 infrequent. Figure 4-4 below combines the results for all the collocation types: A1,

A2. A3.

70% low-intermediate advanced 60% 48.16% 50% 41.01% 39.71% 35.55% 40%

30% 19.28% 16.30% 20%

10%

0% 6~4.5 4.33~2.67 2.5~1 Total A1, A2, A3

Figure 4-4. Native speakers’ Grammatical Judgment. Total. A1, A2, A3.

There was a statistically significant difference in performance between low-intermediate and advanced learners (χ2 = 36.51, df = 2, p <.0001): Advanced learners had higher scores (41.01%) for acceptable collocations (4.5~6) than did low-intermediate learners

(35.55%), while low-intermediate learners had higher scores (48.16%) for unacceptable collocations than did advanced learners (39.71%). Low-intermediate learners had, in general, a slightly deeper hollow between the two hills.

Next, we examine the influence of four variables on Grammaticality Judgments:

(1) Frequency, (2) Existence of equivalent collocations in L1, (3) Light verb vs. Content verb, and (4) Same vs. Different light verb. Table VI-2 in Appendix VI shows the distributions of collocation tokens and percentages by the proportions of native speakers’

Grammaticality Judgments: high (4.5-6), mid (2.67-4.33) and low (1-2.5). Figures 4-5,

4-6, 4-7 and 4-8 show numbers of tokens and percentages of high, mid and low

Grammaticality scores.

70% low-intermediate advanced 60% 53.15%

50% 46.69% 43.41% 43.18% 42.35%

40% 36.84% 35.76%

30% 27.66% 21.88%

20% 16.47% 19.20% 13.41%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Frequent Infrequent Figure 4-5. Grammaticality Judgments. Frequent vs. infrequent collocations.

70% low-intermediate advanced 60%

50.19% 50.67% 48.51% 50% 42.48% 40.85% 37.33% 40% 37.38%

29.06% 30% 21.82% 20.27% 20% 14.11%

10% 7.33%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 With equivalent collocations in Japanese Without equivalent collocation in Japanese Figure 4-6. Grammaticality Judgments. With vs. without equivalent collocations in L1.

70% low-intermediate advanced 60% 52.06%

50% 46.16% 44.84%

41.35% 39.25% 40.61% 40% 33.45%

30% 25.94% 24.20% 23.74%

20% 15.91% 12.49%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Light verb + Noun Content verb + Noun Figure 4-7. Grammaticality Judgments. Light verb + noun vs. content verb + noun.

70% low-intermediate advanced 60% 50.19% 48.51% 49.34% 50% 42.48% 41.22% 41.10% 40% 37.38% 33.70%

30%

17.68% 20% 16.96% 14.11%

10% 7.33%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Same light V + N Different light V + N Figure 4-8. Grammaticality Judgment. Same light verb + noun vs. different light verb + noun.

Figures 4-5, 4-6, 4-7 and 4-8 show very similar trends. As we saw in the graphs of each collocation type A1, A2 and A3, all these figures are also characterized by a sharp v- shaped hollow between two hills. Again, this result suggests that native speakers’ judgments are quite clear-cut though there are some individual variations. The majority of learners’ productions were rated as either acceptable (4.5-6) or unacceptable (1-2.5), with not many being rated in between on the 6-point-scale.

Both low-intermediate and advanced learners had higher scores (High 4.5-6) for the following types of collocations: frequent ones, those with L1 equivalents, light verb +

Noun, and same light verb + Noun. Lower scores were given to the following types: infrequent collocations, collocations without L1 equivalents, content verb + Noun, and different light verb + Noun.

4.3.3. Two-scale analysis of Grammaticality Judgment

In this section, I also tried analyzing the data in two groups to compare the rates at acceptable collocations produced by the learners. I combined Mid (2.67-4.33) and Low

(1-2.5) ratings from native speakers’ Grammaticality Judgments into one proportional group and compared the rates at which collocations were judged to be acceptable (High

4.5-6) or not fully so (Mid/Low1-4.33). Chi-square tests were again used to see whether the percentage with data in a given range varied significantly between these four variable groups as well as between the two proficiency groups. I did four analyses focusing on each of the following four variables: frequency, existence of L1 equivalents, light vs. content verbs, and same vs. different light verbs.

First, I compared frequent vs. infrequent collocations. Table 4-K and Figure 4-K show the distributions of tokens and percentages of frequent and infrequent collocations divided into High and Mid-Low grammaticality groups.

Frequent collocations Infrequent collocations Grammaticality (frequent A1, A2, A3) (infrequent A1, A2, A3) Judgments by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 560 730 1290 356 931 1287 intermediate 43.41% 56.59% 27.66% 72.35% Advanced 550 628 1178 456 819 1275 46.69% 53.31% 35.76% 64.23% Total 1110 1358 2468 812 1750 2562 45.05% 54.95% 63.42% 68.29% TABLE 4-K. Grammaticality Judgments. Frequent vs. infrequent collocations.

72.35% 70% 64.23% low-intermediate 56.59% 60% advanced 53.31%

50% 46.69% 43.41%

40% 35.76%

27.66% 30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Frequent Infrequent

Figure 4-K. Grammaticality Judgment. Frequent vs. infrequent collocations.

Both low-intermediate and advanced learners did better with frequent than infrequent

collocations. The chi-square test shows that the difference was statistically significant

(Low-intermediate: χ2 = 69.752, df = 1, p = *0; Advanced: χ2 = 30.206, df = 1, p =

*<.0001; Total: χ2 = 93.922, df = 1, p = *0). With respect to infrequent collocations, advanced learners produced significantly more highly rated collocations than did low- intermediate learners (χ2 = 19.427, df = 1, p <.000). There was no significant difference

between advanced and low-intermediate learners in their performance on frequent

collocations (χ2 = 2.674, df = 1, p = 0.102).

In my second analysis I tested existence of L1 equivalents. Tokens and

percentages of collocations with vs. without equivalent collocations divided into High

and Mid-Low grammaticality groups are displayed in Table 4-L and Figure 4-L.

With equivalent collocations in L1 Without equivalent collocations in L1 Grammaticality (A1) (A2, A3) Judgments by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 397 394 791 519 1267 1786 intermediate 50.19% 49.81% 29.06% 70.94% Advanced 392 416 808 614 1031 1645 48.51% 51.49% 37.33% 62.85% Total 789 810 1599 1133 2298 3431 49.35% 50.65% 33.20% 66.90% TABLE 4-L. Grammaticality Judgments. Frequent vs. infrequent collocations.

70.94% 70% 62.85%

60% 50.19% 51.49% 48.51% 49.81% 50%

37.33% 40%

29.06% 30%

20% low-intermediate 10% advanced

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 With equivalent collocation in Japanese Without equivalent collocation in Japanese Figure 4-L. Grammaticality Judgment. Collocations with vs. without equivalent collocations in L1.

Both low-intermediate and advanced learners did significantly better with collocations with Japanese equivalents than on those without equivalents (Low-intermediate: χ2 =

106.836, df = 1, p = *0; Advanced: χ2 = 28.044, df = 1, p* <.000; Total: χ2 = 123.053, df

= 1, p = *0). Advanced learners produced significantly more highly rated collocations

without L1 equivalents than did low-intermediate learners (χ2 = 26.452, df = 1, p *<.000).

No statistically significant difference was observed in collocations with L1 equivalents

between low-intermediate and advance learners (χ2 = 0.449, df = 1, p = 0.503).

The third analysis compares content verbs vs. light verbs. Table 4-M and Figure

4-M compare low-intermediate and advanced learners’ performance in light verb + noun

collocations and content verb + noun collocations.

Grammaticality Light verb + Noun (A1, A2) Content verb + Noun (A3) Judgments by natives High Mid/Low Total High Mid/Low Total (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 705 1000 1705 211 661 872 intermediate 41.35% 58.65% 24.20% 75.40% Advanced 730 898 1628 276 549 825 44.84% 55.16% 33.45% 66.55% Total 1435 1898 3333 487 1210 1697 43.10% 56.91% 28.83% 70.98% TABLE 4-M. Grammaticality Judgments. Light verb + noun vs. content verb + noun.

75.40% 70% 66.55% 58.65% 60% 55.16%

50% 44.84% 41.35% 40% 33.45%

30% 24.20%

20% low-intermediate advanced 10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Light verb + Noun Content verb + Noun Figure 4-M. Grammaticality Judgment. Light verb + noun collocations vs. content verb + noun collocations.

Learners did significantly better with light verb + noun collocations than with content verb + noun collocations. The difference between light verb vs. content verb collocation types was statistically significant (Low-intermediate: χ2 = 74.081, df = 1, p = *0;

Advanced learners: χ2 = 29.340, df = 1, p = *<.000; Total: χ2 = 98.164, df = 1, p = *0).

The advanced learners’ performance was significantly better than that of low-

intermediate learners both on light verb collocations and content verb collocations (Light

verb + Noun: χ2 = 4.140, df = 1, p = *0.042; content verb + noun: χ2 = 17.754, df = 1, p =

*<.000).

In my fourth analysis, I tested same vs. different light verbs. Table 4-N and

Figure 4-N display distributions of native speakers’ Grammaticality of same vs. different

light verb + Noun collocations.

Same light verb (A1) Different light verb (A2) Grammaticality Judgments High Mid/Low Total High Mid/Low Total by natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 397 394 791 519 1267 1786 intermediate 50.19% 49.81% 33.70% 66.30% Advanced 392 416 808 614 1031 1645 48.51% 51.49% 41.22% 58.78% Total 789 810 1599 1133 2298 3431 49.35% 50.65% 37.46% 62.54% TABLE 4-N. Grammaticality Judgments. Same light verb + noun vs. different verb + noun.

70% 66.30%

58.78% 60% 50.19% 51.49% 48.51% 49.81% 50% 41.22%

40% 33.70%

30%

20% low-intermediate advanced 10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Same light verb + Noun Different light verb + Noun

Figure 4-N. Grammaticality Judgments. Same light verb + noun vs. different light verb + noun.

The learners produced significantly more highly-rated same light verb + noun

collocations than highly-rated different light verb + noun collocations (Low-intermediate:

χ2 = 106.836, df = 1, p = *0; Advanced: χ2 = 28.044, df = 1, p *<.000; Total: χ2 = 123.053,

df = 1, p = *0). Advanced learners’ performance with type different light verb + noun

was significantly better than low-intermediate learners’ performance (χ2 = 26.452, df = 1,

p *<.000). The difference between Low-intermediate and advanced learner’s

performance with same light verb + noun was statistically insignificant (χ2 = 0.449, df = 1,

p = 0.503).

In summary, Chi-square tests show that learners produced significantly more

acceptable collocations of the following types: frequent collocations, collocations with

L1 equivalents, light verb + noun, and same light verb + noun. They did more poorly in:

infrequent collocations, collocations without L1 equivalents, content verb + Noun, and

different light verb + noun. Advanced learners performed significantly better than low-

intermediate learners on all collocations with more difficult variable types (i.e. infrequent

collocations, collocations without equivalent collocations in L1, content verb + noun

collocations and different light verb + noun collocations. Among collocations with easier

variable types, advanced learners produced significantly more acceptable light verb +

noun collocations than low-intermediate learners, but no significant difference was

observed in other types: frequent collocations, collocations with L1 equivalents, and

same light verb + noun collocations.

4.3.4. Hierarchy of difficulty based on Grammaticality Judgments

In the previous sections, we used two- and three-scale analyses to capture learners’ performance on Verb + Noun collocations. I looked at the relationship between native speakers’ Grammaticality Judgments and the four attributes that influence learners’ performance. In this section, I attempt to extract a hierarchy of difficulty based on the percentages of collocations judged acceptable (4.5~6) when native speakers made

Grammaticality Judgments.

A series of Chi-square tests reveal the following difficulty hierarchy from the

73 easiest collocation type (A1 frequent) on the left through the most difficult collocation types (A2 infrequent, and A3 frequent and infrequent) on the right.

A2 infrequent (same light verb) A1 frequent A2 frequent A1 infrequent 30.68% (same light verb) (different light (different light A3 frequent 60.19% verb) verb) (content verb) (GJ=4.5-6) 44.03% 37.66% 29.94% A3 infrequent

(content verb) 27.63% Easier <------> More difficult TABLE 4-2. Difficulty hierarchy of collocations

The percentage of learners’ productions that were rated highly was significantly greater for the frequent A1 type (60.19%) than for the frequent A2 type (44.03%) (χ2 = 44.038; df

= 1; p* = 0). There were significantly more highly rated collocations of the frequent A2 types (44.03%) than for the infrequent A1 type (37.66%) (χ2 = 6.783; df = 1; p* = 0.0092).

Significantly more highly rated collocations were observed for infrequent A1 type

(37.66%) than for infrequent A2 types (30.68%) (χ2 = 8.925; df = 1; p* = 0.0028). The differences between the other three types (infrequent A2, frequent A3, infrequent A3) were not significant (A2 infrequent vs. A3 frequent: χ2 = 0.109; df = 1; p = 0.7413; A3 frequent vs. A3 infrequent: χ2 = 1.095; df = 1; p = 0.2954; A2 infrequent vs. A3 infrequent: χ2 = 2.018; df = 1; p = 0.1554).

This hierarchy shows that, in general, light verb collocations are easier for learners than content verb collocations. The percentages of highly rated collocations involving light verbs (A1 and A2) were higher than those of highly rated collocations involving content verbs (A3). Among light verb collocations, frequent collocations are

74 easier than infrequent collocations. However, among content verb collocations, there was no significant difference in learners’ performance between frequent and infrequent collocations. There was a lot of difference in the percentages of highly rated collocations within the frequent A3 type. There were 60 subjects (out of 66) who gave the target verb as their response in ask questions, 44 gave the target verb in keep an eye, and 43 in commit suicide, while only 5 subjects responded with the target verb in put a stop, 0 subjects gave the target verbs in stand trial and fell victim although a few subjects answered with alternative acceptable combinations, such as went on trial and went to trial.

One possible reason for this notable difference is that the frequency of collocations exhibited in learners’ textbooks and other pedagogical materials is probably different from the frequency information used here, drawn from the British National

Corpus. There are a few pedagogic corpora including a corpus (Chujo et al., 2007) which consists of 35 high school English textbooks published in Japan. Moreover, in the current experiment, only 6 tokens were prepared for each type of collocation. Thus, examining variations in learners’ performance within a particular type of collocations is beyond the scope of this study, but see Appendix 2 to see the ratings obtained in this experiment.

4.4. Adequacy of meaning conveyed

4.4.1. Average scores of native speakers’ judgments of Meaning Conveyed

Table 4-O and Figure 4-O show the average score of native speakers’ ratings when asked ‘How well is the intended meaning conveyed when a collocation is replaced with a phrase produced by a learner?’, for each type of collocation using a 6-point-scale.

Type of collocations Low-intermediate Advanced learners Total learners (# = 35) (# = 31) (# = 66) A1 Frequent 4.71 4.80 4.76 same light V Infrequent 3.91 3.88 3.90 A2 Frequent 4.32 4.50 4.41 different light V Infrequent 3.35 3.80 3.58 A3 Frequent 3.47 3.77 3.62 content V Infrequent 3.78 4.26 4.02 Total (# = 66) 3.92 4.17 4.05 TABLE 4-O. Native speakers’ judgments of Meaning Conveyed. Experiment I (6-point- scale).

5 4.8 4.71 4.76 low-intermediate 4.5 4.41 advanced 4.5 4.32 total 4.26 3.91 3.9 4.02 4 3.88 3.8 3.77 3.78 3.58 3.62 3.47 3.5 3.35

2.5

2 M e a n i n g ( 6 - p o i n t - s c a l e ) l e a c t - s i n o - p ( 6 g i n n a e M 1.5

1 frequent infrequent frequent infrequent frequent infrequent A 1 A 2 A 3

Figure 4-O. Native speakers’ judgments of Meaning Conveyed (6-point-scale).

The average judgments of performance by low-intermediate learners were lower (3.78) than for advanced learners (4.17). These meaning scores are higher than the average scores of native speakers’ Grammaticality Judgments (low-intermediate: 3.29; advanced:

3.60). The averages for frequent collocations of types A1 and A2 were higher (A1: 4.76 and A2: 4.41) than for infrequent collocations (A1: 3.90 and A2: 3.58). In contrast, the average for infrequent combinations of Type A3 was higher (4.02) than for frequent

combinations (3.62).

Table 4-P and Figure 4-P compare the meaning ratings of frequent vs. infrequent

collocations.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Frequent collocations (freq A1, A2, A3) 4.17 4.37 4.27 Infrequent collocations (infreq A1, A2, A3) 3.66 4.00 3.83 TABLE 4-P. Native speakers’ judgments of Meaning Conveyed: Frequent vs. infrequent collocations. (6-point-scale)

5 ) low-intermediate advanced 4.5 4.37 4.27 4.17 total 4 4 3.83 3.66

3.5

2.5

1.5 M e a n iM v o n e n g C y e ( d 6 - p o it n - s c a l e 1 Frequent collocations Infrequent collocations

Figure 4-P. Native speakers’ judgments of Meaning Conveyed: Frequent vs. infrequent collocations (6-point-scale)

The average for frequent collocations was higher (4.27) than for infrequent collocations

(3.83). Scores for advanced learners were higher than for low-intermediate learners for both frequent and infrequent collocations.

Next, Table 4-Q and Figure 4-Q show the average scores of Meaning Conveyed for all types of collocations with and without Japanese equivalents.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) With L1 equivalents (A1) 4.31 4.34 4.33 Without L1 equivalents (A2, A3) 3.73 4.08 3.91 TABLE 4-Q. Native speakers’ judgments of Meaning Conveyed: Collocations with vs. without Japanese equivalents. (6-point-scale)

5 ) low-intermediate advanced 4.5 4.31 4.34 4.33 total 4.08 4 3.91 3.73

3.5

2.5

1.5 M e a n i n g C o n v e y e d ( d y e v e - 6 i n o o t p n i C - n e a g n a ls c M e 1 With translation Without translation

Figure 4-Q. Native speakers’ judgments of Meaning Conveyed: Collocations with vs. without Japanese equivalents. (6-point-scale)

Both low-intermediate and advanced learners attained higher scores for collocations with

L1 equivalents (low-intermediate: 4.31; advanced: 4.34) than for those without L1 (low- intermediate: 3.73; advanced: 4.08). Judgment scores for advanced learners’ production of collocations both with and without L1 equivalents were higher than for low- intermediate learners.

Next, Table 4-R and Figure 4-R show the average Meaning Conveyed scores for

‘Light verb + Noun’ and ‘Content verb + Noun’ collocations.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Light verb + Noun (A1, A2) 4.07 4.25 4.16 Content verb + Noun (A3) 3.63 4.01 3.82 TABLE 4-R. Native speakers’ judgments of Meaning Conveyed for ‘light verb + noun’ vs. ‘content verb + noun’ collocations (6-point-scale)

5 ) low-intermediate advanced 4.5 4.25 total 4.16 4.07 4.01 4 3.82 3.63 3.5

2.5 o n v e y e d ( v e y d n o 6 - i o p t n - s c a l e 2

1.5 e a n i n g C M 1 Light verb + Noun Content verb + Noun

Figure 4-R. Native speakers’ judgments of Meaning Conveyed for ‘light verb + noun’ vs. ‘content verb + noun’ collocations (6-point-scale)

Both low-intermediate and advanced learners attained higher Meaning scores for ‘Light verb + Noun’ collocations (low-intermediate: 4.07; advanced: 4.25) than for ‘Content verb + Noun’ collocations (low-intermediate: 3.63; advanced: 4.01). Scores for advanced learners were higher than for low-intermediate learners for both set of collocations.

Finally, Table 4-S and Figure 4-S present average Meaning Conveyed scores for collocations with the same vs. different light verb.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Same light verb + Noun (A1) 4.31 4.34 4.33 Different light verb + Noun (A2) 3.84 4.15 4.00 TABLE 4-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs. (6-point-scale)

5 low-intermediate ) advanced total 4.5 4.31 4.34 4.33 4.15 4 4 3.84

3.5

2.5

1.5 e a n i n g C o n v e y e d (yev e d n -6 o i i o e a n C p t g n n - s c a l e M 1 Same light verb + Noun Different light verb + Noun Figure 4-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs (6-point-scale)

The learners attained higher scores when the light verb was the same across languages

(low-intermediate: 4.31; advanced: 4.34) than when it differed (low-intermediate: 3.84; advanced: 4.15). Advanced learners were more successful on both types than were low- intermediate learners.

In summary the learners attained higher scores for the following collocation types: (1) Frequent collocations, (2) Collocations with Japanese equivalents, (3) ‘Light verb + Noun’ collocations, and (4) ‘Same light verb + Noun’ collocations. These results are the same as for native speakers’ Grammaticality Judgments. In general, scores for

Meaning Conveyed were slightly higher than for Grammaticality Judgments. Finally, the scores of advanced learners were higher for all types of collocations than were those of low-intermediate learners.

4.4.2. Three-scale analysis of Meaning Conveyed

In this section, we examine more closely the results of the Meaning Conveyed judgments by native speakers. I used the same criteria as for the three-scale analysis of

Grammaticality Judgments in section 4.3.2. A combination was judged to convey the intended meaning very well when the native speakers rated it as 4.5 or higher on the six- point scale; it conveyed the intended meaning to some extent when it was rated between

2.67 and 4.33; and it not convey the intended meaning when it was rated at 2.5 or lower.

This three-scale analysis helps better understand learners’ performance and native speakers’ reactions to learners’ productions. Table VI-3 in Appendix VI and Figures 4-9,

4-10 and 4-11 below display this grouping of the results from the Meaning judgments for both groups of learners for each collocation type. The first two sets of rows are graphed in Figure A1, the next two in Figure A2, the next two in Figure A3.

68.22 70% low-intermediate 62.35 advanced 60%

50% 40.11 38.44 37.94 40% 33.96

30% 25.9423.62

19.42 18.23 20% 17.11 14.67

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 A1 frequent A1 infrequent Figure 4-9. A1 Native speakers’ grading of Meaning Conveyed.

70% low-intermediate advanced 60% 55.06 49.56 47.58 50% 42.25 38.43 40%

30.79 30.43 27.65 30% 21.98 19.65 19.32 20% 17.28

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 A2 frequent A2 infrequent

Figure 4-10. A2 Native speakers’ grading of Meaning Conveyed.

70% low-intermediate advanced 60% 54.07

50% 43.82

37.92 40% 35.11 34.7834.57 30.32 30.34 27.29 30% 25.84 26.37

19.56 20%

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 A3 frequent A3 infrequent Figure 4-11. A3 Native speakers’ grading of Meaning Conveyed

The learners attained relatively high scores for all collocation types except frequent A3.

Figures 4-9, 4-10 and 4-11 generally look like staircases going down to the right. This is different from the results from three-scale analysis of Grammaticality Judgments, where the figures generally have v-shapes. In the Meaning analysis a v-shaped hollow was found only in infrequent Type A2. Figure 4-12 below combines the results for all three collocation types: A1, A2 and A3.

70% low-intermediate advanced

60% 49.86%

50% 43.66%

40% 27.73% 29.24% 27.10% 30% 22.42%

20%

10%

0% 6~4.5 4.33~2.67 2.5~1 Total A1, A2, A3

Figure 4-12. Meaning Conveyed. Total. A1, A2, A3.

There was a statistically significant difference in performance between low-intermediate

and advanced learners (χ2 = 33.195, df = 2, p < .000): Advanced learners were judged

more successful at conveying meaning (49.86%) than were low-intermediate learners

(43.66%). In contrast, low-intermediate learners were judged unsuccessful (1~2.5) more

often (29.24%) than were advanced learners (22.42%).

Next, we examine the influence of our four variables on Meaning Judgments: (1)

Frequency, (2) Existence of equivalent collocations in L1, (3) Light verb vs. Content verb,

and (4) Same vs. Different light verb. Table VI-4 in Appendix VI shows the distributions of collocation tokens and percentages by the proportions of Meaning Conveyed rated by

native speakers: high (4.5-6), mid (2.67-4.33) and low (1-2.5). Figures 4-13, 4-14, 4-15

and 4-16 graph numbers of tokens and percentages of high, mid and low scores of

Meaning Judgments.

70% low-intermediate advanced 60% 53.10%

50% 46.55% 47.04%

40.68% 40% 33.46% 28.39% 28.57% 30% 26.47% 25.06% 25.85% 24.39% 20.44% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Frequent Infrequent Figure 4-13. Meaning Conveyed. Frequent vs. infrequent collocations.

70% low-intermediate advanced 60% 53.53% 51.83% 48.25% 50%

39.98% 40% 32.53% 27.39% 27.49% 30% 26.30% 27.63% 21.87% 24.12% 19.08% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 With equivalent collocation in L1 Without equivalent collocation in L1

Figure 4-14. Meaning Conveyed. With vs. without equivalent collocations in L1.

70% low-intermediate advanced 60% 52.40%

50% 47.56% 45.27%

40% 35.71% 30.42% 34.07% 26.08% 30.22% 30% 26.86% 25.58% 24.31% 21.53% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Light verb + Noun Content verb + Noun Figure 4-15. Meaning Conveyed. Light verb + noun vs. content verb + noun.

70% low-intermediate advanced 60% 53.53% 51.83% 51.28% 50% 43.92%

40% 31.11% 27.39% 24.97% 30% 26.30% 24.79% 21.87% 23.93% 19.08% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Same light verb + Noun Different light verb + Noun Figure 4-16. Meaning Conveyed. Same light verb + noun vs. different light verb + noun.

Figures 4-13, 4-14, 4-15 and 4-16 show very similar distributions. As we saw in the graphs for each collocation type, these figures can all be characterized as staircases going down to the right. These results suggest that around half of learners’ productions successfully conveyed the intended meaning, although 20-30% of their collocations are unacceptable (i.e. low scores for Grammaticality Judgments).

4.4.3. Two-scale analysis of Meaning Conveyed

In this section, I have combined the Meaning judgments into two groups to directly compare the rates of acceptable vs. unacceptable collocations produced by the learners. To do this, I combined the Mid (2.67-4.33) and Low (1-2.5) ratings into a single proportional group; I thus compared the rates at which collocations were judged to convey the intended meaning (High 4.5-6) or not fully (Mid/Low1-4.33). Chi-square tests were again used to see whether the percentage with data in a given range varied significantly in each of these four variable groups as well as between the two proficiency groups. I thus did four analyses focusing on each of the following four variables:

85 frequency, existence of L1 equivalents, light vs. content verbs, and same vs. different light verbs.

The first analysis tested the Meaning distributions for frequent vs. infrequent collocations. Table 4-T and Figure 4-T present the findings.

Meaning Frequent collocations Infrequent collocations Conveyed (frequent A1, A2, A3) (infrequent A1, A2, A3) rated by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 600 689 1289 524 764 1288 intermediate 46.55% 53.45% 40.68% 59.31% Advanced 634 560 1194 596 671 1267 53.10% 46.91% 47.04% 52.96% Total 1234 1249 2483 1120 1435 2555 49.83% 50.18% 43.86% 56.14% TABLE 4-T. Meaning Conveyed. Frequent vs. infrequent collocations.

70%

59.31% 60% 53.10% 53.45% 52.96% 47.04% 50% 46.55% 46.91% 40.68% 40%

30%

20% low-intermediate advanced 10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Frequent Infrequent Figure 4-T. Meaning Conveyed. Frequent vs. infrequent collocations.

Both low-intermediate and advanced learners did better on frequent than infrequent collocations. A chi-square test shows that this difference is statistically significant (Low- intermediate: χ2 = 9.010, df = 1, p = *0.003; Advanced: χ2 = 9.025, df = 1, p* < .000;

Total: χ2 = 17.385, df = 1, p* < .000). Advanced learners produced significantly more highly rated collocations than did low-intermediate learners for both frequent (χ2 =

10.641, df = 1, p = *0.001) and infrequent collocations (χ2 = 10.484, df = 1, p = *0.001).

My second analysis compared Meaning scores for collocations with vs. without

L1 equivalents. Tokens and percentages of collocations with vs. without L1 equivalents,

divided into High and Mid-Low ranges of Meaning Conveyed, are displayed in Table 4-U

and Figure 4-U.

Meaning With equivalent collocation in L1 Without equivalent collocation in L1 Conveyed (A1) (A2, A3) rated by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 410 381 791 714 1072 1786 intermediate 51.83% 48.17% 39.98% 60.02% Advanced 432 375 807 798 856 1654 53.53% 46.47% 48.25% 51.75% Total 842 756 1598 1512 1928 3440 52.68% 47.32% 44.12% 55.89% TABLE 4-U. Meaning Conveyed. Collocations with vs. without L1 equivalents.

70% 60.02%

60% 53.53% 51.83% 51.75% 48.25% 48.17% 46.47% 50% 39.98% 40%

30%

20% low-intermediate advanced 10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 With equivalent collocation in L1 Without equivalent collocation in L1 Figure 4-U. Meaning Conveyed. Collocations with vs. without L1 equivalents.

Both low-intermediate and advanced learners attained significantly higher scores for collocations with Japanese equivalents than for those without equivalents (Low- intermediate: χ2 = 31.332, df = 1, p* <.000; Advanced: χ2 = 6.059, df = 1, p = *0.014;

Total: χ2 = 33.463, df = 1, p* < .000). With respect to collocations without L1

equivalents, advanced learners produced significantly more highly rated collocations than

did low-intermediate learners (χ2 = 23.836, df = 1, p* <.000). There was no significant difference between advanced and low-intermediate learners in their performance on

collocations with L1 equivalents (χ2 = 0.462, df = 1, p = 0.497).

The third analysis compares performance on content verbs vs. light verbs. Table

4-V and Figure 4-V compare low-intermediate and advanced learners’ success with in

‘Light verb + Noun’ collocations and ‘Content verb + Noun’ collocations.

Meaning Light verb + Noun (A1, A2) Content verb + Noun (A3) Conveyed rated by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 818 902 1720 306 551 857 intermediate 47.56% 52.44% 35.71% 64.29% Advanced 852 774 1626 378 457 835 52.40% 47.61% 45.27% 54.73% Total 1670 1676 3346 684 1008 1692 49.98% 50.03% 40.49% 59.51% TABLE 4-V. Meaning Conveyed. Light verb + noun vs. content verb + noun.

70% 64.29%

60% 54.73% 52.40% 52.44% 47.61% 50% 47.56% 45.27%

40% 35.71%

30%

20% low-intermediate advanced 10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Light verb + Noun Content verb + Noun Figure 4-V. Meaning Conveyed. Light verb + noun vs. content verb + noun.

Learners did significantly better with ‘Light verb + Noun’ collocations than with ‘Content

88 verb + Noun’ collocations. The difference between light verb vs. content verb collocation types was statistically significant (Low-intermediate: χ2 = 32.673, df = 1, p *< .000;

Advanced learners: χ2 = 11.216, df = 1, p = *<.000; Total: χ2 = 40.612, df = 1, p = *0).

The advanced learners’ performance was significantly better than that of low- intermediate learners both on Light verb collocations and Content verb collocations

(Light verb + Noun: χ2 = 7.833, df = 1, p = *0.005; Content verb + Noun: χ2 = 16.061, df

= 1, p = *<.000).

In my fourth analysis, I tested success with same vs. different light verbs. Table

4-W and Figure 4-W display distributions of native speakers’ ratings of Meaning

Conveyed for ‘Same’ vs. ‘Different’ light verb + Noun collocations.

Meaning Same light verb + Noun (A1) Different light verb + Noun (A2) Conveyed rated by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 410 381 791 408 420 929 intermediate 51.83% 48.17% 43.92% 51.28% Advanced 432 375 807 521 399 819 53.53% 46.47% 56.08% 48.72% Total 842 756 1598 828 920 1748 52.68% 47.32% 47.60% 52.40% TABLE 4-W. Meaning Conveyed. Same light verb + noun vs. different light verb + noun.

70% low-intermediate advanced 60% 53.53% 56.08% 51.83% 51.28% 48.17% 48.72% 50% 46.47% 43.92%

40%

30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Same light verb + Noun Different light verb + Noun

Figure 4-W. Meaning Conveyed. Same light verb + noun vs. different light verb + noun.

Low-intermediate learners produced significantly more highly-rated ‘Same light verb +

Noun’ collocations than highly-rated ‘Different light verb + Noun’ collocations (χ2 =

10.731, df = 1, p = *0.001). There was no significant difference between same vs. different light verb collocations by advanced learners (χ2 = 0.825, df = 1, p = 0.364).

However, the overall difference in the performance of all the learners on same vs.

different light verb was statistically significant (χ2 = 9.46, df = 1, p = *0.002). Advanced

learners’ performance on the type ‘Different light verb + Noun’ was significantly better

than that of low-intermediate learners (χ2 = 9.467, df = 1, p = *0.002). The difference

between Low-intermediate and advanced learner’s performance with ‘Same light verb +

Noun’ was statistically insignificant (χ2 = 0.462, df = 1, p = 0.497).

In summary, Chi-square tests show that learners produced significantly more

collocations which conveyed the intended meaning for collocations with the following

attributes: frequent collocations, collocations with L1 equivalents, light verb + Noun, and

same light verb + Noun. They did more poorly for: infrequent collocations, collocations

without L1 equivalents, content verb + Noun, and different light verb + Noun. Advanced

90 learners performed significantly better than low-intermediate learners on all collocations with more difficult characteristics (i.e. infrequent collocations, collocations without L1 equivalents, content verb + Noun collocations and different light verb + Noun collocations. Among collocations with easier characteristics, advanced learners produced significantly more highly-rated frequent collocations and ‘Light verb + Noun’ collocations than did low-intermediate learners, but no significant difference was observed for collocations with L1 equivalents, or same light verb + Noun collocations.

4.4.4. Hierarchy of difficulty based on Meaning Conveyed

In the previous sections, I looked at the relationship between native speakers’ ratings of Meaning Conveyed and the four attributes that influenced learners’ performance, using two- and three-scale analyses. In this section, I attempt to infer a hierarchy of difficulty based on the percentages of collocations judged to adequately convey the intended meaning (4.5~6).

A series of Chi-square tests reveal the following difficulty hierarchy from the easiest collocation type (A1 frequent) on the left through the most difficult collocation types (A3 frequent) on the right.

A2 frequent A2 infrequent A1 frequent (different light verb) (different light verb) A3 frequent (same light verb) 52.14% 42.71% (content verb)

30.86% 65.25% A1 infrequent A3 infrequent (MC=4.5-6) (content verb) (same light verb) 48.89% 39.25%

Easier <------> More difficult TABLE 4-3. Difficulty hierarchy of collocations.

The percentage of learners’ productions that were rated highly was significantly greater

for the frequent A1 type (65.25%) than for the frequent A2 type (52.14%) (χ2 = 29.893; df

= 1; p* < .000). There was no significant difference between the frequent A2 type and

the infrequent A3 type (χ2 = 1.868; df = 1; p = 0.172) but there were significantly more

highly rated productions of the infrequent A3 type (48.89%) than the infrequent A2 type

(42.71%) (χ2 = 6.845; df = 1; p = 0.009). The difference between the infrequent A2 type and the infrequent A1 type was not significant (χ2 = 2.043; df = 1; p = 0.153), but significantly more highly rated collocations were observed for the infrequent A1 type than for the frequent A3 type (χ2 = 12.119; df = 1; p* < .000).

This hierarchy shows that, similar to the hierarchy based on Grammaticality

Judgments, there is a tendency for light verb collocations to be easier for learners than

content verb collocations, except for the comparatively high percentage of highly rated

collocations which was observed for the infrequent A3 type. Among light verb

collocations, frequent collocations are easier than infrequent ones. However, among

content verb collocations, the learners conveyed intended meanings more successfully in

the infrequent set.

4.5. Learners’ confidence in their own production

The subjects rated each item they produced using two 10-point scales: (1)

Confidence (How correct they think each answer is?); and (2) Familiarity (Whether they

have seen or read the combination they produced). In this section we examine the first

rating: learners’ confidence in their productions. Table 4-X and Figure 4-X show the

average confidence scores for learners’ own productions of each collocation type. On

this scale 10 is the highest (very confident) and 1 is the lowest (least confident).

Type of collocations Low-intermediate Advanced learners Total learners (# = 35) (# = 31) (# = 66) A1 Frequent 7.02 8.25 7.64 same light V Infrequent 5.40 6.56 5.98 A2 Frequent 6.62 7.67 7.15 different light V Infrequent 6.05 6.95 6.50 A3 Frequent 6.29 7.92 7.11 content V Infrequent 6.44 7.56 7.00 Total (# = 66) 6.30 7.49 6.90 TABLE 4-X. Average confidence scores for each collocation type: A1, A2, A3 (10-point- scale).

9 low-intermediate 8.25 advanced 7.92 8 7.67 total 7.64 7.56 6.95 7.15 7.02 7.11 7 7 6.62 6.56 6.5 6.44 6.29 5.98 6.05 6 5.4

L e a rL e r n s ' f n c o I c e e (n d i 1 o 0 p t n s c a l e ) 1 frequent Infrequent frequent infrequent frequent infrequent A 1 A 2 A 3

Figure 4-X. Average confidence scores for each collocation type (10-point-scale)

Advanced learners were on average more confident in their performance (7.49) than low- intermediate learners (6.30). The averages for all frequent collocations were higher (A1:

7.64; A2: 7.15; and A3: 7.11) than for infrequent collocations (A1: 5.98; A2: 6.50; and

A3: 7.00).

I again tested these distributions for the four variables: (1) Frequency, (2)

Existence of L1 equivalents, (3) Light verb vs. Content verb, and (4) Same vs. Different light verbs. Frequency: Table 4-Y and Figure 4-Y show learners’ confidence for frequent vs. infrequent collocations.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Frequent (frequent A1, A2, A3) 6.64 7.95 7.30 Infrequent (infrequent A1, A2, A3) 5.96 7.02 6.49 TABLE 4-Y. Average confidence scores for frequent vs. infrequent collocations. (10- point-scale)

7.95 8 7.3 7.02 7 6.64 6.49 5.96 6

3 low-intermediate advanced total 2

L e a r e r e 'L n s f n c o i i c e ( o e n p d t 1 0 n s c a l e ) 1 Frequent collocations Infrequent collocations Figure 4-Y. Average scores of learners’ confidence: frequent vs. infrequent collocations. (10-point-scale)

The average for frequent collocations was higher (6.76) than for infrequent ones (5.81).

Scores for advanced learners were higher than for low-intermediate learners for both frequencies.

L1 equivalents: Table 4-Z and Figure 4-Z show the average confidence scores for all types of collocations which do vs. do not have L1 equivalents.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) With L1 equivalent (A1) 6.21 7.41 6.81 Without L1 equivalent (A2, A3) 6.35 7.53 6.94 TABLE 4-Z. Average confidence scores: with vs. without L1 equivalents. (10-point-scale)

8 7.41 7.53 6.94 7 6.81 6.21 6.35 6

3 low-intermediate advanced total 2

L e aL r e rn s ' f c n o i c d e n e ( 1 i0 p o t n s c a l e ) 1 With equivalent collocation in L1 Without equivalent collocation in L1

Figure 4-Z. Average confidence scores: with vs. without L1 equivalents. (10-point-scale)

The overall confidence for collocations without L1 equivalents was slightly higher (6.94)

than for those with equivalents (6.81). Advanced learners were more confident in their

performance than low-intermediate learners for types with and without L1 equivalents.

Light verbs vs. content verbs: Table 4-AA and Figure 4-AA show the average

scores of learners’ confidence for ‘Light verb + Noun’ and ‘Content verb + Noun’ collocations.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Light verb + Noun (A1, A2) 6.27 7.36 6.82 Content verb + Noun (A3) 6.37 7.74 7.06 TABLE 4-AA. Average confidence scores: light verb + noun vs. content verb + noun. (10- point-scale)

8 7.74 7.36 7.06 7 6.82 6.27 6.37

3 low-intermediate advanced total 2

L eL a r n e r s ' c o n f i d e n c e ( 1 0 i p o tn s c a l e ) 1 Light verb + Noun Content verb + Noun

Figure 4-AA. Average confidence scores: light verb + Noun vs. Content verb + Noun. (10-point-scale)

Both low-intermediate and advanced learners were somewhat more confident for their

content verb + noun collocations (7.06) than for their light verb + noun collocations

(6.82). Scores for advanced learners were higher for both types.

Same vs. different light verb + noun collocations: Table 4-BB and Figure 4-BB present average confidence scores for these two groupings.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Same light verb + Noun (A1) 6.21 7.41 6.81 Different light verb + Noun (A2) 6.34 7.31 6.83 TABLE 4-BB. Average confidence scores: same vs. different light verb. (10-point-scale)

8 7.41 7.31

7 6.81 6.83 6.21 6.34 6

3 low-intermediate advanced total 2

L e a r n e r s e r e a r s 'f n L ic e (n i c o o 0 e n t 1 p n d s c a l e ) 1 Same light verb + Noun Different light verb + Noun

Figure 4-BB. Average confidence scores: same vs. different light verb. (10-point-scale)

Low-intermediate learners gave slightly higher ratings for ‘Different verb’ collocations

(6.34) than for ‘Same verb’ collocations (6.21), whereas advanced learners gave slightly higher ratings for ‘Same verb’ (7.41) than for ‘Different verb’ collocations (7.31). The total score was slightly higher for ‘Different verb’ collocations (6.83) than for ‘Same verb’ collocations (6.81).

In summary, learners were more confident for the following collocation types: (1)

Frequent collocations, (2) Collocations without L1 equivalents, (3) ‘Content verb +

Noun’ collocations, and (4) ‘Different light verb + Noun’ collocations. A striking finding is that the learners were often more confident of their productions of the more difficult types of collocations even though native speakers judged them less highly. I next present two-variable analyses to compare learners’ own confidence ratings with native speakers’ judgments of Grammaticality and Meaning Conveyed.

4.6. Learners’ familiarity with their own productions (Heard or Read).

Table 4-CC and Figure 4-CC present the averages of learners’ responses to the

question ‘Have you ever heard or read this combination?’ using a 10-point scale.

Type of collocations Low-intermediate Advanced learners Total (# = 66) learners (# = 35) (# = 31) A1 Frequent 6.59 7.87 7.23 Same light V Infrequent 4.43 5.91 5.17 A2 Frequent 6.10 7.25 6.68 different light V Infrequent 5.35 6.28 5.82 A3 Frequent 5.69 7.01 6.35 content V Infrequent 5.93 7.06 6.42 Total (# = 66) 5.68 6.90 6.28 TABLE 4-CC. Average familiarity scores for combinations A1, A2 and A3. (10-point scale)

9 low-intermediate advanced 8 7.87 total 7.23 7.25 7.01 7.06 7 6.59 6.68 6.28 6.35 6.42 6.1 5.91 5.82 5.93 6 5.69 5.35 5.17 5 4.43

H e arH d o r r ea ( d 1 0 p o in t s c al e ) 2

1 frequent Infrequent frequent infrequent frequent infrequent A 1 A 2 A 3

Figure 4-CC. Average familiarity scores for combinations: A1, A2 and A3. (10-point- scale)

The average of the familiarity ratings by advanced learners was higher (6.90) than for low-intermediate learners (5.68) for all types of collocations taken together. The averages for frequent collocations of types A1 and A2 were higher (A1: 7.23 and A2:

6.68) than for infrequent collocations (A1: 5.17 and A2: 5.82). The average for infrequent combinations of Type A3 was slightly higher (6.42) than for frequent

98 combinations (6.35). The same trend has already been noted for native speakers’ judgments of Grammaticality (4.3) and Meaning Conveyed (4.4).

In the same manner, I tested the influence of each of our four variables: (1)

Frequency, (2) Existence of L1 equivalents, (3) Light verb vs. Content verb, and (4) Same vs. Different light verbs. Frequency: Table 4-DD and Figure 4-DD show learners’ judgments of their familiarity with frequent vs. infrequent collocations.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Frequent (freq A1, A2, A3) 6.13 7.38 6.76 Infrequent (infreq A1, A2, A3) 5.20 6.42 5.81 TABLE 4-DD. Average familiarity for frequent vs. infrequent collocations. (10-point- scale)

8 7.38

7 6.76 6.42 6.13 6 5.81 5.2 5

3 low-intermediate advanced total 2 H e a r o r ea ( d d H 1 0 p o i n ts c a l e )

1 Frequent Infrequent Figure 4-DD. Average familiarity scores for frequent vs. infrequent collocations. (10- point-scale)

The average for frequent collocations was higher (6.76) than that for infrequent ones

(5.81). Scores for advanced learners were higher than for low-intermediate learners for both frequencies.

Collocations with vs. without L1 equivalents: Table 4-EE and Figure 4-EE show the average scores for collocations which do vs. do not have Japanese equivalents.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) With L1 equivalents (A1) 5.51 6.89 6.20 Without L1 equivalents (A2, A3) 5.77 6.90 6.34 TABLE 4-EE. Average familiarity scores for collocations with vs. without L1 equivalents. (10-point-scale)

6.89 6.9 7 6.2 6.34 6 5.77 5.51

3 low-intermediate advanced total 2 H e a r o r r d ea (H d 1 0 p o i n ts c a l e )

1 With equivalent collocation in L1 Without equivalent collocation in L1 Figure 4-EE. Average scores of learners’ familiarity with collocations with vs. without L1 equivalents. (10-point-scale)

The average for collocations without L1 equivalents was only slightly higher (6.34) than for those with L1 equivalents (6.20). Advanced learners produced higher ratings than did low-intermediate learners for collocations both with and without L1 equivalents.

Collocations containing light vs. content verbs: Table 4-FF and Figure 4-FF show the average familiarity scores for ‘Light verb’ vs. ‘Content verb’ collocations.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Light verb + Noun (A1, A2) 5.62 6.83 6.23 Content verb + Noun (A3) 5.81 7.04 6.43 TABLE 4-FF. Average familiarity scores for combinations using light vs. content verbs (10-point-scale)

100

7.04 7 6.83 6.43 6.23 5.81 6 5.62

3 low-intermediate advanced total 2 H e a r d o r e a d (i e a r d 1 0 p n t s c a l H e )

1 Light verb + Noun Content verb + Noun

Figure 4-FF. Average familiarity scores for combinations using light vs. content verbs (10-point-scale)

The average for ‘Content verb + Noun’ collocations was only slightly higher (6.43) than for ‘Light verb + Noun’ collocations (6.23). Advanced learners produced higher ratings than did low-intermediate learners for both light verb and content verb collocations.

Familiarity with collocations containing same vs. different verbs: Table 4-GG and

Figure 4-GG show the average scores for collocations grouped this way.

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) Same light verb + Noun (A1) 5.51 6.89 6.20 Different light verb + Noun (A2) 5.73 6.77 6.25 TABLE 4-GG. Average familiarity scores for same vs. different light verb collocations. (10-point-scale)

101

6.89 7 6.77 6.2 6.25 6 5.73 5.51

3 low-intermediate advanced total 2 H e a r d o r r e a d i ( r e a d t 1 0 p o r o n e a r d la s c e ) H

1 Same light verb + Noun Different light verb + Noun

Figure 4-GG. Average familiarity scores for same vs. different light verb collocations. (10-point-scale)

Low-intermediate learners produced higher ratings for ‘Different light verb + Noun’ collocations (5.73) than for ‘Same light verb’ collocations (5.51), while advanced learners claimed to be slightly more familiar with ‘Same light verb’ collocations (6.89) than

‘Different light verb’ collocations (6.77). The overall score was only slightly higher for

‘Different light verb + Noun’ collocations (6.25) than that for ‘Same light verb + Noun’ collocations (6.20).

In summary, the learners indicated more familiarity with the following collocation types: (1) Frequent collocations, (2) Collocations without Japanese equivalents, (3)

‘Content verb + Noun’ collocations, and (4) ‘Different light verb + Noun’ collocations.

Except for type (1) frequent collocations, these results are in reverse proportion to both

Grammaticality Judgments and Meaning Conveyed as rated by native speakers. It is striking that the learners judge themselves as more familiar with the more difficult types of collocations. I will discuss the relationship between learners’ familiarity and native speakers’ Grammaticality & Meaning judgments in section 4.7.

102

4.7. Multi-variable analyses of results

In sections 4.3-4.6, we separately examined two 6-point ratings by native speakers

(Grammaticality Judgments and Meaning Conveyed) and two 10-point ratings by learners

of their productions (Confidence and Familiarity). We also considered how our four

variables influenced learners’ performance. In this section, we consider pairs of ratings.

First, we look at Grammaticality Judgments paired with learners’ confidence (4.7.1), then

Meaning Conveyed with learners’ confidence (4.7.2). After that, we compare learners’ confidence vs. familiarity with their own productions (4.7.3). By considering multiple variables together, we can capture learners’ performance more deeply: this analysis affords same insights about psychological aspects, such as how many of the subjects were confident in their answers when their answers were actually wrong, or what proportion of the learners’ productions were acceptable though the subjects lacked confidence in their productions.

4.7.1. Grammaticality Judgments vs. learners’ confidence

Table VI-5 in Appendix VI combines the Grammaticality ratings by native speakers (6-point scale) with the subjects’ ratings of their confidence (10-point scale). I split the scores for each of these variables into two groups as follows: Grammaticality

Judgments were divided into: (1) Grammatical (High score: 4.5-6) vs. (2) Ungrammatical

(Mid/Low scores: 1-4.33). This is the same division used in the two-scale analysis in

4.3.3. Learners’ confidence ratings were split as follows: (1) Confident (scores: 7-10); vs.

(2) Not confident (scores: 1-6).

Figures 4-17 and 4-18 summarize the pooled data. Figure 4-17 compares

103

Grammaticality when learners were vs. were not confident in their production, and Figure

4-18 compares learners’ confidence when their productions were grammatical vs.

ungrammatical. First, we examine Figure 4-17.

Figure 4-17. Grammaticality when learners were vs. were not confident in their productions.

A significant difference was observed between the low-intermediate and the advanced groups in the proportion of grammatical combinations for which the subjects were confident in their productions (χ2 = 3.945; df = 1; p* = 0.0470). Among all the

productions that the subjects rated confidently, a larger percentage of advanced learners’

combinations were regarded as grammatical (49.99%) than of low-intermediate learners’

(46.60%). No such difference was observed between the proficiency groups when the

subjects lacked confidence in their productions (χ2 = 1.224; df = 1; p = 0.2686).

104

Figure 4-18. Learners’ confidence when their productions were judged grammatical vs. ungrammatical.

Statistical analysis shows that the differences in the proportion of confident productions between grammatical and ungrammatical productions were significant for both proficiency groups (Low: χ2 = 127.666; df = 1; p* = 0; Advanced: χ2 = 163.247; df =

1; p* = 0; Total: χ2 = 301.939; df = 1; p* = 0); both groups gave high confidence scores

when their answers were grammatical, and low confidence scores when their answers

were ungrammatical.

There were also statistically significant differences between the proficiency groups in the proportions of productions about which they were confident when their combinations were judged grammatical or not. When their productions were

grammatical, more advanced learners rated their productions confidently by (82.10%)

than did low-intermediate learners (64.27%) (χ2 = 69.864; df = 1; p* = 0). When the

subjects’ answers were ungrammatical, more advanced learners again were confident

(57.37%) than were low-intermediate learners (42.10%) (χ2 = 73.603; df = 1; p* = 0).

These results reveal that advanced learners are more confident in their

105

productions in general, even when their combinations are ungrammatical. Advanced

learners’ overall confidence in their productions might be one factor impeding the

attainment of more native-like proficiency. It is possible that advanced learners are so

confident that they don’t try to continue improving their collocational competence by

regular study at home, or that they don’t even notice when their productions are not fully

acceptable to native speakers.

4.7.2. Meaning Conveyed vs. learners’ confidence

In this section, we combine native speakers’ ratings of Meaning Conveyed (6-

point scale) with learners’ confidence in their productions (10-point scales). I used the

same procedure as in the previous section (4.7.1.) to divide these rating scores into just

two groups. Meaning Conveyed was split as follows: (1) Meaning successfully conveyed

(High score: 4.5-6); and (2) Meaning is NOT conveyed (Mid/Low scores: 1-4.33). This

is the same criterion used in the two-scale analysis in 4.4.3. As in section 4.7.1, learners’

confidence ratings were divided into the following two groups: (1) Confident (scores: 7-

10); and (2) Not confident (scores: 1-6).

Table VI-6 (see Appendix VI) summarizes Meaning Conveyed as rated by native

speakers with the subjects’ confidence ratings. Figures 4-19 and 4-20 graph the pooled

data. Figure 4-19 compares ratings of Meaning Conveyed when learners were confident

vs. not confident in their production, and Figure 4-20 compares learners’ confidence when their productions convey vs. do not convey the intended meaning.

106

Figure 4-19. Ratings of Meaning Conveyed when learners were confident vs. not confident in their productions.

A difference was found between the two groups of learners. Among confident productions, significantly more of those produced by advanced learners (59.73%) conveyed the intended meaning than those from low-intermediate learners (55.04 %) (χ2

= 6.637; df = 1; p* = 0.0010). When the subjects lacked confidence in their productions, there was no significant difference between the two proficiency groups in the proportion of productions which were judged to convey the intended meaning or not (χ2 = 2.798; df

= 1; p = 0.0944).

107

Figure 4-20. Learners’ confidence when their productions did vs. did not convey intended meaning.

As with Meaning Conveyed vs. learners’ confidence, across both proficiency groups there was a statistically significant difference in the amount of confident productions when Meaning was vs. was not judged to be conveyed (Low: χ2 = 142.422;

df = 1; p* = 0; Advanced: χ2 = 219.335; df = 1; p* = 0; Total: χ2 = 368.373; df = 1; p* =

0). All subjects gave higher confidence scores for productions that were judged to

convey the intended meaning.

When their productions were judged to convey the intended meaning,

significantly more advanced learners rated their combinations confidently by (80.90%)

than did low-intermediate learners (64.57%) (χ2 = 79.697; df = 1; p* = 0). The difference

was also significant when their combinations did not convey the intended meaning.

Advanced learners were more confident in their productions even when their

combinations did not convey intended meaning (52.96%) as compared with low-

intermediate learners (40.88%) (χ2 = 39.692; df = 1; p* = 0).

108

These results are similar to those we found between Grammaticality Judgments and learners’ confidence in section 4.7.1. All the subjects are more confident about their productions when their combinations are judged to convey intended meaning. However, advanced learners are more confident overall regardless of accuracy, even when their combinations do not convey the meaning.

4.7.3. Learners’ confidence vs. familiarity

In sections 4.5. and 4.6., we separately examined learners’ confidence and familiarity with their productions. In this section we briefly consider learners’ confidence and familiarity ratings together. Overall the subjects rated their confidence and familiarity using the same 10-point-scales. No significant difference was observed in the way they rated their confidence and familiarity. That is, when they were confident about a production, they felt they had heard or read that combination produced by native speakers; when they were not confident, they were more likely to feel they had never heard or read it. However, overall the subjects gave slightly higher scores for confidence than for familiarity. Table VI-7 (see Appendix VI) summarizes the average differences between confidence and familiarity ratings for each group for each collocation subtype.

Figures 4-21, 4-22, 4-23 and 4-24 show the average differences for each subtype and

Figure 4-25 divides the data into easy vs. difficult subtypes, and Figure 4-26 collapses easy vs. difficult subtypes.

109

Figure 4-21. Learners’ confidence and familiarity (10-point scale): Frequent vs. infrequent.

Figure 4-22. Learners’ confidence and familiarity (10-point scale): With vs. without L1 equivalents.

110

Figure 4-23. Learners’ confidence and familiarity (10-point scale): Light vs. content verbs.

Figure 4-24. Learners’ confidence and familiarity (10-point scale): Same vs. different light verbs.

111

Figure 4-25. Learners’ confidence and familiarity (10-point scale): Easy vs. difficult types.

Figure 4-26. Learners’ confidence and familiarity (10-point scale): Total

The differences between the confidence and familiarity ratings were slightly greater for

low-intermediate learners than for advanced learners. No clear systematic difference was

observed between the easy and difficult subtypes. Low-intermediate learners gave slightly lower familiarity scores for the easier types except for the frequent subtypes,

112

while advanced learners gave slightly lower familiarity scores for the more difficult types.

The differences are very small and not significant.

4.8. Discussion of Experiment 1

In Experiment 1, I have reported on the analysis of four factors (frequency,

existence of L1 equivalence, light vs. content verbs, and same vs. different light verbs)

which influence learners’ productions and perception. The results show that these four

variables can be divided into easy vs. difficult types as below:

Variables Easy types More difficult types (1) Frequency Frequent Infrequent (2) Existence of L1 equivalents With L1 equivalents Without L1 equivalents (3) Light verbs vs. content verbs Light verbs Content verbs (4) Same vs. different light verbs Same light verbs Different light verbs TABLE 4-4. Easy vs. difficult types of collocations.

All the results from native speaker ratings of Grammaticality and Meaning

Conveyed show that subjects attained higher scores for the easy types than for difficult types. The results from the three-scale analyses of Grammaticality Judgments show

similar patterns. All the figures (Figures 4-5, 4-6, 4-7, and 4-8) resemble a v-shaped

hollow between 2 hills. This suggests that native speakers’ judgments are quite clear-cut

as to acceptability although there are some individual variations. In contrast, the results

from the Meaning Conveyed judgments (Figures 4-13, 4-14, 4-15, and 4-16) resemble

staircases going down to the right. Around half of learners’ productions successfully conveyed the intended meaning with 20-30% judged to be grammatically unacceptable.

113

I now discuss these results in relation to the points discussed in 4.5 and 4.6: (1) number of collocations produced; (2) the four factors which influence productions; and

(3) learners’ confidence and familiarity.

4.8.1. Number of collocations produced In general, advanced learners produced more collocations (2.21) for all the subtypes (both easy and difficult) than did low-intermediate learners (2.06) (see 4.2.). It is possible that low-intermediate learners produced fewer because of their overall lack of confidence in giving answers (see 4.5) and used an avoidance strategy. Another reason could be the difference in their vocabulary size.

Both proficiency groups produced more collocations for more difficult types than for the easy types. A possible reason is that uncertainty about their answers for the difficult types led them to make more attempts.

Bardovi-Harlig’s study (2008) on the developmental stages of L2 acquisition of formulas shows that learners produced a greater variety of formulae at the early stages of acquisition; as their proficiency increases, their productions exhibit less variation.

Similar to Bardovi-Harlig, the results of my study show that the subjects produced a larger variety of collocations for all the difficult types except for the frequent vs. infrequent types. Figure 4-27 shows the average numbers of different tokens produced for each item grouped by of subtype.

114

Figure 4-27. The average numbers of different tokens produced per item of each subtype.

These results suggest that difficulty with collocations could lead learners to make a wider range of attempts.

4.8.2. Observed frequency

Supporting Hatch’s observation (1974) that frequency of input may influence accuracy in second language acquisition, my subjects generally did better with frequent collocations, attaining significantly higher scores for frequent collocations than infrequent ones (χ2 = 93.922, df = 1, p* = 0). This result also supports Kellerman and

Jorden’s notion of psychological markedness (1977, 1978, 1979, 2000), i.e., that learners work from the hypothesis that there are constraints on how similar the L2 can be to the

L1, and that forms which are perceived as marked (less frequent, semantically odd, syntactically less producible, etc.) tend to be more difficult than those which they perceive to be less marked. In other words, frequent collocations are acquired more

115

easily because they are perceptually less complex. It appears that learners’ intuitions

about the range of prototypicality affect their performance with collocations.

However, there were some differences in their accuracy with specific collocations,

especially among frequent types. For example, even though the following collocations

are all frequent, 61 of 68 subjects produced make a decision, 60 produced ask questions,

63 produced take note, while only 32 produced do business, 16 produced take account, and 5 produced put a stop.

Such differences might be the result of insufficient input in formal English education in Japan. Koya (2005) counted numbers of V-NP collocations in four English textbooks used in Japanese high schools. She found that on average each textbook contained only 29 collocations, although these collocations were all highly frequent in the

BNC and the TIMES corpora. She states that these textbooks select appropriate collocations, but that it is necessary to present many more in order to increase learners’ collocational competence. Koya’s findings suggest that although the subjects did very well on the small number of frequent collocations which they learned in formal education in Japan, they had not acquired many collocations both frequent and infrequent which they had not encountered in their formal instruction.

4.8.3. With vs. without L1 equivalents

My subjects did better on collocations with L1 equivalents than on those without.

It appears that, although English and Japanese are typologically different languages,

Japanese learners of English perceive English collocations that have Japanese equivalents as easier than those without. The existence of L1 equivalents facilitates learning. Thus,

116

the greater accuracy for collocations with L1 equivalents may reflect successful transfer

from L1 (Bahns, 1993).

This finding supports other empirical studies which investigated English learners

with different L1 backgrounds, such as Caroli (1998) who collected data from 73 Italian learners of English and found that her subjects acquired English collocations with Italian equivalents more easily than those without. On the other hand, as suggested by many studies (Bahns & Eldaw, 1990, Koya, 2002, Murano, 2004, Nesselhauf, 2003), the lack of

L1 equivalents often causes difficulty for learners and makes them susceptible to potential influence from their L1. The learners’ L1 thus plays a role in creating both

positive and negative transfer of forms from L1 to L2 (Bahns and Eldaw, 1993, Fayez-

Hussein, 1990, Marton, 1978).

4.8.4. Light verbs vs. content verbs

Greater accuracy was observed for the light verb types than for the content verb types. The restricted meaning of content verbs might cause difficulty for learning.

This result somewhat conforms to Caroli (1998), who could find no substantial difference between light and content verbs. She felt that learners were familiar with light verb collocations.

However, partly because of the design of this study, we are not yet justified in concluding that light verbs are always easier than content verbs. First, among the 66 collocations tested in Experiments 1 and 2 together, the target for 44 of them was a light verb. This uneven distribution of light and content verbs might have prompted learners to produce more light verbs than they otherwise would have.

117

A second reason is that since the subjects were asked to produce as many combinations as they could think of, they might have produced more light verbs in an attempt to make their collocations sound native-like. Misuse of light verbs (Low- intermediate: 36.81% and Advanced: 37%) is a very serious problem (see section 2.3).

High accuracy in collocations involving light verbs might be partly because the subjects produced so many light verbs, whether or not they were actually familiar with their use in a specific collocation. The semantic vagueness of light verbs may cause difficulty in learning collocations.

4.8.5. Same vs. different light verbs

Same light verbs are generally easier than different light verbs. The existence of

L1 equivalents appears to facilitate acquisition. However, as discussed in section 4.8.4., misuse of light verbs was very frequent. This result indicates that most learners do not have sufficient knowledge of the use of light verbs, and that the acquisition of light verbs is often a very big challenge for L2 learners.

Each of these four factors contributes to learners’ difficulty with collocations.

The difficulty hierarchy (see Table 4-2) shows the easiest type (A1 frequent, same light verb) on the left end with the more difficult types (A2 infrequent, different light verb; and

A3 frequent & infrequent, content verbs) toward the right. The more difficult factors a collocation carries, the more difficult it generally is for L2 learners to acquire.

4.8.6. Learners’ confidence and familiarity

Advanced learners’ indicated greater confidence in their productions (7.49 on the

118

10-point scale) than did low-intermediate learners (6.30) (See Table 4-X). The scores for

frequent types were higher (7.30) than for infrequent types (6.49) (see Figure 4-Y), but a

clear difference was not observed for the other contrasts (with vs. without L1 equivalents

types, light vs. content verbs, and same vs. different light verbs).

Advanced learners’ familiarity ratings (Heard or Read) were also higher (6.90) than for low-intermediate learners’ (5.68) (See Table 4-CC). As with the confidence results, subjects indicated greater familiarity (Heard and Read) for the frequent types

(6.76) than for infrequent types (5.81). No clear difference was observed for the other contrasts (with vs. without L1 equivalents types, light vs. content verbs, and same vs. different light verbs).

A notable finding in relation to the subjects’ confidence and native speakers’ ratings of Grammaticality & Meaning Conveyed is that advanced learners are generally more confident in their productions. Clement and his co-workers’ research, using a questionnaire with 293 francophone students at the University of Ottawa (Clement 1986) demonstrated that self-confidence is the best predictor of language proficiency. They suggest that overall self-confidence significantly contributes to learners’ willingness to communicate in a foreign language. It is possible that learners’ confidence in their productions may increase their overall self-confidence or vice versa. Thus confidence may positively affect L2 proficiency.

However, one problem is that advanced learners may be overconfident in their combinations even when those combinations are unacceptable. This overconfidence might be one factor which makes it difficult to attain more native-like proficiency.

Although I did not collect information about the subjects’ regular study of English

119 at home, my impression is that the majority of the subjects, except for those who were enrolled in the English language programs, spend no or very little time studying English at home. After the pretest, some subjects gave such comments as “I could have answered these questions much better when I was studying English for entrance examinations in

Japan” and “I’ve forgot a lot, but this test helped me refresh my old memory”. It is possible that most Japanese students in the US are so busy with school work that they find little time to study English at home. At the same time, many learners, especially, advanced learners might be so confident in their productions that they don’t try to continue improving their collocational competence by regular study at home, or they don’t even notice that some of their productions are not acceptable to native speakers.

120

CHAPTER 5

Experiment 2: The role of Meaning in acquiring Verb+Noun collocations

5.1. Stimuli

In the previous experiment all collocations had relatively literal meanings, now we consider semantic transparency as a variable.

Thirty Verb+Noun collocations were selected on the basis of their semantic transparency. In order to calibrate this variable, four native speakers of English rated a list of collocations on the following 10-point scale: 10 = literal meaning, 1 = non-literal meaning. For the experiment, a collocation was considered to be literal when all the native speaker judges responded with 6 or higher; the meaning of a collocation was regarded as abstract or metaphorical when all the native judges rated it with 5 or lower.

The frequency of all the collocations were checked with the BNC. They occurred between 22 and 82 times, that is, they are neither extremely frequent nor infrequent.

The thirty Verb + Noun collocations were divided into the following six types using the same categories A1, A2 and A3 as in Experiment 1: (1) Phrases of Type A1 with literal meanings, (2) Phrases of Type A1 with metaphorical meanings, (3) Phrases of Type

A2 with literal meanings, (4) Phrases of Type A2 with metaphorical meanings, (5)

Phrases of Type A3, with literal meanings (6) Phrases of Type A3 with metaphorical meanings. Five tokens were prepared for each type. All the test sentences are attached in

Appendix VII. The data were collected at the same time as the data for Experiment 1.

121

Types VERB TYPE MEANING EXAMPLES same light verb in (1) Literal Do a project A1 English and Japanese (2) Abstract Have an ear (for) different light verb (3) Literal Take an exam (Receive an exam in Japanese) A2 in English and Japanese (4) Abstract Take a dive (Do a dive in Japanese) content verb in (5) Literal Develop cancer (Become cancer in Japanese) English, and light A3 (6) Abstract Bear witness (do witness in Japanese) verb in Japanese TABLE 5-1. The six collocational types for Experiment 2.

5.2. Overall results and single variable analyses of results

A total of 3883 responses for Experiment 2 were collected from the sixty-six subjects at the same time as responses were collected for Experiment 1. The data collected in Experiment 3 is attached in Appendix VIII. Table 5-A and Figure 5-A display the average numbers of combinations produced by the subjects for each type.

Low-intermediate Advanced learners Total (# = 66) Types of collocations learners (# = 35) (# = 31) Token avg/prsn token avg/prsn token avg/prsn A1 Literal 387 2.21 390 2.52 777 2.35 same light V Abstract 313 1.79 326 2.10 639 1.94 A2 Literal 316 1.81 316 2.04 632 1.92 different light V Abstract 271 1.55 248 1.60 519 1.57 A3 Literal 382 2.18 344 2.22 726 2.20 content V Abstract 316 1.81 274 1.77 590 1.79 Total (# = 66) 1985 1.89 1898 2.04 3883 1.96 TABLE 5-A. Number of responses produced by the subjects for Experiment 2.

122

2.52 low-intermediate 2.5 2.35 advanced 2.22 2.21 total 2.18 2.2 2.1 2.04 1. 9 4 2 1. 9 2 1. 7 9 1. 8 1 1. 8 1 1. 7 7 1. 79 1. 55 1. 6 1. 5 7 1. 5

0.5

0 literal abstract literal abstract literal abstract

A 1 A 2 A 3

Figure 5-A. Average numbers of literal and abstract combinations produced.

Low-intermediate learners produced an average of 1.89 combinations for each

question while advanced learners averaged 2.04 combinations. A chi-square test shows that the difference is significant (χ2 = 5.6, df = 1, p = 0.018). Overall, more combinations

were produced for collocations with literal or transparent meanings as compared with

collocations with non-literal or abstract meanings. (Type A1 literal (2.35) vs. abstract

(1.94): χ2 = 13.26, df = 1, p = 0.0003; Type A2 literal (1.92) vs. abstract (1.57): χ2 = 10.9,

df = 1, p = 0.001; Type A3 literal (726) vs. abstract (590): χ2 = 13.84, df = 1, p = 0.0002).

Not much difference was observed between low-intermediate and advanced

learners. The only significant difference was in the abstract Type A1 collocations where

advanced learners produced more (2.10) than did low-intermediate learners (1.79) (χ2 =

4.04, df = 1, p = 0.0444).

Table 5-B and Figure 5-B summarize the numbers of combinations produced for literal vs. abstract collocations:

123

Type of Low-intermediate Advanced Total (# = 66) Collocation learner (# = 35) learners (# = 31) Tokens average/ token average/ token average/ person person person Literal collocations (literal A1, A2, A3) 1085 2.07 1050 2.26 2135 2.16 Abstract collocations (abst A1, A2, A3) 900 1.71 848 1.82 1748 1.77 TABLE 5-B. Number of responses produced by the subjects for literal vs. abstract collocations.

2.3 2.26 low-intermediate 2.2 2.16 advanced total

2.1 2.07

1.9 1.82 1.8 1.77 1.71 1.7

1.6 A v e A r a g be e rn u m o f b ic o m n a t i o n s

1.5 Literal collocations Abstract collocations

Figure 5-B. Average number of responses produced by each subject for literal vs. abstract collocations.

More combinations were produced for literal collocations (2.16) than for abstract

collocations (1.77). This difference is statistically significant (χ2 = 38.38, df = 1, p

< .0001). Advanced learners produced significantly more combinations for literal

collocations (2.26) than low-intermediate learners (2.07) (χ2 = 4.1, df = 1, p = 0.0429).

There was no statistical difference between low-intermediate and advanced learners in abstract collocations.

124

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token avg/psn Token avg/psn Token avg/psn With L1 equivalents (A1) 700 1.84 716 2.07 1416 1.96 Without L1 equivalents (A2, A3) 1285 2.00 1182 1.91 2467 1.96 TABLE 5-C. Number of responses produced by the subjects for collocations with vs. without L1 equivalents. A1, A2, A3.

2.3

2.2

2.1 2.07

2 2 1.96 1.96 1.91 1.9 1.84

1.8

1.7 low-intermediate advanced total 1.6 A v er b e A ra o fi e b g n u m c o m n a ti o n s

1.5 With equivalent collocation in Without equivalent collocation in Japanese Japanese

Figure 5-C. Number of responses produced by the subjects for collocations with vs. without L1 equivalents. A1, A2, A3.

Collocations with L1 equivalents are Type A1 and collocations without L1

equivalents are Types A2 and A3. Low-intermediate learners produced significantly

more combinations for collocations without L1 equivalents than for collocations with L1 equivalents (χ2 = 4.51, df = 1, p = *0.0337), while advanced learners produced significantly more combinations for collocations with L1 equivalents (χ2 = 18.94, df = 1, p <*.0001). The total average for collocations both with and without L1 equivalents was

identical: 1.96. Advanced learners produced significantly more collocations with L1

equivalents than did low-intermediate learners (χ2 = 7.08, df = 1, p = 0.0078)

Table 5-D and Figure 5-D display the numbers of combinations produced for

‘Light verb + Noun’ and ‘Content verb + Noun’ collocations.

125

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token Avg/psn Token avg/psn Token avg/psn Light verb + Noun (A1, A2) 1287 1.84 1280 2.07 2567 1.96 Content verb + Noun (A3) 698 2.00 618 2.00 1316 2.00 TABLE 5-D. Number of responses produced by the subjects for collocations with light verbs vs. content verbs.

2.3

2.2

2.1 2.07

2 2 2 2 1.96

1.9 1.84

1.8

1.7 low-intermediate advanced total 1.6 A v e b e r rA a e i o f m g b n u c a o m n t i o n s

1.5 Light verb + Noun Content verb + Noun

Figure 5-D. Number of responses produced by the subjects for collocations with light verbs vs. content verbs.

Light verb collocations are Types A1 and A2, and content verb collocations are Type A3.

Statistical difference. Low-intermediate learners produced significantly more content verb collocations than light verb collocations (χ2 = 4.1, df = 1, p = 0.0429). The average numbers of light verb + Noun collocations produced by advanced learners were higher than for those produced low-intermediate learners (chi-square = 8.34, df = 1, p = 0.0039).

No statistical difference was observed in advanced learners’ performance on light vs. content verb collocations, or in learners’ performance on content verb collocations.

The last variable is same vs. different light verbs. Table 5-E and Figure 5-E present the numbers of responses produced by the subjects for collocations with same vs.

126

different light verbs.

Low-intermediate Advanced learners Total (# = 66) Types of learners (# = 35) (# = 31) Collocation token avg/psn Token avg/psn Token avg/psn Same light verb + Noun (A1) 700 2.00 716 2.31 1416 2.15 Different light verb + Noun (A2) 587 1.68 564 1.82 1151 1.74 TABLE 5-E. Number of responses produced by the subjects for collocations with same vs. different light verbs.

2.31 2.3 s

2.2 2.15

2.1 low-intermediate advanced 2 2 total

1.9 1.82 1.8 1.74

1.7 1.68

1.6 A vee rb A r io f b u m ae n g m co a n t i n o

1.5 Same light verb + Noun Different light verb + Noun

Figure 5-E. Number of responses produced by the subjects for collocations with same vs. different light verbs.

Collocations of Type A1 are grouped as ‘Same light verb + Noun’, while collocations of

Type A2 are grouped as ‘Different light verb + Noun’. Overall, the learners produced significantly more collocations of type A1 ‘Same light verb + Noun’ as compared to type

A2 ‘Different light verb + Noun’ (Low-intermediate: χ2 = 9.74. df = 1, p = *0.0018;

Advanced: chi-square = 17.82, df = 1, p * < .0001; Total: chi-square = 27.16, df = 1, p

*< .0001). The average for same light verb collocations for advanced learners was significantly higher than for low-intermediate learners (χ2 = 7.08, df = 1, p = *0.0078).

No significant difference was found for numbers of ‘Different light verb’ collocations

127 between low-intermediate and advanced learners (χ2 = 1.77, df = 1, p = 0.1834).

In summary, advanced learners generally tended to produce more combinations than low-intermediate learners. The difference was significant for all the easy collocation types: literal collocations, collocations with L1 equivalents, light verb collocations, and same light verb + Noun collocations. No significant difference was observed between low-intermediate and advanced learners for any of more difficult types of collocations.

Learners produced more combinations for literal collocations and same light verb collocations than for abstract collocations and different light verb collocations. They produced approximatelly the same numbers of combinations for collocations with vs. without L1 equivalents and Light verb vs. Content verb collocation.

In the next section, we consider the native speakers’ Grammaticality Judgments for each collocation type, followed by analyses using three- and then two-scales to better understand learners’ performance.

5.3. Native speakers’ Judgments of Grammaticality

5.3.1. Average scores of native speakers’ Grammaticality Judgments

Table 5-F and Figure 5-F show the average scores of native speakers’

Grammaticality Judgments for each type of collocation produced by the subjects using a

6point-scale.

128

Type of collocations Low-intermediate Advanced learners Total (# = 66) learners (# = 35) (# = 31) A1 Literal 4.59 4.47 4.53 same light V Abstract 3.70 3.66 3.68 A2 Literal 3.91 4.05 3.98 different light V Abstract 2.81 3.21 3.01 A3 Literal 3.16 3.30 3.23 content V Abstract 2.25 2.53 2.39 Total (# = 66) 3.40 3.54 3.47 TABLE 5-F. Native speakers’ Grammaticality Judgment in Experiment 2. (6-point-scale)

5 4.59 low-intermediate 4.474.53 4.5 advanced total 4.05 3.98 4 3.91 3.7 3.66 3.68

3.5 3.3 3.21 3.16 3.23 3.01 3 2.81 2.53 2.5 2.39 2.25

1.5

1 literal abstract literal abstarct literal abstract G r a m m a t i m l c a t r e n ( i a m G 6 - m i t g o p d t y J u n -l s c a e ) A 1 A 2 A 3

Figure 5-F. Native speakers’ Grammaticality Judgment (6-point-scale). Experiment 2.

The averages for literal collocations of all types (A1, A2 and A3) were higher than for abstract collocations (A1, A2 and A3). The average judgments for Types A2 and A3 by advanced learners were higher than for low-intermediate learners, whereas low- intermediate learners were slightly more successful on Type A1 than were advanced learners.

Next, I compared the influence of my four variables on Grammaticality

Judgments: (1) Collocations with literal vs. abstract meaning, (2) Equivalent collocations in L1, (3) Light verb vs. Content verb, and (4) Same vs. Different light verb. First, Table

129

5-G and Figure 5-G display the average Grammaticality Judgments for literal vs. abstract collocations.

Type of Low-intermediate Advanced Total Collocation learner (# = 35) learners (# = 31) (# = 66) Literal collocations (literal A1, A2, A3) 3.89 3.94 3.91 Abstract collocations (abstr A1, A2, A3) 2.92 3.13 3.03 TABLE 5-G. Native speakers’ Grammaticality Judgments: Literal vs. abstract collocations. (6-point-scale)

5 low-intermediate

t 4.5 advanced total 3.94 4 3.89 3.91

3.5 3.13 3.03 3 2.92

2.5 (I - t 6 o n - p a l c s ) e 2

G r a m m a t I c a l i t y J u d g m e n e n t a Ia l c m m i g d t u J r y m a G 1.5

1 Literal meaning Abstract meaning

Figure 5-G. Native speakers’ Grammaticality Judgments: Literal vs. abstract collocations. (6-point-scale)

The average for literal collocations was much higher (3.91) than for abstract collocations

(3.03). Scores for advanced learners were only slightly higher than for low-intermediate learners for both literal and abstract collocations.

Next, Table 5-H and Figure 5-H show the average scores of Grammaticality

Judgments for all types of collocations with and without Japanese equivalents.

130

Types of Low-intermediate Advanced learners Total (# = 66) Collocation learners (# = 35) (# = 31) With L1 equivalents (A1) 4.15 4.07 4.11 Without L1 equivalents (A2, A3) 3.03 3.27 3.15 TABLE 5-H. Native speakers’ Grammatical Judgments: collocations with vs. without L1 equivalents. (6-point-scale)

5 low-intermediate

t 4.5 advanced 4.15 4.07 4.11 total 4

3.5 3.27 3.15 3.03 3

2.5 ( 6 - p o I n t - s c a l e ) l e a c t - s I n o p - ( 6 2

G r a m m a t I e n m c a l i r a m tm g G d y J u 1.5

1 With L1 equivalent Without L1 equivalent

Figure 5-H. Native speakers’ Grammatical Judgments: collocations with vs. without L1 equivalents. (6-point-scale)

Both low-intermediate and advanced learners did better on collocations with L1 equivalents (4.11) than without L1 equivalents (3.15). Low-intermediate learners attained slightly higher scores (4.15) for collocations with L1 equivalents than did advanced learners (4.07). In contrast, advanced learners were more successful on collocations without L1 equivalents (3.27) than were low-intermediate learners (3.03).

Table 5-I and Figure 5-I show the average scores of Grammaticality Judgments for ‘Light verb + Noun’ and ‘Content verb + Noun’ collocations.

131

Types of Low-intermediate Advanced learners Total (# = 66) Collocation learners (# = 35) (# = 31) Light verb + Noun (A1, A2) 3.75 3.85 3.80 Content verb + Noun (A3) 2.71 2.92 2.82 TABLE 5-I. Native speakers’ Grammatical Judgments for collocations with light verbs vs. content verbs. (6-point-scale)

5 low-intermediate 4.5 advanced total

4 3.85 3.75 3.8

3.5

2.92 3 2.82 2.71

2.5

( 6 - p o I n t - s c a l e ) l e a c t - s I n o - p ( 6 2

1.5 G r a m m a t r m IG a m c a l i t e n t y J u d g m

1 Light verb + Noun Content verb + Noun

Figure 5-I. Native speakers’ Grammatical Judgments for collocations with light verbs vs. content verbs. (6-point-scale)

The learners attained much higher scores for ‘Light verb + Noun’ collocations (low-

intermediate: 3.75; advanced: 3.85) than for ‘Content verb + Noun’ collocations (low- intermediate: 2.71; and advanced: 2.92). The scores for advanced learners were higher than for low-intermediate learners for both sets of collocations.

Finally, Table 5-J and Figure 5-J present the average scores of Grammaticality

Judgments for collocations with the same vs. different light verbs.

Types of Low-intermediate Advanced learners Total (# = 66) Collocation learners (# = 35) (# = 31) Same light verb + Noun (A1) 4.1 4.07 4.09 Different light verb + Noun (A2) 3.36 3.63 3.50 TABLE 5-J. Native speakers’ Grammatical Judgment for collocations with same vs. different light verbs. (6-point-scale)

132

5 low-intermediate

t 4.5 advanced total 4.1 4.07 4.09 4 3.63 3.5 3.5 3.36

2.5

( 6 - p o I n t - s c a l e ) 2

G r a m m a t rm G Ia m c a l i t y e n J u d g m 1.5

1 Same light verb + Noun Different light verb + Noun

Figure 5-J. Native speakers’ Grammatical Judgment for collocations with same vs. different light verbs. (6-point-scale)

The averages of judgments for same verb collocations (4.09) were higher than for different verb collocations (3.5). The average for same verb collocations of the performance by low-intermediate learners was slightly higher (4.10) than for advanced learners (4.07), while advanced learners did better on ‘Different verb’ collocations (3.63 vs. 3.36).

In summary, the learners attained higher scores for the following ‘easier’ collocation types: (1) Literal collocations, (2) Collocations with Japanese equivalents, (3)

Light verb + noun collocations, and (4) Same light verb + noun collocations. The scores of advanced learners were always higher than for low-intermediate learners for all the more difficult types: (1) Abstract collocations, (2) Collocations without Japanese equivalents, (3) Content verb + Noun collocations, and (4) Different light verb + Noun collocations. With respect to easier types of collocations, the low-intermediate learners attained very slightly higher scores for collocations with L1 equivalents and same light verb collocations, while advanced learners did very slightly better on literal collocations

133

and light verb + noun collocations. We will examine which of these differences are

statistically significant in section 5.3.3.

5.3.2. Three-scale analysis of Grammaticality Judgments

In this section, we examine the results of the Grammaticality Judgments more

closely. I used the same criteria as in Experiment I; a combination was judged acceptable

if native speakers rated it as 4.5 or higher on the six-point scale; it was regarded as

neither acceptable or unacceptable when it was rated between 2.67 and 4.33; and it was

judged unacceptable when it was rated 2.5 or lower.

Just as we saw in Experiment I, this three-scale analysis helps better understand learners’ performance because the average scores of native speakers’ Grammaticality

Judgments do not correspond to the most typical performance by Japanese learners. The average scores generally fall in the ‘neither acceptable nor unacceptable’ range when they are grouped in this way. Most commonly, however, many combinations were rated either

‘acceptable’ or ‘not acceptable’ with relatively few combinations in the middle. This can be seen in Table IX-1 (Appendix IX) and Figures 5-1, 5-2 and 5-3 which display the results from the grammaticality judgments of the low-intermediate learners and advanced learners for each collocation type.

134

70% low-intermediate

advanced 60% 55.56 52.31

50% 46.33 42.45 44.41 42.45

40% 35.13

30.49 30%

20% 13.9512.56 15.09 9.27 10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 B-Type 1 literal B-Type 1 less literal

Figure 5-1. Native speakers’ Grammaticality Judgment. A 1.

70% low-intermediate

advanced 60% 51.27 49.37 48.89 50% 43.95

39.56

40% 34.18 31.48 28.63 27.42 30%

19.63

20% 14.56 11.08 10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 B-Type 2 literal B-Type 2 less literal

Figure 5-2. Native speakers’ Grammaticality Judgment. A 2. 68.99 70% low-intermediate 64.68

advanced 60%

50% 46.67

41.03 37.39 40% 32.5

30%

20.83 21.58 18.67 18.59 20% 16.73 12.34

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 B-Type 3 literal B-Type 3 less literal

Figure 5-3. Native speakers’ Grammaticality Judgment. A 3.

135

A v-shaped hollow was observed in three of the figures: the abstract type of A1

and the literal types of A2 and A3. The Figure for literal A1 looks like a steep staircase

going down to the right, while the abstract type of A2 looks like a gentle slope going up

to the right (unacceptable). The abstract type of A3 looks like a steep wall towering on

the right. Here we see a greater variety of distributions than in Experiment I. One reason

for this could be that many of the variables, such as the meanings (literal vs. abstract) as

well as the types of verbs (light vs. content verb) exert a very strong influence on the

learners’ performance.

Figure 5-4 combines the results for all three collocation types into a single graph

figure: A1, A2 and A3.

70% low-intermediate advanced 60%

50% 39.41% 42.61% 36.95% 37.97% 40%

22.62% 30% 20.44%

20%

10%

0% 6~4.5 4.33~2.67 2.5~1 Total A1, A2, A3

Figure 5-4. Native speakers’ Grammaticality Judgments. Total A1, A2, A3.

There was a statistically significant difference in performance between low-intermediate

and advanced learners (χ2 = 8.751, df = 2, p = 0.0126): Advanced learners were generally

more successful than were low-intermediate learners. But overall, the performance of

136 both sets of learners had a similar v-shaped hollow, in spite of differences in the distributions observed for each collocation type.

Next, we examine the influence of our four variables on native speakers’

Grammaticality Judgments: (1) Literal vs. Abstract meaning, (2) Existence of equivalent collocations in L1, (3) Light verb vs. Content verb, and (4) Same vs. Different light verb.

Table IX-2 (in Appendix IX) shows the distributions of tokens of collocations and then percentages sorted by the proportions of Gramatticalty Judgments by native speakers in three ratings: high (4.5-6), mid (2.67-4.33) and low (1-2.5). Figures 5-5, 5-6, 5-7 and 5-8 graph numbers of tokens and percentages of high, mid and low scores of Grammaticality

Judgments.

70% low-intermediate advanced 60% 54.40%

47.25% 50.06% 50% 45.91%

40% 32.64% 29.70% 30% 28.21% 24.54% 26.36% 21.45% 20.24% 19.24% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Literal Abstract

Figure 5-5. Grammaticality Judgments. Literal vs. abstract collocations.

137

70% low-intermediate advanced 60% 51.43% 50.95% 50% 47.88% 45.27%

40% 34.25%

27.57% 28.92% 30% 26.13% 25.99% 20.48% 21.00% 20.13% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 With equivalent collocation in Japanese Without equivalent collocation in Japanese

Figure 5-6. Grammaticality Judgments. Collocations with vs. without L1 equivalents.

70% low-intermediate advanced 60% 57.10%

51.67% 50% 44.73% 44.25%

40% 34.99% 31.53% 30% 28.09% 23.74% 23.08% 20.23% 20.76% 19.82% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Light verb + Noun Content verb + Noun Figure 5-7. Grammaticality Judgments. Light vs. content verbs.

70% low-intermediate advanced 60% 51.43%

50% 47.88% 43.86% 40.78% 38.48% 40% 35.67%

27.57% 30% 26.13% 25.99% 20.74% 21.00% 20.48% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Same light verb Different light verb Figure 5-8. Grammaticality Judgments. Same vs. different light verb + noun.

138

A first glance at these figures suggests several different types of distributions. Indeed the

figures for the literal and abstract types, collocations without L1 equivalents, Light verb +

Noun, and Different light verb collocations all look like hollows between two hills. In contrast, the figures for collocations with L1 equivalents and for Same light verb collocations have a wall rising on the left, while the figure for Content verb collocations looks like a wall rising on the right.

One feature which should be noted here is that in each of these figures the tallest bar always appears on the left in figures for easy type collocations ((1) Literal collocations, (2) Collocations with L1 equivalents, (3) Light verb + Noun, and (4) Same light verb + Noun). In contrast, the tallest bar always appears on the right for figures for difficult type collocations ((1) Abstract collocations, (2) Collocations without L1

equivalents, (3) Content verb + Noun, and (4) Different light verb + Noun). All the left

bars (High 4.5~6) of the easy types are as high as 44.25~51.43% while the right bars

(Low 1~2.5) of these types are between 25.99~34.99%. In contrast, the right bars (Low

1~2.5) for the difficult types reach 38.48~57.10% while the left bars (High 4.5~6) of

these types only reach 23.08~40.78%.

Such distributions were not observed for the native speakers’ Grammaticality

Judgments in Experiment I. It is possible that the meaning of collocations may influence

learners’ performance more strongly than frequency.

5.3.3. Two-scale analysis of Grammaticality Judgments

In this section, I have combined the Grammaticality Judgments into just two

139

groups to directly compare the rates of acceptable vs. unacceptable collocations produced by the learners. I combined the Mid (2.67-4.33) and Low (1-2.5) ratings into a single proportional group; I was thus able to compare the rates at which collocations were judged to acceptable (High 4.5-6) or not fully so (Mid/Low1-4.33). Chi-square tests were

used to see whether the percentages varied significantly for each of these four variable

groups as well as between the two proficiency groups. I thus did four analyses focusing

on each of the four variables: (1) Literal vs. abstract meaning, (2) Existence of L1

equivalents, (3) Light vs. content verbs, and (4) Same vs. different light verbs.

The first analysis compared the Grammaticality distributions for literal vs.

abstract collocations. Table 5-K and Figure 5-K present the findings.

Literal collocations (A1, A2, A3 Abstract collocations (A1, A2, A3 Grammaticality literal) abstract) Judgments by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 488 575 1063 237 662 899 intermediate 45.91% 54.09% 26.36% 73.64% Advanced 489 546 1035 248 587 835 47.25% 52.75% 29.70% 70.30% Total 977 1121 2098 485 1249 1734 46.58% 53.42% 28.03% 71.97% TABLE 5-K. Grammaticality Judgments: Collocations with literal vs. abstract meanings.

140

80% 73.64% low-intermediate 70.30% 70% advanced

60% 54.09% 52.75% 47.25% 50% 45.91%

40% 29.70% 30% 26.36%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Literal Abstract

Figure 5-K. Grammaticality Judgments: Collocations with literal vs. abstract meanings.

Both sets of learners did better on literal than on abstract collocations. A chi-square test shows that this difference is statistically significant (Low-intermediate: χ2 = 79.866, df =

1, p = *0; Advanced: χ2 = 59.582, df = 1, p = *0; Total: χ2 = 139.163, df = 1, p = *0).

There was no statistical difference between low-intermediate and advanced learners’

performance (Literal collocations: χ2 = 0.378, df = 1, p = 0.539; Abstract collocations: χ2

= 2.394, df = 1, p = 0.123).

My second analysis compared Grammaticality Judgment scores for collocations with vs. without L1 equivalents. Tokens and percentages of collocations with vs. without

L1 equivalents, divided into High and Mid-Low ranges of Grammatical Acceptability are displayed in Table 5-L and Figure 5-L.

141

Grammaticality With L1 equivalents (B1) Without L1 equivalents (B2, B2) Judgments by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 360 340 700 365 897 1262 intermediate 51.43% 48.57% 28.92% 71.08% Advanced 339 369 708 398 764 1162 47.88% 52.12% 34.25% 65.75% Total 699 709 1408 763 1661 2424 49.66% 50.35% 31.59% 68.42% TABLE 5-L. Grammaticality Judgments: Collocations with vs. without L1 equivalents.

80% low-intermediate advanced 71.08% 70% 65.75%

60% 52.12% 51.43% 48.57% 47.88% 50%

40% 34.25% 28.92% 30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 With L1 equivalent Without L1 equivalent

Figure 5-L. Grammaticality Judgments: Collocations with vs. without L1 equivalents.

Both sets of learners attained significantly higher scores for collocations with Japanese equivalents than for those without equivalents (Low-intermediate: χ2 = 97.894, df = 1, p =

*0; Advanced: χ2 = 34.228, df = 1, p = *0; Total: χ2 = 124.589, df = 1, p = *0). With

respect to collocations without L1 equivalents, advanced learners produced significantly

more highly rated collocations than did low-intermediate learners (χ2 = 7.965, df = 1, p*

= 0.005). There was no significant difference between advanced and low-intermediate

learners in their performance on collocations with L1 equivalents (χ2 = 1.772, df = 1, p =

0.183).

The third analysis compares performance on content verbs vs. light verbs. Table

142

5-M and Figure 5-M compare low-intermediate and advanced learners’ success with light

verb + noun collocations and content verb + noun collocations.

Grammaticality Light verb + Noun (B1, B2) Content verb + Noun (B3) Judgments by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low-intermediate 569 717 1286 156 520 676 44.25% 55.75% 23.08% 76.92% Advanced 569 703 1272 168 430 598 44.73% 55.27% 28.09% 71.90% Total 1138 1420 2558 324 950 1274 44.49% 55.51% 25.59% 74.41% TABLE 5-M. Grammaticality Judgments. Light verb + noun vs. content verb + noun.

80% 76.92% low-intermediate 71.90% 70% advanced

55.27% 60% 55.27%

44.73% 50% 44.73%

40%

28.09% 30% 23.08%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Light verb + Noun Content verb + Noun

Figure 5-M. Grammaticality Judgments. Light verb + noun vs. content verb + noun.

Overall, learners did significantly better with ‘Light verb + Noun’ collocations than with

‘Content verb + Noun’ collocations. The difference between light verb vs. content verb collocation types was statistically significant (Low-intermediate: χ2 = 85.226, df = 1, p =

*0; Advanced learners: χ2 = 47.162, df = 1, p = *0; Total: χ2 = 130.879. df = 1, p = *0).

The advanced learners’ performance was significantly better than that of low-

intermediate learners on Content verb collocations (χ2 = 4.211, df = 1, p = *0.040). No

statistical difference between low-intermediate and advanced learners was found for

143

Light verb collocations (χ2 = 0.061, df = 1, p = 0.805).

In my fourth analysis, I tested success with same vs. different light verbs. Table

5-N and Figure 5-N display distributions of native speakers’ ratings of Grammaticality

for ‘Same’ vs. ‘Different’ light verb + Noun collocations.

Grammaticality Same light verb (A1) Different light verb (A2) Judgments by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low-intermediate 360 340 700 209 377 586 51.43% 48.57% 35.67% 64.34% Advanced 339 369 708 230 334 564 47.88% 52.99% 40.78% 59.22% Total 699 709 1408 439 711 1150 49.66% 50.78% 38.23% 61.78% TABLE 5-N. Grammaticality Judgments: Same vs. different light verb + noun.

80% low-intermediate 70% advanced 64.34% 59.22% 60% 52.99% 51.43% 48.57% 47.88% 50% 40.78% 40% 35.67%

30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Same light verb + Noun Different light verb + Noun

Figure 5-N. Grammaticality Judgments: Same vs. different light verb + noun.

Both low-intermediate and advanced learners produced significantly more highly-rated same light verb + noun collocations than highly-rated different light verb + noun collocations (Low-intermediate: χ2 = 32.128, df = 1, p <*.000; Advanced: χ2 = 6.403, df =

1, p = *0.011; Total: χ2 = 33.726, df = 1, p <*.000). There was no significant difference

144

between low-intermediate and advanced learners’ performance both for ‘Same light verb’

(χ2 = 1.772, df = 1, p = 0.183) or for ‘Different light verb’ collocations (Chi-square =

3.185, df = 1, p = 0.074).

In summary, Chi-square tests show that learners produced significantly more

acceptable collocations with the following ‘easy’ attributes: literal collocations,

collocations with L1 equivalents, light verb + Noun, and same light verb + Noun. They did more poorly with: abstract collocations, collocations without L1 equivalents, content verb + Noun, and different light verb + Noun. Little statistical difference was observed between the performance of low-intermediate and advanced learners. The differences were significant only for two attributes: collocations without L1 equivalents, and

‘Content verb + Noun’ collocations.

5.3.4. Hierarchy of difficulty based on Grammaticality Judgments

In the previous sections, we used two- and three-scale analyses to capture learners’ performance on Verb + Noun collocations. I looked at the relationship between native speakers’ Grammaticality Judgments and the four attributes that affected learners’ performance. In this section, I will again extract a hierarchy of difficulty based on the percentages of collocations judged acceptable (4.5~6) by native speakers

(Grammaticality Judgments).

A series of Chi-square tests reveal the following difficulty hierarchy from the easiest (A1 literal on the left) through the most difficult collocation types (A3 abstract on the right).

145

A1 literal (same light verb) 53.92% A1 abstract A3 literal A2 abstract A3 abstract (GJ=4.5-6) (same light (content verb) (different light (content verb)

A2 literal verb) 34.83% verb) 14.36% (different light 44.37% 23.36% verb) 50.32% Easier <------> More difficult TABLE 5-2. Difficulty hierarchy of collocations.

There was no significant difference between the literal A1 type (53.92%) and the literal

A2 type (50.32%) (χ2 = 1.82; df = 1; p = 0.177). Statistically significant differences were

observed among the other pairs (A2 literal vs. A1 abstract: χ2 = 4.473; df = 1; p* = 0.034;

A1 abstract vs. A3 literal: χ2 = 12.558; df = 1; p* < .000; A3 literal vs. A3 abstract: χ2 =

18.57; df = 1; p* < .000; A2 abstract vs. A3 abstract: χ2 = 14.707; df = 1; p* < .000).

This hierarchy shows that, in general, literal collocations are easier for learners than abstract ones. For both light verb collocations (A1, A2) and content verb collocations (A3), literal collocations are easier than abstract ones.

5.4. Adequacy of meaning conveyed

5.4.1. Average scores of Meaning Conveyed

Table 5-O and Figure 5-O show the average scores of native speakers’ ratings when asked ‘How well is the intended meaning conveyed when a collocation is replaced with a phrase produced by a learner?’, for each type of collocation using a 6-point-scale.

146

Low-intermediate Advanced Total (# = 66) learners (# = 35) learners (# = 31) A1 Literal 4.80 4.80 4.80 same light V Abstract 4.11 4.11 4.11 A2 Literal 4.52 4.57 4.55 different light V Abstract 3.26 3.68 3.47 A3 Literal 3.65 3.79 3.72 content V Abstract 2.67 2.93 2.80 Total (# = 66) 3.84 3.98 3.91 TABLE 5-O. Native speakers’ judgments of Meaning Conveyed in Experiment 2. (6- point-scale)

5 4.8 4.8 low-intermediate 4.8 4.57 4.52 4.55 advanced 4.5 4.11 total 4.11 4.11 3.79 4 3.68 3.65 3.72 3.47 3.5 3.26 2.93 3 2.8 2.67

2.5

1.5

1 literal abstract literal abstract literal abstract

Figure 5-O. Native speakers’ judgments of Meaning Conveyed in Experiment 2. (6- point-scale)

The average judgments of performance by low-intermediate learners was slightly lower

(3.84) than for advanced learners (3.98). As with the results of Experiment I, these

meaning scores are higher than the average scores of native speakers’ Grammaticality

Judgments (low-intermediate: 3.40; advanced: 3.54). Advanced learners attained higher

scores than did low-intermediate learners for both literal and abstract types of A2 and A3.

Low-intermediate and advanced learners attained the same average scores for both literal

147

and abstract types of A1. The averages for all literal collocations were higher (A1: 4.80;

A2: 4.55; and A3: 3.72) than for abstract collocations (A1: 4.11; A2: 3.47; and A3: 2.80)

Table 5-P and Figure 5-P compare the meaning ratings for literal vs. abstract collocations.

Type of Low-intermediate Advanced learners Total (# = 66) Collocation learner (# = 35) (# = 31) Literal collocations (literal A1, A2, A3) 4.32 4.39 4.36 Abstract collocations (abst A1, A2, A3) 3.35 3.57 3.46 TABLE 5-P. Native speakers’ judgments of Meaning Conveyed in Experiment 2. Literal vs. abstract collocations. (6-point-scale)

5 ) low-intermediate advanced 4.5 4.32 4.39 4.36 total

4 3.57 3.46 3.5 3.35

2.5

1.5 M e ai n M n v o eC yg n e( d 6 - p o i t n - s c a l e 1 Literal meaning Abstract meaning

Figure 5-P. Native speakers’ judgments of Meaning Conveyed in Experiment 2. Literal vs. abstract collocations. (6-point-scale)

The average for literal collocations was higher (4.36) than that for abstract collocations

(3.46). Scores for advanced learners were a little higher than for low-intermediate learners for both literal and abstract collocations.

Next, Table 5-Q and Figure 5-Q show the average Meaning Conveyed scores for all types of collocations which do vs. do not have Japanese equivalents.

148

Types of Low-intermediate Advanced learners Total Collocation learners (# = 35) (# = 31) (# = 66) With L1 equivalents (B1) 4.46 4.46 4.46 Without L1 equivalents (B2, B3) 3.53 3.74 3.64 TABLE 5-Q. Native speakers’ grading of Meaning Conveyed: Collocations with vs. without translation equivalent in Japanese and English. (6-point-scale)

5 ) low-intermediate 4.46 4.46 4.46 advanced 4.5 total

4 3.74 3.64 3.53 3.5

2.5

1.5 M e a n i n g C o n v e y e d (d y e v e 6 -i n o o t p n i C -e a n g n l s c a M e 1 With translation Without translation

Figure 5-Q. Native speakers’ grading of Meaning Conveyed: Collocations with vs. without translation equivalent in Japanese and English. (6-point-scale)

Both low-intermediate and advanced learners were more successful in conveying intended meaning for collocations with L1 equivalents (4.46) than for collocations without L1 equivalents (3.64). Low-intermediate and advanced learners attained the same average Meaning scores for collocations with L1 equivalents. For collocations without L1 equivalents, the average score for advanced learners was somewhat higher

(3.57) than for low-intermediate learners (3.35).

Next, Table 5-R and Figure 5-R show the average Meaning Conveyed scores for light verb + noun and content verb + noun collocations.

149

Types of Low-intermediate Advanced learners Total (# = 66) Collocation learners (# = 35) (# = 31) Light verb + Noun (B1, B2) 4.17 4.29 4.23 Content verb + Noun (B3) 3.16 3.36 3.26 TABLE 5-R. Native speakers’ judgments of Meaning Conveyed: Collocations with light verbs vs. content verbs. (6-point-scale)

5 ) low-intermediate advanced 4.5 4.29 4.17 4.23 total

3.5 3.36 3.26 3.16

2.5

1.5 M e a n i n g C o n v e y e d ( e d e y v - 6 i n o o t p n i C n -e a g n a l s c M e 1 Light verb + Noun Content verb + Noun

Figure 5-R. Native speakers’ judgments of Meaning Conveyed: Collocations with light verbs vs. content verbs. (6-point-scale)

Again, both low-intermediate and advanced learners attained higher Meaning scores for

‘Light verb + Noun’ collocations (low-intermediate: 4.17; advanced: 4.29) than for

‘Content verb + Noun’ collocations (low-intermediate: 3.16; advanced: 3.36). Scores for advanced learners were higher than for low-intermediate learners for both sets of collocations.

Finally, Table 5-S and Figure 5-S present average Meaning Conveyed scores for collocations with the same vs. different light verbs.

Types of Low-intermediate Advanced Total (# = 66) Collocation learners (# = 35) learners (# = 31) Same light verb + Noun (B1) 4.46 4.46 4.46 Different light verb + Noun (B2) 3.89 4.13 4.01 TABLE 5-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs. (6-point-scale)

150

5 ) low-intermediate advanced 4.46 4.46 4.46 4.5 total 4.13 4.01 4 3.89

3.5

2.5

1.5 M e a n i n g C o n v e y e d (d y e v e 6 -i n o o t p n i C -e a n g n ls c a M e 1 Same light verb + Noun Different light verb + Noun

Figure 5-S. Native speakers’ judgments of Meaning Conveyed: Collocations with same vs. different light verbs. (6-point-scale)

The learners attained higher scores when the light verb was the same across languages

(low-intermediate: 4.46; advanced: 4.46) than when it differed (low-intermediate: 3.89; advanced: 4.13). Low-intermediate and advanced learners attained the same average score for ‘Same light verb’ collocations (4.46), while advanced learners were more successful on ‘Different light verb’ collocations (4.13) than were low-intermediate learners (3.89).

In summary the learners attained higher scores for the following collocation types: (1) Literal collocations, (2) Collocations with Japanese equivalents, (3) ‘Light verb

+ Noun’ collocations, and (4) ‘Same light verb + Noun’ collocations. These results are the same as for native speakers’ Grammaticality Judgments. In general, scores for

Meaning Conveyed were slightly higher than for Grammaticality Judgments. Finally, the scores of advanced learners were higher for all difficult types of collocations than were those of low-intermediate learners. For easier types of collocations, low-intermediate and advanced learners attained very similar scores.

151

5.4.2. Three-scale analysis of Meaning Conveyed

In this section, we examine more closely the results of the Meaning Conveyed judgments by native speakers just as we did in Experiment I, section 4.4.2. Again, I used the same criteria as for the three-scale analysis of Grammaticality Judgments in section

4.3.2. A combination was judged to convey the intended meaning very well when the native speakers rated it as 4.5 or higher on the six-point scale; it conveyed the intended meaning to some extent when it was rated between 2.67 and 4.33; and it not convey the intended meaning when it was rated at 2.5 or lower.

This three-scale analysis helps better understand learners’ performance and native speakers’ reactions to learners’ productions. Table IX-3 (in Appendix IX) and Figures 5-9,

5-10 and 5-11 display this grouping of the results from the Meaning judgments for both groups of learners for each collocation type. The first two sets of rows in Table IX-3 are graphed in Figure A1, the next two in Figure A2, the next two in Figure A3.

152

70% 65.45 63.25 low-intermediate advanced 60%

47.87 50% 44.48

40%

30.37 27.82 28.2 30% 25.15 23.93 22.34

20% 12.21 8.92 10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 B-Type 1 literal B-Type 1 less literal Figure 5-9. Native speakers’ judgments of Meaning Conveyed. A1.

70% low-intermediate

advanced 60% 54.05 50.96

50%

40.96 38.38 37.9 40% 33.06 29.81 29.03 30% 26.21

19.23 19.74 20.66 20%

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 B-Type 2 literal B-Type 2 less literal Figure 5-10. Native speakers’ judgments of Meaning Conveyed. A2.

70% low-intermediate

advanced 60% 56.01

48.91 47.38 50% 45.29

40% 33.21 30.81 31.01 30% 27.49 27.23 21.8 17.88 20% 12.97

10%

0% 6~4.5 4.33~2.67 2.5~1 6~4.5 4.33~2.67 2.5~1 B-Type 3 literal B-Type 3 less literal Figure 5-11. Native speakers’ judgments of Meaning Conveyed. A3.

153

All the figures except for abstract types A2 and A3 look like staircases going down to the

right. This result indicates that many learners attained high scores (4.5~6) for these types

of collocations. In contrast, the figure for abstract type A2 for low-intermediate learners

looks like moderate slope going up to the right, and the figure for advanced learners

shows a small hillock in the middle. The figures for abstract type A3 have steep

staircases going up to the right. These results suggest that the abstract types of A2 and

A3 are more difficult for learners. Figure 5-12 combines the results for all these

collocation types: A1, A2 and A3.

70% low-intermediate advanced 60%

44.58% 50% 41.96%

40% 30.78% 29.68% 28.36% 30% 24.64%

20%

10%

0% 6~4.5 4.33~2.67 2.5~1 Total A1, A2, A3

Figure 5-12. Meaning Conveyed. Total A1, A2, A3.

There was a statistically significant difference in performance between low-intermediate and advanced learners (χ2 = 12.103, df = 2, p = *0.002): Advanced learners were judged

more successful at conveying meaning (44.58%) than were low-intermediate learners

(41.96%). In contrast, low-intermediate learners were judged unsuccessful (1~2.5) more

often (29.68%) than were advanced learners (24.64%).

154

Next, we examine the influence of our four variables on Meaning Judgments: (1)

Literal vs. Abstract collocations, (2) Existence of equivalent collocations in L1, (3) Light verb vs. Content verb, and (4) Same vs. Different light verb. Table IX-4 (Appendix IX) shows the distributions of collocation tokens and percentages by the proportions of

Meaning Conveyed rated by native speakers: high (4.5-6), mid (2.67-4.33) and low (1-

2.5). Figures 5-13, 5-14, 5-15 and 5-16 graph numbers of tokens and percentages of high, mid and low scores of Meaning Judgments.

70% low-intermediate advanced 60% 55.41% 54.12%

50%

41.93% 40% 33.49% 32.55% 33.96% 28.34% 30.83% 30% 26.32% 27.24%

19.56% 20% 16.24%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Literal Abstract Figure 5-13. Meaning Conveyed. Literal vs. abstract collocations.

70% low-intermediate advanced 57.68% 60% 54.80%

50%

39.23% 40% 35.28% 33.52% 31.66% 29.11% 31.23% 30% 29.11% 23.04% 19.28% 20% 16.10%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 With L1 equivalent Without L1 equivalent Figure 5-14. Meaning Conveyed. With vs. without L1 equivalents.

155

70% low-intermediate advanced 60% 50.00% 48.15% 50%

40.26% 40% 34.30% 31.88% 33.82% 30.20% 30.66% 29.08% 30% 27.97% 23.88% 19.80% 20%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Light verb + Noun Content verb + Noun Figure 5-15. Meaning Conveyed. Light verb vs. Content verb.

70% low-intermediate advanced 57.68% 60% 54.80%

50% 44.70%

40% 36.88% 33.79% 31.42% 29.11% 29.33% 30% 23.04% 23.88% 19.28% 20% 16.10%

10%

0% 6-4.5 4.33-2.67 2.5-1 6-4.5 4.33-2.67 2.5-1 Same light verb + Noun Different light verb + Noun Figure 5-16. Meaning Conveyed. Same vs. Different light verbs.

The results of these figures above can be divided into two types of distributions: (1) Steep staircases going down to the right, and (2) A very gentle slope or an almost horizontal shape. All the easy types of collocations (literal collocations, collocations with L1 equivalents, light verb collocations, and same light verb collocations) belong to the

‘Steep staircases’ type, whereas all difficult types of collocations (abstract collocations, collocations without L1 equivalents, content verb collocations, and different light verb collocations) fall into the ‘Horizontal’ type. These results imply that more than half of learners’ productions for the easy types of collocations successfully conveyed the

156

intended meaning, while a large part of the more difficult types of collocations were unsuccessful in conveying the intended meaning.

5.4.3. Two-scale analysis of Meaning Conveyed.

In this section, I have again collapsed the Meaning judgments into two groups to directly compare the rates of acceptable vs. unacceptable collocations produced by the learners. To do this, I combined the Mid (2.67-4.33) and Low (1-2.5) ratings into a single

group; I was thus able to compare the rates at which collocations were judged to convey

the intended meaning well (High 4.5-6) or not fully (Mid/Low1-4.33). Chi-square tests

were again used to see whether the percentage with data in a given range varied

significantly in each of these four variable groups as well as between the two proficiency groups. I thus did four analyses focusing on each of the following four variables: literal vs. abstract collocations, existence of L1 equivalents, light vs. content verbs, and same vs. different light verbs.

The first analysis tested the Meaning distributions for literal vs. abstract collocations. Table 5-T and Figure 5-T present the findings.

Literal collocations (A1, A2, A3 literal) Abstract collocations (A1, A2, A3 Meaning abstract) Conveyed High Mid/Low Total High Mid/Low Total rated by natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 584 495 1079 243 649 892 intermediate 54.12% 45.88% 27.24% 72.76% Advanced 571 463 1034 276 572 848 55.41% 44.58% 32.55% 67.45% Total 1155 958 2113 519 1221 1740 54.77% 45.23% 29.90% 70.11% TABLE 5-T. Meaning Conveyed. Literal vs. abstract collocations.

157

80% low-intermediate 72.76% advanced 70% 67.45%

60% 55.41% 54.12% 45.88% 50% 44.58%

40% 32.55% 27.24% 30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Literal Abstract

Figure 5-T. Meaning Conveyed. Literal vs. abstract collocations.

Both low-intermediate and advanced learners did much better on literal than abstract collocations. A chi-square test shows that this difference is statistically significant (Low- intermediate: χ2 = 144.899, df = 1, p = *0; Advanced: χ2 = 96.787, df = 1, p* < .000;

Total: χ2 = 239.513, df = 1, p = *0). Advanced learners produced significantly more

highly rated collocations than did low-intermediate learners for abstract collocations (χ2 =

5.845, df = 1, p = *0.016). No significant difference was observed for low-intermediate

vs. advanced learners’ performance on literal collocations (χ2 = 0.257, df = 1, p = 0.612).

My second analysis compared Meaning scores for collocations with vs. without

L1 equivalents. Tokens and percentages of collocations with vs. without L1 equivalents,

divided into High and Mid-Low ranges of Meaning Conveyed, are displayed in Table 5-U

and Figure 5-U.

158

Meaning With equivalent collocation in L1 (A1) Without equivalent collocation in L1 Conveyed (A2, A3) rated by High Mid/Low Total High Mid/Low Total natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 398 292 690 429 852 1281 intermediate 57.68% 42.32% 33.52% 66.51% Advanced 386 321 707 461 714 1175 54.80% 45.21% 39.23% 60.77% Total 784 613 1397 890 1566 2456 56.24% 43.77% 36.38% 63.64% TABLE 5-U. Meaning Conveyed. Collocations with vs. without L1 equivalents.

80% low-intermediate advanced 70% 66.51% 60.77% 57.68% 60% 54.80%

50% 42.32%45.21% 39.23% 40% 33.52%

30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 With L1 equivalent Without L1 equivalent

Figure 5-U. Meaning Conveyed. Collocations with vs. without L1 equivalents.

Both low-intermediate and advanced learners attained significantly higher scores for collocations with Japanese equivalents than for those without equivalents (Low- intermediate: χ2 = 107.767, df = 1, p = *0; Advanced: χ2 = 42.092, df = 1, p = *0; Total: χ2

= 143.269, df = 1, p = *0). For collocations without L1 equivalents, advanced learners produced significantly more highly rated collocations than did low-intermediate learners

(χ2 = 8.753, df = 1, p = *0.003). There was no significant difference between advanced

and low-intermediate learners in their performance on collocations with L1 equivalents

(χ2 = 1.349, df = 1, p = 0.245).

The third analysis compares performance on content verbs vs. light verbs. Table

159

5-V and Figure 5-V compare low-intermediate and advanced learners’ success with

‘Light verb + Noun’ collocations and ‘Content verb + Noun’ collocations.

Meaning Conveyed Light verb + Noun (A1, A2) Content verb + Noun (A3) rated by natives High Mid/Low Total High Mid/Low Total (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low-intermediate 613 660 1273 214 484 698 48.15% 51.85% 30.66% 69.34% Advanced 625 629 1254 212 406 618 50.00% 50.00% 34.30% 65.70% Total 1238 1289 2527 426 890 1316 49.08% 50.93% 32.48% 67.52% TABLE 5-V. Meaning Conveyed. Light verbs vs. content verbs.

80% low-intermediate advanced 69.34% 70% 65.70%

60% 51.85% 50.00% 50.00% 48.15% 50%

40% 34.30% 30.66% 30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Light verb + Noun Content verb + Noun

Figure 5-V. Meaning Conveyed. Light verb vs. content verb.

Learners did significantly better with ‘Light verb + Noun’ collocations than with ‘Content verb + Noun’ collocations. The difference between light verb vs. content verb collocation types was statistically significant (Low-intermediate: χ2 = 56.658, df = 1, p = *0;

Advanced: χ2 = 40.422, df = 1, p = *0; Total: χ2 = 97.361, df = 1, p = *0). No significant difference was found between low-intermediate and advanced learners for either of these types (Light verb + Noun: χ2 = 0.719, df = 1, p = 0.396; Content verb + Noun: χ2 = 1.989, df = 1, p = 0.158).

160

In my fourth analysis, I tested success with same vs. different light verbs. Table

5-W and Figure 5-W display distributions of native speakers’ ratings of Meaning

Conveyed for ‘Same’ vs. ‘Different’ light verb + Noun collocations.

Meaning Same light verb (A1) Different light verb (A2) Conveyed High Mid/Low Total High Mid/Low Total rated by natives (4.5-6) (1-4.33) (4.5-6) (1-4.33) Low- 398 292 690 429 852 1281 intermediate 57.68% 42.32% 36.88% 63.12% Advanced 386 321 707 461 714 1175 54.80% 45.21% 44.70% 55.30% Total 784 613 1397 890 1566 2456 56.24% 43.76% 40.79% 59.21% TABLE 5-W. Meaning Conveyed. Same vs. different light verb.

80% low-intermediate advanced 70% 63.12% 57.68% 60% 54.80% 55.30%

45.21% 50% 42.32% 44.70%

40% 36.88%

30%

20%

10%

0% High: 6-4.5 Mid/Low: High: 6-4.5 Mid/Low: 4.33-1 4.33-1 Same light verb + Noun Different light verb + Noun

Figure 5-W. Meaning Conveyed. Same vs. different light verb.

The learners produced significantly more highly-rated ‘Same light verb + Noun’ collocations than highly-rated ‘Different light verb + Noun’ collocations (Low- intermediate: χ2 = 85.226, df = 1, p = *0; Advanced: χ2 = 47.162, df = 1, p = *0; Total: χ2

= 130.879, df = 1, p = *0). Advanced learners’ performance on the type ‘Different light verb + Noun’ was significantly better than that of low-intermediate learners (χ2 = 4.211, df = 1, p = *0.040). The difference between Low-intermediate and advanced learner’s

161 performance with ‘Same light verb + Noun’ was statistically insignificant (χ2 = 0.061, df

= 1, p = 0.805).

In summary, Chi-square tests show that learners produced significantly more collocations which conveyed the intended meaning for collocations with the following attributes: literal collocations, collocations with L1 equivalents, light verb + Noun, and same light verb + Noun. They did more poorly for: abstract collocations, collocations without L1 equivalents, content verb + Noun, and different light verb + Noun. Advanced learners performed significantly better than low-intermediate learners on collocations without L1 equivalents, and content verb + Noun collocations. No significant difference was observed in rest of the collocation types.

5.4.4. Hierarchy of difficulty based on Meaning Conveyed.

In the previous sections, we used two- and three-scale analyses to capture learners’ performance on Verb + Noun collocations. I looked at the relationship between native speakers’ ratings of Meaning Conveyed and the four attributes that influenced performance. In this section, I will extract a hierarchy of difficulty based on the percentages of collocations judged to carry intended meaning adequately.

A series of Chi-square tests reveal the following difficulty hierarchy from the easiest collocation type (A1 literal) on the left through the most difficult types (A3 abstract) on the right.

162

A1 abstract A1 literal A2 literal A3 literal (different light A3 abstract (same light (different light (content verb) verb) 31.43% (content verb) verb) verb) 46.28% A2 abstract 15.25% 64.36% 52.50% (same light verb) (MC=4.5-6) 26.59% Easier <------> More difficult TABLE 5-3. Difficulty hierarchy of collocations.

The percentage of learners’ productions that were rated highly was significantly greater

for the literal A1 type (64.36%) than for the literal A2 type (52.50%) (χ2 = 19.964; df = 1;

p* < .000). There were significantly more highly rated collocations of the literal A2 type

(52.50%) than for the literal A3 type (46.28%) (χ2 = 5.173; df = 1; p* = 0.023).

Significantly more high ratings were observed for the literal A3 type than for the abstract

A2 type (26.59%) (χ2 = 38.137; df = 1; p* = 0). The difference between the abstract A1

and A2 types was not significant (χ2 = 3.726; df = 1; p = 0.056). There was significant

difference between the abstract A2 and A3 types (χ2 = 21.723; df = 1; p* < .000).

This hierarchy shows that learners conveyed the intended meanings more

successfully with literal collocations than with abstract collocations with both light and

content verbs.

5.5. Learners’ confidence in their own productions.

As in Experiment 1, we now consider two 10-point scale ratings made by the subjects. First, we consider learners’ confidence in their productions. Table 5-X and

Figure 5-X show the average scores of learners’ confidence in each combination they produced, using a 10-point-scale (10: confident; 1: unconfident).

163

Type of collocations Low-intermediate Advanced learners Total learners (# = 35) (# = 31) (# = 66) A1 Literal 7.39 8.40 7.90 same light V Abstract 5.31 6.04 5.68 A2 Literal 7.27 8.48 7.88 different light V Abstract 4.48 4.95 4.72 A3 Literal 6.31 7.74 7.03 content V Abstract 4.48 6.14 5.31 Total (# = 66) 5.87 6.96 6.42 TABLE 5-X. Average confidence scores for collocation types: A1, A2, A3 (10-point-scale). Experiment 2.

9 8.4 8.48 low-intermediate 7.9 7.88 advanced 7.74 8 7.39 7.27 total 7.03 7 6.31 6.04 6.14 5.68 6 5.31 5.31 4.954.72 5 4.48 4.48

1 literal abstract literal abstract literal abstract

A1 A2 A3

Figure 5-X. Average confidence scores for each collocation type (10-point-scale)

Higher ratings were obtained in the advanced learners’ group (6.96) than in the low- intermediate learners’ group (5.87). Pooling all the subjects, the averages for all literal types were higher (A1: 7.90; A2: 7.88; A3: 7.03) than for abstract types (A1: 5.68; A2:

4.72; A3: 5.31).

As I did for native speakers’ ratings of Grammaticality and Meaning Conveyed, I

tested the distributions for the four variables: (1) Literal vs. abstract meaning; (2) The

164

existence of L1 equivalents; (3) Light vs. content verbs; and (4) Same vs. Different light

verbs.

Literalness of meaning: Table 5-Y and Figure 5-Y show learners’ confidence in

their productions of literal vs. abstract collocations.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Literal (Literal A1, A2, A3) 6.99 8.21 7.60 Abstract (Abstract A1, A2, A3) 4.76 5.71 5.24 TABLE 5-Y. Average confidence for literal vs. abstract collocations. (10-point scale)

9 8.21 low-intermediate 8 7.6 advanced total 6.99 7

6 5.71 5.24

5 4.76

1 L e a r e a e rL n ' s f n i c o i c e ( o e n t 1 0 p d n l s c a e ) Literal collocations Abstract collocations

Figure 5-Y. Average confidence scores for literal vs. abstract collocations. (10-point- scale)

The average for literal collocations was much higher (7.60) than for abstract collocations

(5.24). Also, advanced learners were more confident. Scores for the advanced group were higher than for the low-intermediate group. These results echo those from

Experiment 1 (see 4.5).

Collocations with vs. without L1 equivalents: Table 5-Z and Figure 5-Z display the average confidence scores for collocations produced by each proficiency group on these subsets of collocations.

165

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) With L1 equivalent (A1) 6.35 7.22 6.79 Without L1 equivalent (A2, A3) 5.64 6.83 6.24 TABLE 5-Z. Average confidence scores: collocations with vs. without L1 equivalents. (10-point-scale)

9 low-intermediate advanced 8 total 7.22 7 6.79 6.83 6.35 6.24 6 5.64

1 L e a re r e if L s ' n ( o te n il c a p n c o s c 1 0 n e ) d e With L1 equivalent Without L1 equivalent

Figure 5-Z. Average confidence scores for collocations with vs. without L1 equivalent. (10-point-scale)

The averages for collocations with L1 equivalents were slightly higher (6.79) than for those without (6.24). This result is different from learners’ average confidence scores in

Experiment 1 where learners attained slightly higher scores for collocations without L1 equivalents (see section 4.5). Advanced learners were more confident in their

performance than low-intermediate learners for both subtypes.

Collocations with light vs. content verbs: The average confidence scores for these

subsets of collocation are shown in Table 5-AA and Figure 5-AA.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Light verb + Noun (A1, A2) 6.11 6.97 6.54 Content verb + Noun (A3) 5.40 6.94 6.17 TABLE 5-AA. Average confidence scores: light verb vs. content verb. (10-point-scale)

166

9 low-intermediate advanced 8 total 6.97 6.94 7 6.54 6.11 6.17 6 5.4

1 L e a rL e r n s ' f n c o i c e ( i e n o d 1 0 p t n s c a l e ) Light verb + Noun Content verb + Noun

Figure 5-AA. Average confidence scores: light verb vs. content verb. (10-point-scale)

Overall, learners in both proficiency groups were more confident in their productions with light verb types (6.54) than with content verb types (6.17). This difference was more notable in the low-intermediate group than in the advanced group. For both subtypes, advanced learners were more confident than low-intermediate learners.

Confidence on productions with same vs. different light verbs: These average scores are given in Table 5-BB and Figure 5-BB.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Same light verb + Noun (A1) 6.35 7.22 6.79 Different light verb + Noun (A2) 5.88 6.72 6.30 TABLE 5-BB. Average confidence scores: same vs. different light verb. (10-point-scale)

167

9 low-intermediate advanced 8 total 7.22 7 6.79 6.72 6.35 6.3 5.88 6

1 L e a rL e r n s ' f n c o i c e ( e n i d o 1 0 p t n s c a l e ) Same light verb + Noun Different light verb + Noun

Figure 5-BB. Average confidence scores: same vs. different light verb. (10-point-scale)

Both proficiency groups were more confident with collocations with the same verb types than for different ones. This result differs slightly from Experiment 1 where learners were a little more confident in collocation types with different light verbs. The average ratings by the advanced group were higher than for the low-intermediate group for both subtypes.

To summarize, overall the learners indicated higher confidence for all the easier types: (1) Literal; (2) With L1 equivalents; (3) Light verbs; (4) Same light verbs. Among these four variables, the difference in the confidence ratings between literal vs. abstract types (2.36) was greater than for other subtypes (With vs. without L1 equivalent: 0.55;

Light vs. content verb: 0.37; Same vs. different light verb: 0.49). The average confidence score for the literal types (7.60) was higher than scores for any other subtypes, while the score for the abstract types (5.24) was the lowest of all. This result suggests that learners were most confident in literal types and least confident in abstract types, with abstract subtypes more difficult than any of the other difficult subtypes (i.e., Without L1 equivalent, content verbs, and different light verbs).

168

In the next section, we examine the second 10-point scale rating: familiarity with

their own productions.

5.6. Learners’ familiarity with their own productions

In this section, we consider learners’ familiarity ratings. They rated each item

they produced: from 10: Has heard or read this combination produced by native speakers

of English; to 1: Has never heard or read this combination produced by native speakers.

The average ratings of learners’ familiarity with their own productions for each collocation type are displayed in Table 5-CC and Figure 5-CC.

Type of collocations Low-intermediate Advanced learners Total learners (# = 35) (# = 31) (# = 66) A1 Literal 7.08 8.17 7.63 same light V Abstract 4.30 5.23 4.77 A2 Literal 6.95 8.26 7.61 different light V Abstract 3.55 4.31 3.93 A3 Literal 5.86 7.08 6.47 content V Abstract 3.77 5.35 4.56 Total (# = 66) 5.25 6.40 5.83 TABLE 5-CC. Average familiarity scores for each collocation type: A1, A2, A3 (10-point- scale). Experiment 2

169

9 low-intermediate 8.17 8.26 advanced 8 7.63 7.61 total 7.08 6.95 7.08 7 6.47

6 5.86 5.23 5.355.31 5 4.77 4.3 4.31 3.93 3.77 4 3.55

2 H e a r d o r r e a d ( 1 0 p o i n t s c a l e ) n t s c a l e a r d o ( 1 0 p i H 1 literal abstract literal abstract literal abstract

A1 A2 A3

Figure 5-CC. Average familiarity scores for each collocation type: A1, A2, A3 (10-point- scale). Experiment 2.

As with learners’ confidence, their average familiarity ratings for all the literal types were

higher than for the abstract types. In addition, the advanced group claimed greater

familiarity for all the subtypes than did the low-intermediate group.

I tested the influence of each of the four variables: (1) Literal vs. abstract meaning; (2) With vs. without L1 equivalents; (3) Light vs. content verbs; and (4) Same

vs. different light verbs. Literal vs. abstract meaning: Learners’ average familiarity

ratings for these subtypes are shown in Table 5-DD and Figure 5-DD.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Literal (Literal A1, A2, A3) 6.63 7.84 7.24 Abstract (Abstract A1, A2, A3) 3.87 4.96 4.42 TABLE 5-DD. Average familiarity ratings: Literal vs. abstract collocations. (10-point- scale)

170

) 9 low-intermediate 8 7.84 advanced 7.24 total 7 6.63

4.96 5 4.42 3.87 4

1 L e a r n e r s ' f a m i l i a r i t y ( 1 0 p o i n t s c a l e i n t o ar i y ( 1 0 p i l fam e r s ' n L e a r Literal collocations Abstract collocations

Figure 5-DD. Average familiarity ratings: Literal vs. abstract collocations. (10-point- scale)

As with their confidence scores, both proficiency groups claimed much greater familiarity with literal collocations (7.24) than abstract ones (4.42). Advanced learners gave higher ratings than low-intermediate learners for both subtypes.

Existence of L1 equivalents: Table 5-EE and Figure 5-EE summarize the average scores of the subjects’ familiarity scores for these subtypes.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) With L1 equivalent (A1) 5.69 6.70 6.20 Without L1 equivalent (A2, A3) 5.03 6.25 5.64 TABLE 5-EE. Average familiarity scores: collocations with vs. without L1 equivalent. (10-point-scale)

171

) 9 low-intermediate 8 advanced total

7 6.7 6.2 6.25 6 5.69 5.64 5.03 5

1 L e a r n e r s ' f a m i l i a r i t y ( 1 0 p o i n t sin t c a l e i l a r to ( 1 p y f 0 e r s ' e a m L a r n With L1 equivalent Without L1 equivalent

Figure 5-EE. Average familiarity scores: Collocations with vs. without L1 equivalent. (10-point-scale)

The subjects’ familiarity ratings for collocations which have L1 equivalents were

somewhat higher than for those without direct equivalents. The difference between these

subtypes was slightly greater in the low-intermediate group (0.66) than in the advanced group (0.45). Advanced learners claimed greater familiarity than did low-intermediate

learners for both types of collocations.

Light vs. content verbs: The average familiarity rating for these produced

collocations by each proficiency group are shown in Table 5-FF and Figure 5-FF.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Light verb + Noun (A1, A2) 5.47 6.49 5.98 Content verb + Noun (A3) 4.82 6.22 5.52 TABLE 5-FF. Average familiarity scores: light vs. content verb. (10-point-scale)

172

) 9 low-intermediate 8 advanced total

7 6.49 6.22 5.98 6 5.47 5.52

5 4.82

1 L e a r n e r i e ' e a r s n lL f i a m a r i t i y ( t o n 1 0 p s c a l e Light verb + Noun Content verb + Noun

Figure 5-FF. Average familiarity scores: light vs. content verb. (10-point-scale)

Both proficiency groups indicated higher familiarity with light verb types than with content verb types. The difference between these subtypes was greater for low- intermediate learners (0.65) than for the advanced learners (0.27). The advanced group’s rating scores were higher for both subtypes.

Same vs. different light verbs: Table 5-GG and Figure 5-GG display the average familiarity scores for each proficiency group.

Types of Collocation Low-intermediate Advanced Total learners (# = 35) learners (# = 31) (# = 66) Same light verb + Noun (A1) 5.69 6.70 6.20 Different light verb + Noun (A2) 4.82 6.22 5.52 TABLE 5-GG. Average familiarity scores: same vs. different light verb. (10-point-scale)

173

) 9 low-intermediate 8 advanced total

7 6.7 6.2 6.22 6 5.69 5.52

5 4.82

1 L e a r n e r s ' f a m i l i a r i t y ( 1 0 p o i n ta li is c e l n i o r i ' t a fr s ( p e m y 1 0 a n a r e L Same light verb + Noun Different light verb + Noun

Figure 5-GG. Average familiarity scores: same vs. different light verb. (10-point-scale)

The average scores for same light verb type were higher than for different light verb type.

As with the other variables, the difference between these subtypes was greater for the low-intermediate learners (0.87) than for the advanced learners (0.48). Advanced learners scored higher for both types of collocations.

In summary, the results for this analysis are quite similar to the results from learners’ confidence in their productions (see 5.5). In general, learners’ familiarity scores were higher for all the easier types. Among the four variables, the difference was the greatest in literal vs. abstract types. Pooling the data from both groups, the highest average ratings were observed for the literal type (7.24) while the lowest ratings were for the abstract type (4.42). This suggests that the subjects felt they had heard or read more of the literal collocations than the abstract ones. Advanced learners’ scores were higher than lower-intermediate learners’ for all the subtypes. Moreover, the differences between each variable type were greater in the low-intermediate group than in the advanced group.

That is, the low-intermediate learners felt they had not received much input for the more difficult types of collocations.

174

5.7. Multi-variable analyses of results

In sections 5.3-5.6, we examined Grammaticality and Meaning Conveyed as rated

by native speakers, and learners’ Confidence and Familiarity, one variable at a time. We

also considered how the four easy vs. difficult factors (literalness, with vs. without L1 equivalents, light vs. content verbs, same vs. different verbs) influenced learners’ performance. In this section, we consider pairs of native speakers’ ratings

(Grammaticality or Meaning Conveyed) and learners’ confidence ratings. First, we look at Grammaticality Judgments paired with learners’ confidence (4.7.1), then Meaning

Conveyed against learners’ confidence (4.7.2). Finally we compare learners’ confidence vs. familiarity with their productions (4.7.3).

5.7.1. Grammaticality Judgments vs. learners’ confidence

The first analysis is the relationship between native speakers’ Grammaticality ratings and learners’ confidence in their own productions. I used the same procedure as in the corresponding analysis in Experiment 1 (see 4.7.1) to summarize the data. Table

IX-5 (in Appendix IX) displays Grammaticality Judgments and the subjects’ confidence ratings, each of which were divided into two groups: Grammatical vs. Ungrammatical and Confident vs. Not confident. Figure 5-17 compares Grammaticality when learners were vs. were not confident in their productions, and Figure 5-18 compares learners’ confidence when their productions were judged to be grammatical vs. ungrammatical.

175

Figure 5-17. Grammaticality when learners were vs. were not confident in their productions.

There was no significant difference across the proficiency groups in the proportions of grammatical combinations when the subjects were confident in their productions (χ2 = 0.002; df = 1; p = 0.9643) or not so (χ2 = 0.302; df = 1; p = 0.5826).

Statistical analysis shows that the differences in the proportions of confident

productions judged to be grammatical or not were significant in both proficiency groups

(Low: χ2 = 91.017; df = 1; p* = 0; Advanced: χ2 = 90.596; df = 1; p* = 0; Total: χ2 =

183.748; df = 1; p* = 0). Both groups gave higher confidence scores when their

combinations were grammatical.

176

Figure 5-18. Learners’ confidence when their productions were grammatical vs. ungrammatical.

A significant difference was found in the proportion of confident productions when the combinations were grammatical vs. ungrammatical. Advanced learners were more confident with their grammatical production (74.63%) than were low-intermediate learners (61.55%) (χ2 = 28.353; df = 1; p* < 0.000). A difference was observed when the

subjects produced ungrammatical combinations. Advanced learners rated their

productions more confidently (52.70%) than did low-intermediate learners (39.37%) (χ2 =

42.291; df = 1; p* = 0).

As we saw in the corresponding analyses for Experiment 1 (section 4.7.1),

advanced learners are more confident of their productions regardless of their grammaticality; they are often confident even when they are not grammatical.

5.7.2. Meaning Conveyed vs. learners’ confidence

This section combines native speakers’ ratings of Meaning Conveyed with learners’ confidence in their productions. Using the same criteria as for Experiment 1

177

(see 4.7.2.) these variables were divided into two groups: Meaning Conveyed scores

between 4.5 and 6 vs. not conveyed (scores: 1-4.33), and Confident (7-10 on the 10-point

scale) vs. Not confident (1-6) in their productions. Table IX-6 (See Appendix IX)

juxtaposes Meaning Conveyed as rated by native speakers with the subjects’ confidence

ratings. Figures 5-19 and 5-20 graph pooled data. Figure 5-19 compares ratings of

Meaning Conveyed when learners were confident vs. not confident, and Figure 5-20

compares learners’ confidence when their productions did vs. did not convey the intended meaning.

Figure 5-19. Meaning conveyed or not conveyed when learners were confident vs. not confident in their productions.

As we saw in the Grammaticality vs. Confidence in section 5.7.1., there was no significant difference between the low-intermediate and the advanced groups in the amount of combinations which were judged to convey the intended meaning when the subjects were vs. were not confident in their production (Confident: χ2 = 1.092; df = 1; p

= 0.2960; Unconfident: χ2 = 1.044; df = 1; p = 0.3069).

178

Figure 5-20. Learners’ confidence when their combinations did vs. did not convey the intended meaning.

The differences in the proportions of confident productions when Meaning was vs. was not conveyed were significantly different in both proficiency groups (Low: χ2 =

93.919; df = 1; p* = 0; Advanced: χ2 = 124.565; df = 1; p* = 0; Total: χ2 = 220.461; df =

1; p* = 0). All the subjects gave higher confidence scores when their productions

actually conveyed the intended meaning.

Significant differences were found in the percentages of confident combinations

when their productions were judged to convey vs. not to convey the meaning. When

productions convey the intended meaning, advanced learners rated more of them

confidently (73.67%) than did low-intermediate learners (60.10%) (χ2 = 34.858; df = 1;

p* = 0). A significant difference was observed across the proficiency groups for

combinations which did not convey the intended meaning. Advanced learners were more confident in these productions (48.31%) than were low-intermediate learners (38.04%)

179

(χ2 = 23.473; df = 1; p* < 0.0000).

5.7.3. Learners’ confidence vs. Heard or Read

In sections 5.7.1. and 5.7.2., we examined learners’ confidence and familiarity with their productions separately. In this section we briefly consider learners’ confidence and familiarity ratings together. Overall the subjects rated their confidence and familiarity with the same numbers on the 10-point-scales. The results were quite similar to Experiment 1 (see 4.7.3): there was no significant difference in the way subjects rated their confidence vs. familiarity. However, as we saw in Experiment 1, the subjects generally gave slightly higher confidence than familiarity scores. Table IX-7 (in

Appendix IX) summarizes the average differences between confidence and familiarity ratings by group for each collocation subtype. Figures 5-21, 5-22, 5-22, 5-23 and 5-24 show the average differences for each subtype and Figure 5-25 divides the data into easy vs. difficult subtypes.

Figure 5-21. Learners’ confidence and familiarity (10-point scale): literal vs. abstract.

180

Figure 5-22. Learners’ confidence and familiarity (10-point scale): With vs. without L1 equivalent.

Figure 5-23. Learners’ confidence and familiarity (10-point scale): light vs. content verbs.

181

Figure 5-24. Learners’ confidence and familiarity (10-point scale): same vs. different light verbs.

Figure 5-25. Learners’ confidence and familiarity (10-point scale): easy vs. difficult types.

182

Figure 5-26. Learners’ confidence and familiarity (10-point scale): Total.

The differences between the confidence and familiarity ratings were slightly greater for low-intermediate learners than for advanced learners for all the subtypes except for literal types and content verb types. There was no systematic difference between the easy and difficult subtypes among the subtypes except for literal vs. abstract types. Compared to the differences in the confidence and the familiarity scores of literal types (Low- intermediate: 0.36; Advanced: 0.37; Total: 0.36), there were much greater differences between these two scores for abstract types (Low-intermediate: 0.89: Advanced: 0.75;

Total: 0.82). No large differences were observed for the other subtype pairs. This suggests that literalness of meaning influences learners’ performance more than do other variables. Learners are more comfortable with literal collocations than with the other easy types (with L1 translation, light verbs, and same light verbs), while they are less comfortable with abstract collocations than any of the other difficult types (without L1 translation, content verbs, and different light verbs).

183

5.8. Discussion of Experiment 2

Experiment 2 examined four factors (meaning, existence of L1 equivalence, light

vs. content verbs, and same vs. different light verbs). Meaning is the new factor in this

experiment. The other three factors are the same as in Experiment 1. The results in

Experiment 2 were quite similar to Experiment 1 in that all four factors influence

learners’ productions and perception. The collocations can again be divided into easy vs.

difficult types as below:

Variables Easy types More difficult types (1) Meaning Literal meaning Abstract meaning (2) Existence of L1 equivalents With L1 equivalents Without L1 equivalents (3) Light verbs vs. content verbs Light verbs Content verbs (4) Same vs. different light verbs Same light verbs Different light verbs TABLE 5-4. Easy vs. difficult types of collocations.

All the results from native speaker ratings of Grammaticality and Meaning

Conveyed show that subjects attained higher scores for the easy types than for the

difficult types. The results from the three-scale analyses of Grammaticality Judgments show patterns similar to those in Experiment 1. All the figures (see Figures 5-6, 5-7, 5-7 and 5-8) resemble a v-shaped hollow between 2 hills, suggesting that native speakers’ judgments are quite clear-cut although there are some individual variations. In contrast, the results from the Meaning Conveyed judgments are somewhat different from those in

Experiment 1. The easy types resemble staircases going down to the right in both experiments. The results from the difficult types in Experiment 1 generally resemble a

184

gentle staircase going down to the right. On the other hand, in Experiment 2, the difficult

types exhibit nearly the same percentages for high, mid and low ratings. The averages for

the easy types were: high = 54.08%; mid = 30.14%; low = 18.78%; while the averages

for the difficult types were: high = 34.89%; mid = 33.96%; low = 31.15%.

Figure 5-27. High, mid low scores of the Grammaticality Judgments. Easy vs. difficult types.

These results suggest that conveying intended meanings for the difficult types in

Experiment 2 was more difficult than conveying meaning of the difficult types in

Experiment 1.

In the following sections I discuss these results in relation to the points discussed in 5.5 and 5.6: (1) number of collocations produced; (2) the four factors which influence productions; and (3) learners’ confidence and familiarity.

5.8.1. Number of collocations produced

Advanced learners generally produced more collocations (2.04) than did low-

185

intermediate learners (1.89) for most of the subtypes (both easy and difficult), except for

two: (1) without L1 equivalents (low-intermediate = 2; advanced = 1.91), and (2) those

with content verbs (both low-intermediate and advanced = 2). As discussed in 4.8.1.,

fewer productions by low-intermediate learners may indicate their overall lack of

confidence in giving answers, their use of an avoidance strategy or their smaller

vocabulary size.

No systematic relation was observed between the number of collocations

produced by learners for easy vs. difficult subtypes. Both proficiency groups produced

more collocations for the literal and same light verb types than for the abstract and

different light verb types. There was no clear difference in the other two pairs: with vs.

without L1 equivalents, and light vs. content verbs. This result is somewhat different

from Experiment 1 where the subjects produced more collocations for all the difficult

types. One possible explanation of why, in Experiment 2, the subjects did not always

produce more collocations for the difficult types is that, all the collocations tested in

Experiment 2 were relatively difficult for the subjects, therefore, no clear difference was

observed between the easy vs. difficult types. As mentioned in 4.8.2., Japanese students only learn a very small number of highly frequent collocations in formal instruction in

Japan. Although all the collocations tested in Experiment 2 were relatively frequent, the majority of them probably do not appear in textbooks in Japan. As a result, the subjects probably perceived most of the collocations as relatively difficult.

However, slight differences were observed for the easy vs. difficult types in the ranges of collocations produced by the subjects. Figure 5-28 shows the average numbers of different tokens produced for each item grouped by of subtype.

186

Figure 5-28. The average numbers of different tokens produced per item of each subtype.

Although the differences in the ranges between easy vs. difficult subtypes are not as large as in Experiment 1 (see Figure 4-27), both proficiency groups produced a slightly

greater range of collocations for difficult types than for the easy types. This result concurs with Bardovi-Harlig’s study (2008) on the developmental stages of L2 acquisition of formulas where learners produced a greater variety of formulae at the early stages of acquisition; as their proficiency increases, their productions exhibit less

variation.

5.8.2. Literal vs. abstract meaning

Literal collocations are easier than abstract collocations for learners (χ2 = 139.163,

df = 1, p* = 0). Literalness of meaning seems to influence learners’ performance more

strongly than the other factors tested in this study. Part of the reason for this is that abstract meanings seem to cause great difficulty. Irujo (1986) proposed non-literalness of

187

meaning as one of several explanations for the fact that figurative expressions are

difficult to learn in a second language. In my data, 66/68 subjects produced the literal

collocation take a bath while only 7 produced the abstract collocation take a dive, in spite of the fact that these two collocations are both relatively frequent, do not have an L1 equivalent, and both use the verb take. Take a bath denotes the literal action of washing, whereas take a dive means to pretend to be knocked out and has nothing to do with the literal meaning of dive.

Another possible explanation is Kellerman and Jordens’ notion of psychological markedness (1977). They state that: “Psychologically marked features tend to be perceptually more complex, and are thus less likely to serve as models of transfer than unmarked ones.” In other words, core, literal meanings are acquired easily while abstract meanings are difficult for L2 learners. It is possible that my subjects underused their L1 knowledge when producing abstract collocations because of their uncertainty about transferability from Japanese to English. Indeed, the subjects committed fewer transfer errors for the abstract collocations (8.02%) than for literal collocations (14.14%). It appears that learners’ intuitions about the range of prototypicality affect their performance of collocations.

5.8.3. With vs. without L1 equivalents

The subjects did better on collocations with L1 equivalents than on those without.

As in Experiment 1, the low-intermediate group did slightly better than the advanced group on collocations with L1 equivalents with regards to both Grammaticality

(percentage of 4.5+: low-intermediate = 51.43%; advanced = 47.88%) and Meaning

188

Conveyed (4.5+: low-intermediate = 57.68%; advanced = 54.80%). In contrast,

advanced learners attained higher scores for collocations without L1 equivalents (GJ

4.5+: low-intermediate = 28.92%; advanced = 34.25%; MC 4.5+: low-intermediate =

33.52%; advanced = 39.23%). It is not certain whether the low-intermediate learners

demonstrated higher proficiency in collocations with L1 equivalents. It is possible that

the low-intermediate learners overused their L1 knowledge more than did advanced

learners, or that advanced learners avoided using their L1 knowledge, as a result, their

accuracy percentages were slightly higher only for (1) those with L1 equivalents and (2)

same light verb types. Advanced learners’ accuracy was higher for rest of all the

subtypes.

These results add more support for the idea that the existence of L1 equivalents

can facilitate learning (Bahn, 1993; Caroli, 1998) in spite of typological linguistic

differences between English and Japanese.

At the same time, learners find collocations without L1 equivalents more difficult

(Bahns & Eldaw, 1990, Koya, 2002, Murano, 2004, Nesselhauf, 2003). The lack of L1

equivalents may cause various problems for learners, including L1 transfer, confusion

with other words which have similar meanings and so forth. As in Experiment 1, these

results show that the learners’ L1 plays a role in creating both positive and negative

transfer of forms from L1 to L2 (Bahns and Eldaw, 1993, Fayez-Hussein, 1990, Marton,

1978).

5.8.4. Light verbs vs. content verbs

Greater accuracy was observed for the light verb types than for the content verb

189

types, in fact content verb collocations were the most difficult. We can speculate that the

restricted meanings of content verbs cause difficulty for learning. The subjects were not

very successful as measured either by Grammaticality Judgments or Meaning Conveyed

(GJ 4.5+: low-intermediate = 23.08%; advanced = 28.09%).

At the same time light verbs do not appear to be very easy, either. Although they

group as an easy type, the accuracy percentages for both the Grammaticality Judgments

(44.49%) and the Meaning Conveyed (49.08%) were not as high as those of other easy

types (literal type: GJ = 46.58%; MC = 54.77%, with L1 equivalents: GJ = 49.66%; MC

= 56.24%; and same light verbs: GJ = 49.66%; MC = 56.24%. The same results were

obtained in Experiment 1.

As discussed in 4.8.3., the more accurate (GJ & MC 4.5+) percentages for light

verb types might overestimate learners’ proficiency, because the subjects may have

overproduced light verbs resulting in frequent misuse. As discussed for Experiment 1,

the semantic vagueness of light verbs may cause difficulty in learning collocations.

5.8.5. Same vs. different light verbs

The existence of L1 equivalents appears to facilitate the acquisition of light verbs.

As with the contrast between collocations with vs. without L1 equivalents, the low- intermediate learners scored slightly higher on same light verbs (4.5+ GJ = 51.43%; MC

= 57.08%) than did the advanced learners (GJ = 47.88%; MC = 54.80%). However, this could be because the low-intermediate learners overused their L1 knowledge and/or the advanced learners’ underused it. The advanced learners scored higher on the different

light verb type (4.5+ GJ = 40.78%; MC = 44.70%) than did the low-intermediate learners

190

(GJ = 35.67%; MC = 36.88%). In addition, the frequent overuse of light verbs could have inflated the learners’ accuracy with light verbs in general.

Each of these four factors contributes to learners’ difficulty with collocations.

The difficulty hierarchy (see Table 5-3) shows the easiest type (A1 literal, same light verb; and A2 literal, different light verb) on the left end with the more difficult types (A3 abstract, content verb) toward the right. The more difficult factors a collocation carries, the more difficult it seems to be for L2 learners to acquire.

5.8.6. Learners’ confidence and familiarity

Advanced learners’ indicated greater confidence in their productions (6.96 on the

10-point scale) than did low-intermediate learners (5.87) (See Table 5-X). The scores for literal types were higher (7.6) than for abstract types (5.24) (see Table 5-Y and Figure 5-

Y), but a clear difference was not observed for the other contrasts (with vs. without L1

equivalents types, light vs. content verbs, and same vs. different light verbs).

Advanced learners’ familiarity ratings (Heard or Read) were also higher (6.40)

than for low-intermediate learners’ (5.25) (See Table 5-CC). As with the confidence

results, subjects indicated greater familiarity (Heard and Read) for the literal types (7.24)

than for the abstract types (4.42). No clear difference was observed for the other contrasts (with vs. without L1 equivalents types, light vs. content verbs, and same vs. different light verbs).

Both the learners’ confidence and familiarity ratings in this experiment were somewhat lower than in Experiment 1. This implies that the subjects perceived many of

the collocations in Experiment 2 as more difficult than those in Experiment 1. It is

191 possible that the lower frequency of the collocations tested in Experiment 2 caused difficulty for the subjects, or that literal vs. abstract meaning may have influenced learners’ performance more strongly than the other variables.

As in Experiment 1, the advanced learners may be overconfident in their combinations even when their combinations are unacceptable. Again, this result poses the problem that overconfidence might be one factor which makes it difficult to attain more native-like proficiency.

5.9. General discussion of Experiments 1 and 2

The results from Experiments 1 and 2 show trends in learners’ development of

Verb + Noun collocations that are, on the whole, consistent with my predictions although there are some discrepancies.

5.9.1. Support for predictions

Prediction 1 that learners would demonstrate better productive knowledge of frequent collocations as compared to less frequent ones, was supported. Both low- intermediate and advanced learners attained higher scores for both Grammaticality and

Meaning Conveyed for the frequent types than infrequent ones. The average percentages of acceptable (GJ and MC 4.5+) combinations in Experiment 2 (all the collocations are neither frequent nor infrequent) fall between the scores for frequent and infrequent collocations as shown in Figure 5-29 (frequent: the frequent collocations of Experiment

1; neither frequent nor infrequent: all the collocations of Experiment 2; infrequent: the infrequent collocations of Experiment 1):

192

Figure 5-29. Acceptable collocations. Frequent, neither frequent nor infrequent and infrequent types.

This result suggests that learners’ difficulties can be ranged on a cline from frequent

collocations being easier to infrequent collocations being more difficult. Japanese

students learn a very small number of highly frequent collocations in formal instruction

in Japan, that is, the majority of collocations of (1) neither frequent nor infrequent and (2)

infrequent types are not presented in school. In spite of that, the students attained higher

grammaticality scores for neither frequent nor infrequent types than infrequent types.

This may because of the input they received in English-speaking countries, or during self

study. In the future, it would be interesting to compare Japanese students in Japan and in

an English-speaking country.

Prediction 2 that learners would demonstrate better productive knowledge of

collocations which have literal meaning, was also supported. In Experiment 2 both proficiency groups performed better with literal collocations than with abstract

193

collocations. The percentages of acceptable collocations (GJ and MC 4.5+) in

Experiment 1 (all the collocations are relatively literal) fall between literal and abstract

types.

Figure 5-30. Acceptable collocations. Frequent, neither frequent nor infrequent and infrequent types.

This result again implies learners’ difficulties can be described as a continuum; from literal collocations on the left being easier to abstract collocations on the right being more difficult. But it is necessary to collect more data to confirm this result because, in this study, the collocations in Experiment 1 were either frequent or infrequent while all the collocations in Experiment 2 are neither frequent nor infrequent.

Prediction 3 that learners would perform better if there is a translational equivalent in L1, was supported. My subjects did significantly better on collocations with L1 equivalents than on those without (χ2 = 246.168, df = 1, p* = 0). This greater

accuracy for collocations with L1 equivalents appears to reflect transfer from L1 (Bahns,

1993). The learners’ L1 plays a role in creating both positive and negative transfer of

194

forms from L1 to L2 (Bahns and Eldaw, 1993, Fayez-Hussein, 1990, Marton, 1978).

However, the existence of L1 equivalents does not always ensure success. For

example, among the eight light verbs tested, my subjects seemed to underuse their L1

knowledge when they were asked to translate suru (do) regardless of the existence of L1 equivalents. The accuracy for translating suru into English do was only 32.59%, while the overall average for all the light verbs with English equivalents was 49.59%. Part of the reason for this underuse is differences in the morphosyntactic behavior of English do

vs. Japanese suru. English do is usually used only with nouns to denote an activity or a

task (i.e., do the garden, do the work, do the teeth, etc). On the other hand, in Japanese

suru can be used with adjectives and adverbs as well as nouns (i.e., shigoto-o suru (work-

acc do) ‘do the work’, kireini suru (clean do) ‘make something clean’, urusaku suru

(noisily do) ‘be noisy’). The more restricted syntactic behavior of English do may

contribute to Japanese speakers’ underuse of it.

Another notable feature is very frequent transfer of eru (get) and ataeru (give).

My subjects frequently used get when translating eru (get) and give when translating

ataeru (give) whether or not an equivalent collocation exists in English. As a result, the

percentages of negative transfer with these verbs were quite high. Figure 5-31 displays

the percentages of negative transfer for each Japanese light verb.

195

Figure 5-31. Percentages of negative transfer for each Japanese light verb.

The high frequency of interference with ataeru and eru may be due to the corresponding

patterns of light verbs in English and Japanese. Ataeru and eru most frequently

correspond to just a single English verb (give and get, respectively) while the other light

verbs mapped onto many English words. Because of these different patterns, learners

may expect ataeru and eru always to be translated into give and get, in spite of

exceptions such as *give a great impression for make/leave a great impression, and *get

the heart for win the heart.

Prediction 4 that light verbs are more difficult than content verbs, was not

supported. Higher accuracy scores were observed for the light verb types than for the

content verb types (χ2 = 224.48, df = 1, p* = 0). The restricted meaning of content verbs might cause difficulty for learning. However, because of the design of this study, we are not yet justified in concluding that light verbs are always easier than content verbs. First, among the 66 collocations tested, the target for 44 of them was a light verb. This uneven distribution of light and content verbs might have prompted learners to produce more

196

light verbs than they usually would.

A second reason is, since the subjects were asked to produce as many

combinations as they could think of, they might have produced many light verbs almost

randomly in an attempt to make their collocations sound native-like. These attempts were

not always successful; misuse of light verbs was extremely frequent. Figure 5-32 shows the numbers of misuse for each light verb.

Figure 5-32. Misuse of light verbs.

Among the eight light verbs on the test, have, get, make and take were misused much more frequently. We have no way to tell whether or not these subjects actually misuse light verbs in their daily life as often as they did during the experiment. However, the results clearly indicate learners’ insufficient knowledge of these four light verbs. A pedagogical implication is that teachers would do well to focus on these four light verbs.

Prediction 5 that advanced learners would perform significantly better than low- intermediate learners was supported. Improved performance was observed at the higher

197 proficiency level; 35.88% of low-intermediate learners’ collocations and 39.92% of advanced learners’ collocations were judged to be native-like (i.e. average scores of

Grammaticality Judgments were 4.5 or higher). This difference is statistically significant

(χ2 = 15.478; df = 1; p* < .000). This suggests that learners’ collocational competence correlates with learners’ general language proficiency.

5.9.2. Other remarks

There are two more points to be discussed in this section. The first point regards types of errors and the second point L1 transfer. These points need to be discussed here since I found more types of errors in addition to the six error types found in the preliminary study (section 2.3), and L1 transfer seems to be related to various factors in addition to the influence of L1 knowledge and negative/positive transfer mentioned in sections 4.8 and 5.8.

5.9.2.1. Types of errors

The following eleven error types were identified: (1) Paraphrases; (2) Misuse of light verbs; (3) L1 transfer; (4) Blending two collocations; (5) Morphological synonymy; (6)

Use of words other than verbs; (7) Unnecessary articles and particles; (8) Intransitive vs. transitive distinctions; (9) Creating collocations from compound nouns; (10) Mistakes with actors; and (11) Phonological errors. Error types (1)-(6) were observed in the preliminary study (See section 2.3 for examples). Error types (7)-(11) are new types found in Experiments 1 and 2. Examples of these are provided below:

198

[7] Unnecessary Articles, Particles and Prepositions: Unnecessary articles, possessive pronouns, particles and prepositions were occasionally inserted between verbs and nouns.

(7) a. Skinheads, who killed an Armenian, stood the trial in Moscow on July 26, 2004. (stood trial)

b. Farmers use the latest technology to grow up rice. (grow rice)

There were 10 instances where unnecessary articles and possessive pronouns were used, and 15 instances of inserting unnecessary particles and prepositions. The insertion of unnecessary elements causes a considerable drop in the Grammaticality and Meaning

Conveyed scores. When unnecessary articles or possessive pronouns were inserted between the target verb and the NP, the average scores were GJ = 3.77 and MC = 4.38.

When an unnecessary particle and/or preposition was inserted between the target verb and the noun phrase, the average scores were GJ = 2.47 and MC = 3.18.

[8] Intransitive vs. Transitive Distinctions: Low-intermediate learners used intransitive verbs with similar semantic content. Errors were particularly common when verbs denoted existence of an element or occurrence of an event.

(8) a. At eight Stella existed a class in General Biology. (had a class)

b. This year we have happened a cut in case-management funding of $150,000. (had a cut)

[9] Creating Collocations from Compound Nouns: Learners decomposed compound

nouns and created collocations from the parts.

199

(9) a. On September 13th we eyed witness to an event that should serve as a turning point in the history of the Middle East. (‘eye-witness’, bore witness)

b. This year we have budget a cut in case-management funding of $150,000. (‘budget-cut’, had a cut)

[10] Mistakes with Actors: There were a few errors in which learners misunderstood the roles of the actors in a sentence. For example, the subject of ‘take an order’ is a person

who collects orders from customers (i.e., waiters, sellers, businessmen); the subject is not

the person who places the order. Similarly, the Subject of deliver a newspaper is the

person who visits houses to distribute newspapers to his/her customers, and not the person who subscribes to the newspaper.

(10) a. Have you ever taken an order online with your credit card? (placed an order)

b. Over 20% of the students are delivering a newspaper each day. (taking a newspaper)

[11] Phonological Errors: Phonological similarity may also lead to errors in collocations.

For example, a learner may produce the erroneous form That is why rising questions about what happened on Haifa Street is very important, because of the phonological similarity between the word rising and raising. Here are a few more examples.

(11) a. Mary’s dream is to find a school for young children and to develop a curriculum that blends art and science. (found a school)

b. In Manila seaweed farmers failed victim to a strong peso. (fell victim)

Among these 11 error types, the most frequent errors were paraphrases and misuse of

light verbs followed by L1 transfer. In the following sections, I discuss L1 transfer and

200 misuse of light verbs.

5.9.2.2. L1 transfer

Some studies have mentioned the influence of learners’ L1 in learning collocations. Kellerman (1977, 1978, 1979, 2000) suggests in his Markedness theory that

L2 learners work on the hypothesis that there are constraints on how similar an L2 can be to their L1. In acceptability tests and translation tasks, L2 learners accept structures which they perceive as unmarked (frequent, literal meaning, syntactically productive) while they tend to reject what they perceive as marked since they think they are language specific items. If Kellerman’s markedness theory is applied to this study, there should be more L1 transfer for easy types of collocations: frequent, literal and light verb types. In contrast, Nesselhauf (2003) investigated French learners’ acquisition of verb-noun combinations involving make and take. She claims that L1 transfer does not seem to decrease with the degree of idiomaticity, but instead, restricted collocations are most susceptible to transfer.

In this study, there were a total of 626 tokens of erroneous collocations which appeared to be result from the influence of L1 Japanese. Because of the small amount of data in this study, transfer does not always seem to reflect marked-unmarked distinctions, including frequent vs. infrequent, literal vs. abstract meaning and light vs. content verbs.

However there was a tendency for transfer to be more common in the following types: (1) infrequent; (2) literal; and (3) content verb types as displayed in Figure 5-33. There were large differences in the percentages of transfer among collocations of the same subtype.

201

Figure 5-33. Percentages of transfer errors.

For example, among frequent types, 42 subjects produced become victim (target = fall victim) and 30 produced take lunch (have lunch), while only one subject produced receive trial (stand trial), do questions (ask question) and do an eye (keep an eye). The numbers of subjects (total = 66) who produced transfer errors are listed in Table 5-J.

Transfer errors Target verb # of Transfer errors Target verb # of subject subject Frequent type Infrequent type become victim fall 42 make rice grow 54 take lunch have 30 make a school found 49 do suicide commit 24 become a sacrifice make 41 take an appointment make 19 take a passport get 31 do a stop put 6 take a balance keep 22 do use make 3 do a goal make 4 receive trial stand 1 do a discovery make 4 do questions ask 1 do emphasis put 4 do an eye keep 1 take a stain remove 3 do a ballot cast 2 do a breakdown have 2 do counsel take 1

202

Literal type Abstract type receive an order accept 47 get a name have 42 give damage cause 41 take court pay 22 take a copy make 28 do liberties take 8 become cancer develop 26 become the sack get 5 do an order place 13 do corners cut 4 receive an operation undergo 13 do a bargain strike 4 do a walk take 11 do a dive take 4 do an argument have 9 do witness bare 1 do a bath take 3 do the knot tie 1 receive an exam take 1 TABLE 5-5. Number of transfer errors in each subtype.

A large part of L1 transfer seems to involve idiosyncratic errors; 55% of the erroneous

forms were produced by fewer than 10 subjects. But at the same time, some erroneous

combinations were produced by quite large numbers of subjects. In the language

classroom, it seems wise to give explicit instruction regarding the most troublesome

collocations. More than half of the subjects produced the 7 erroneous forms: become

victim (target = fall victim), make rice (grow rice), make a school (found a school),

become a sacrifice (make a sacrifice), receive an order (accept an order), give damage

(cause damage), and get a name (have a name).

A final remark about L1 transfer relates to restrictedness. Some collocations are

quite fixed and do not permit internal modifications, such as word replacement and

addition of another word within the collocation, while other collocations are closer to free

combinations and the words can be replaced with a wide range of other words with little

semantic change. For example, the NP a job clusters with the following eight verbs with

little semantic change: get, acquire, attain, find, obtain, be hired for, accept, and receive.

In contrast, the noun permission clusters with smaller number of verbs: give and grant to denote the same content. An increase in transfer was observed in proportion to the

203

number of candidates which can co-occur with the NP. Figure 5-34 shows the

percentages of transfer errors divided by the degree of restrictedness of collocations.

Figure 5-34. Percentages of transfer errors.

The results of this study do not support Nesselhauf’s claim (2003) that restricted

collocations are most susceptible to transfer. In my study the less restricted collocations

were more susceptible to transfer. It is possible that learners’ awareness of collocational

restrictions might lead them to overestimate the range of verbs occurring in less restricted collocations while underestimating the range of verbs which occur in more restricted

collocations. My study generated relatively small amounts of data (626 tokens of transfer

errors in total). Furthermore, most of the L1 transfer seems to be idiosyncratic; 55% of

the erroneous forms were produced by fewer than 10 of the 66 subjects. More research is

needed to better understand the relation between transfer and restrictedness of

collocations.

204

CHAPTER 6

The effects of Instruction in acquiring Verb+Noun collocations.

6.1. Purpose of the experiments, research questions and hypotheses

The second pair of experiments investigated the role of instruction in learning

Verb + Noun collocations. The expectation was that explicit instruction would bring about measurable improvement in learners’ performance and help L2 learners attain more native-like knowledge. Because Swain (1983) states in her output hypothesis that learners internalize better what they actually use or produce, the inclusion of production exercises (Long, 1985; Prabhu, 1987; Long and Crookes, 1992) as part of the instruction was designed to improve the learner’s ability to use L2 collocations. For these reasons, preceding Experiments 3 and 4, learners received explicit explanations regarding collocational usage, including the restrictions and differences between synonymous words; these were followed by production exercises.

The research questions are:

Question 1: Do instruction and exercises have a measurable effect in acquiring

English Verb+Noun collocations? Can these collocations be taught in the ESL

classroom?

Question 2: Does the learners’ performance correlate with the overall frequency of

target collocations?

Question 3: Is the learners’ performance directly influenced by the literalness of the

target collocations?

205

Question 4: Do learners perform better if there is a translational equivalent in L1?

Question 5: Are English collocations involving light verbs more difficult that these

involving content verbs?

Question 6: Is there a significant difference between the performance of low-

intermediate and advanced learners?

Specific hypotheses are:

Hypothesis for Question 1: Instruction will have a measurable role in acquisition.

There will be overall improvement in the learners’ performance after explicit

instruction.

Hypothesis for Question 2: Learners will demonstrate better productive knowledge of

frequent collocations as compared to less frequent ones.

Hypothesis for Question 3: Learners will demonstrate better productive knowledge of

collocations which have literal meanings (e.g., take time, have an idea, make a

mistake) as compared to those which have metaphorical meanings (e.g., take wing,

have the brains, make history).

Hypothesis for Question 4: Learners will perform better if there is a translational

equivalent in L1.

Hypothesis for Question 5: Learners will perform less well on light verbs than on

content verb.

206

Hypothesis for Question 6: Advanced learners will perform significantly better than

low-intermediate learners.

6.2. Method

6.2.1. Subjects

The subjects of this experiment were sixty of the sixty-six Japanese learners of

English who participated in Experiments 1 and 2. There were thirty subjects in each

proficiency group. Seven males and twenty-three females were in the low-intermediate

group, and six males and twenty-four females were in the advanced group. Their ages varied from 19 to 48 years, with the mean age being 22 years 9 months (23 years 5

months for low-intermediate subjects, and 23 years 9 months for advanced subjects). The

subjects have lived in English-speaking countries from 1 month to 22 years, with the

mean length of residence being 4 years 3 months (2 years 7 month for the low-

intermediate group, and 6 years 6 months for the advanced group). When I analyzed the

results from Experiments 1 and 2 in Chapter 4 and 5, I included data from all the sixty-six

subjects. In this Chapter, when analyzing results from the pretest and posttest I excluded

data of six learners who took the pretest but did not come back to take the posttest.

6.2.2. Stimuli and Procedure

Immediately after completing Part 1 (Experiments I and II), all subjects received

approximately forty minutes of instruction, individually or in small groups. The

instruction used a 32-page handout (attached in Appendix-X) which included a brief

introduction to collocations and a discussion of common mistakes with L2 collocations.

207

Then the eight light verbs which are dealt in this study were presented. Small differences

between English and Japanese (the actions denoted by the verbs, the types of objects which often co-occur with these verbs) were discussed along with plenty of examples.

For example, it was pointed out that the verb take can be used when “something is picked up in someone’s hand (example: you can take anything in your hand and have a close

look at it)”, but take cannot be used when “something is picked up in someone’s hand and

then given to somebody else (example: *Will you take me the mustard?)” although using

the corresponding Japanese verb toru (take) is appropriate in this context. Many

examples were given in the handout, including incorrect usages, which were highlighted

with an “incorrect” mark ×.

Another topic discussed during the instruction was degree of fixedness of collocations. Some collocations are more fixed than others. Some allow internal modifications by replacing the verb by another verb, adding adjectives and/or adverbs

without changing the general meaning, while other collocations allow very limited or no

internal modifications. Among a number of modifications, possibilities of replacing

verbs were mainly discussed in the instruction. For example, verbs in such collocations

as get a break, get the nod, become a sensation, take a stand, take the trouble, and take

one’s order cannot be replaced with another verb. In contrast, verbs in some collocations

can be replaced by a few different verbs: form / organize an association, get / attain the

position of President, get / obtain knowledge, give / confer a commission and so on. The

handout listed collocations with possible paraphrases, and added some information about

slight semantic or stylistic changes which may arise by replacing the verb.

The instruction concluded by providing students with a set of techniques and tools

208

which they might find useful in learning collocations, such as using online search engines,

making full use of the five senses (read, write, listen, act out, etc.), translating back from

English to Japanese, learning the etymology, learning related words together (synonyms,

antonyms, common paraphrases), learning similar collocations together, and using mnemonic aids.

Two kinds of take-home exercises were then given to each participant. The first exercise was a paper-based set of multiple choice questions with answer keys at the end

(A copy of the exercises is attached to Appendix-XI). The second exercise constituted a link to online flashcard exercises using WordChamp.com.

(http://www.wordchamp.com/lingua2/Home.do), a website developed by

GlobalLinguist. It provides an open database of user-generated language learning content that includes word pairs in over one hundred languages, self-paced exercises, online homework and drills for any language, resources for language teachers, and community networking tools to assist foreign language learning.

All the subjects were asked to study the collocations at home, using the paper- based exercises and/or the online flashcards along with the handout distributed during the instruction. The subjects used the materials of their choice, and spent as much or as little time as they wanted in studying the collocations at home.

Within two weeks after taking Part 1 (Experiments 1 and 2), each subject came back to take another fill-in-the-blank test. The same sixty-six collocations were tested, presented in different sentences in semantically natural contexts. Unlike in Part 1, the subjects were asked to give only one answer to each question, and there were no additional questions which asked about the subjects’ confidence in their answers and

209

familiarity of the phrases they produced. The subjects were allowed to spend as much

time as they wanted to complete these tasks; they spent between fifteen and forty minutes.

Before the posttest, subjects reported how much time they had spent studying

collocations using the materials provided (A copy of the questionnaire is attached to

Appendix-XII). A brief analysis of this data will be presented in 6.5.

All the combinations produced by the subjects were given to the same six native

speakers of English who had graded the combinations from Experiments 1 and 2. They

graded each combination using the same criteria: two 6-point-scales: (1) Meaning: How

well was the intended meaning conveyed, and (2) Acceptability: Would you produce this

phrase in the same context?

Experiment 3 is discussed in section 6.3, and Experiment 4 in section 6.4. Then

the study data for Experiments 3 and 4 are discussed in section 6.5. A general discussion

is in 6.6.

6.3. Experiment 3 (The post test for Experiment 1): The role of frequency in acquiring Verb + Noun collocations

Experiment 3 served as the posttest for Experiment 1. The same thirty-six

collocations tested in Experiment 1 were tested again (See Appendix-XIII for a list of test

sentences of Experiment 3). In Experiment 1, it was found that frequency as well as

other variables (existence of L1 equivalents, light verbs vs. content verbs, and same light verbs or different light verbs) influence learners’ performance. In Experiment 3, we examine efficacy of instruction and how these variables influence learners’ performance.

First I discuss the general results and compare the four variables.

210

6.3.1. Results for Experiment 3

The data from Experiment 3 is attached in Appendix-XIV. Table XV-1 (in

Appendix XV) and Figures 6-1, 6-2, and 6-3 show the percentages of accurate collocations for each collocation type produced by the two proficiency groups in

Experiments 1 (pretest) and 3 (posttest), and Table 6-A and Figure 6-A compare the percentages of accurate production for all these types of collocations. Collocations were considered to be accurate when native speakers’ average scores for both Grammaticality

Judgments and Meaning Conveyed (6-point-scale) were 5.0 or higher. To analyze learners’ improvement, for each item I compared each learner’s most confident answer in the pretest with their answer in the posttest.

100% 91.11%89.17% 86.67% 87.22% 90% 83.89% 85.28%

80% 75.00% 68.33% 70% Low- 61.67% intermediate 60% 51.11% 52.22% Advanced 50% 53.33%

40% Total 30%

20%

10%

0% pretest posttest pretest posttest

A1 Frequent A1 Infrequent

Figure 6-1. Percentages of accurate collocations of types A1.

211

100% 86.11% 90% 77.22% 80% 71.67% 71.67% 68.33% 70% Low- 59.17% 60.28% 60% intermediate 51.11% 48.89% Advanced 50% 46.67% 41.67% Total 40% 32.22%

30% 20%

10%

0% pretest posttest pretest posttest

A2 Frequent A2 Infrequent

Figure 6-2. Percentages of accurate collocations of types A2.

100%

90% 76.67% 80% 68.33% 66.94% 70% 60.00% 57.22% 60% 51.67% 44.44% 46.11% Low-intermediate 50% 38.89% 38.05% Advanced 40% 31.67% 31.67% Total 30%

20%

10% 0% pretest posttest pretest posttest

A3 Frequent A3 Infrequent

Figure 6-3. Percentages of accurate collocations of types A3.

212

Pretest Posttest Improve Ex. 1 Ex. 3 ment Low-intermediate 503 675 (# = 30) 46.58% 62.50% 15.93% Advanced learner 632 844 (# = 30) 58.52% 78.15% 19.63% Total 1135 1519 (# = 60) 52.55% 70.33% 17.78% TABLE 6-A. Summary of tokens and percentages of accurate collocations in Experiments 1 and 3.

78.15% 80% 70.33% 70% 62.50% 58.52% 60% 52.55% 50% 46.58% Low-intermediate 40% Advanced 30% Total 20%

10%

0% pretest posttest

Figure 6-A. Summary of percentages of accurate collocations in Experiments 1 and 3.

There was a higher percentage of accurate collocations in the posttest than in the pretest for all the types of collocations. That is, the learners did improve in their performance after instruction. The increase in the percentages of accurate collocations observed in the overall performance between Experiments 1 and 3 was statistically significant for both proficiency groups (Low-intermediate: χ2 = 24.82; df = 1; p* < .0001; Advanced: χ2 =

30.16; df = 1; p* <.0001; The total: χ2 = 55.28; df = 1; p* < .0001). There was no significant difference in how much low-intermediate learners improved compared with advanced learners (χ2 = 0.004; df = 1; p* = 0.950). In other words, both of the

213

proficiency groups attained similar rates of improvement.

Next, we examine how the four variables ((1) Frequency; (2) Existence of L1

equivalents; (3) Light vs. content verbs; and (4) Same vs. Different light verbs) affected

the improvement in the learners’ performance.

First, I tested frequency. Table 6-B and Figure 6-B compare the numbers of

tokens and percentages of accurate frequent vs. infrequent combinations produced by

learners in Experiments 1 and 3.

Frequent Infrequent

Pretest Posttest Improve Pretest Posttest Improve Ex. 1 Ex. 3 ment Ex. 1 Ex. 3 ment Low-intermediate 292 373 211 302 (# = 30) 54.08% 69.07% 14.99% 39.07% 55.93% 16.86% Advanced learner 365 442 267 402 (# = 30) 67.59% 81.85% 14.26% 49.44% 74.45% 25.01% Total 657 815 478 704 (# = 60) 60.84% 75.46% 14.62% 44.26% 65.19% 20.93% TABLE 6-B. Tokens and percentages of accurate collocations in Experiments 1 and 3. Frequent vs. infrequent types.

100% 90% 81.85% 80% 74.45% 69.07% 75.46% 70% 67.59% 65.19% 60% 54.08% 60.84% 55.93% 50% 49.44% Low- intermediate 40% 44.26% Advanced 39.07% 30% Total 20% 10% 0% pretest posttest pretest posttest

Frequent Infrequent

Figure 6-B. Percentages of accurate collocations in Experiments 1 and 3. Frequent vs. infrequent types.

214

The subjects improved on average by 14.62% for the frequent types and by

20.93% for the infrequent types. Both proficiency groups produced significantly more

accurate combinations for both frequent and infrequent types in the posttest (Low-

intermediate, frequent types: χ2 = 9.62; df = 1; p* = 0.002; infrequent types: χ2 = 15.78; df

= 1; p* <.0001; Advanced, frequent types: χ2 = 7.16; df = 1; p* = 0.0075; infrequent

types: χ2 = 26.84; df = 1; p* <.0001; Total, frequent types: χ2 = 16.74; df = 1; p* <.0001;

infrequent types: χ2 = 42.82; df = 1; p <.0001.

There was no significant difference in the amount of improvement between the low-intermediate and advanced groups for either frequent or infrequent types (frequent

types: χ2 = 0.257; df = 1; p = 0.6122; infrequent types: χ2 = 0.18; df = 1; p = 0.6714.

Overall, the learners made significantly more improvement with infrequent types than with frequent types (χ2 = 4.709; df = 1; p* = 0.0300) although they still produced greater numbers of accurate collocations for the frequent collocations in the posttest. A significant increase in accurate collocations for frequent types as compared to infrequent types was observed in the advanced group (χ2 = 4.226; df = 1; p* = 0.0398), but no

significant difference was found in the low-intermediate group (χ2 = 0.914; df = 1; p =

0.3391).

Second, I compared existence of L1 equivalents. Tokens and percentages of

accurate collocations with vs. without L1 equivalents in the pretest and the posttest are

displayed in Table 6-C and Figure 6-C.

215

With L1 equivalent Without L1 equivalent

Pretest Posttest Improve Pretest Posttest Improve Ex. 1 Ex. 3 ment Ex. 1 Ex. 3 ment Low-intermediate 247 268 256 407 (# = 30) 68.61% 74.45% 5.84% 35.56% 56.53% 20.97% Advanced learner 248 299 384 545 (# = 30) 68.89% 83.06% 14.17% 53.33% 75.70% 22.37% Total 495 567 640 952 (# = 60) 68.75% 78.76% 10.01% 44.45% 66.12% 21.67% TABLE 6-C. Tokens and percentages of accurate collocations in Experiments 1 and 3. With vs. without L1 equivalents.

100% 90% 83.06% 78.76% 80% 68.89% 74.45% 75.70% 68.75% 68.61% 66.12% 70% 53.33% 56.53% 60% Low-intermediate 44.45% 50% Advanced 35.67% 40% Total 30% 20% 10% 0% pretest posttest pretest posttest

With L1 equivalent Without L1 equivalent

Figure 6-C. Percentages of accurate collocations in Experiments 1 and 3. With vs. without L1 equivalents.

The percentages of accurate combinations with vs. without L1 equivalents were significantly higher in the posttest among the advanced learners (With L1 equivalents: χ2

= 4.58; df = 1; p* = 0.0323; Without L1 equivalents: χ2 = 27.56; df = 1; p* <.0001).

Among the low-intermediate learners, the difference was significant for the collocation types without L1 equivalents (χ2 = 33.94; df = 1; p* <.0001), but not for types with L1

equivalents (χ2 = 0.78; df = 1; p = 0.3771). Overall, a significantly higher percentage of

216

collocations was regarded as accurate in the posttest in both types across all the subjects

(With L1 equivalents: χ2 = 4.74; df = 1; p* = 0.0295; Without L1 equivalents: χ2 = 60.76; df = 1; p* <.0001).

There was no significant difference in the amount of improvement between the two proficiency groups for either type (With L1 equivalent: χ2 = 0.733; df = 1; p =

0.3919; Without L1 equivalents: χ2 = 1.193; df = 1; p = 0.2747).

Pooling all the subjects, the increase in percentages of accurate combinations was

greater for the types without L1 equivalents than for those with L1 equivalents (χ2 =

10.691; df = 1; p* = 0.0011). This trend was observed in the advanced learners’ results

(χ2 = 2.254; df = 1; p = 0.1333) but not for the low-intermediate learners (χ2 = 10.354; df

= 1; p* = 0.0013). In spite of greater improvements in collocations with no L1

equivalents, learners nevertheless produced a greater number of accurate collocations with L1 equivalents than without L1 equivalents.

The third analysis looks at content verbs vs. light verbs. Table 6-D and Figure 6-

D compare low-intermediate and advanced learners’ percentages of accurate collocations

in the pretest and the posttest.

Light verb Content verb

Pretest Posttest Improve Pretest Posttest Improve Ex. 1 Ex. 3 ment Ex. 1 Ex. 3 ment Low-intermediate 389 479 114 196 (# = 30) 54.03% 66.53% 12.50% 31.67% 54.45% 22.78% Advanced learner 469 583 163 261 (# = 30) 65.14% 80.97% 15.83% 45.28% 72.50% 27.22% Total 858 1062 277 457 (# = 60) 59.59% 73.75% 14.17% 38.48% 63.48% 25.00% TABLE 6-D. Tokens and percentages of accurate collocations in Experiments 1 and 3. Light verbs vs. content verbs.

217

Low- 100% intermediate 90% Advanc ed 80.97% 80% 72.50% Total 66.53% 73.75% 70% 65.14% 60% 54.03% 54.45% 59.59% 63.48% 50% 45.28% 38.48% 40% 31.67% 30% 20% 10% 0% pretest posttest pretest posttest

Light verbs Content verbs

Figure 6-D. Percentages of accurate collocations in Experiments 1 and 3. Light verbs vs. content verbs.

Both proficiency groups produced a significantly greater number of accurate collocations in the posttest for both light verb and content verb collocations (Low-intermediate, light verbs: χ2 = 9.12; df = 1; p* = 0.0025; content verbs: χ2 = 21.16; df = 1; p* <.0001;

Advanced, light verbs: χ2 = 12.14; df = 1; p* = 0.0005; content verbs: χ2 = 22.2; df = 1;

p* <.0001; Total, light verbs: χ2 = 21.46; df = 1; p* <.0001; content verbs: χ2 = 43.66; df

= 1; p* <.0001).

No statistically significant difference was observed in the ratio of accurate

collocations between the pretest and the posttest among the low-intermediate vs. the

advanced groups (Light verbs: χ2 = 0.011; df = 1; p = 0.9165; Content verbs: χ2 = 0.212;

df = 1; p = 0.6452).

More significant improvement was found in content verb collocations than in

light verb collocations in both proficiency groups (Low-intermediate: χ2 = 6.037; df = 1;

p* = 0.0140; Advanced: χ2 = 4.651; df = 1; p* = 0.0310; Total: χ2 = 10.476; df = 1; p* =

218

0.0012). In the posttest, productions were more accurate for the light verb collocations

than for content verb collocations.

Finally, we compare performance with same vs. different light verbs. Table 6-E

and Figure 6-E show the average percentages of accurate collocations involving same vs.

different light verbs produced by the low-intermediate and the advanced learner groups.

Same light verb Different light verb

Pretest Posttest Improve Pretest Posttest Improve Ex. 1 Ex. 3 ment Ex. 1 Ex. 3 ment Low-intermediate 247 268 142 211 (# = 30) 68.61% 74.45% 5.84% 39.45% 58.61% 19.16% Advanced learner 248 299 221 284 (# = 30) 68.89% 83.06% 14.17% 61.39% 78.89% 17.50% Total 495 567 363 495 (# = 60) 68.75% 78.76% 10.01% 50.42% 68.75% 18.33% TABLE 6-E. Tokens and percentages of accurate collocations in Experiments 1 and 3. Same light verbs vs. different light verbs.

100% 90% 83.06% 78.89% 80% 68.89% 74.45% 78.76% 68.61% 68.75% 70% 68.75% 61.39% 58.61% 60% 50% 50.42% Low-intermediate 39.45% 40% Advanced 30% Total 20% 10% 0% pretest posttest pretest posttest

Same light verbs Different light verbs

Figure 6-E. Percentages of accurate collocations in Experiments 1 and 3. Same light verbs vs. different light verbs.

The average for the posttest was significantly higher for both same light verbs and

219

different light verb types (Same light verbs: χ2 = 4.74; df = 1; p* = 0.0295; Different light

verbs: χ2 = 20; df = 1; p* < .0001). This result was observed in the advanced learners’

group (Same light verbs: χ2 = 4.58; df = 1; p* = 0.0323; Different light verbs: χ2 = 7.62; df = 1; p* = 0.0058). Among the low-intermediate group, the posttest results for different

light verb type had significantly more accurate collocations (χ2 = 13.1; df = 1; p* =

0.0003), while the difference was not significant for the same light verb type (χ2 = 0.78;

df = 1; p = 0.3771).

In other words, advanced learners improved significantly more than did low-

intermediate learners in the production of collocations involving same light verbs in the

posttest (χ2 = 0.733; df = 1; p = 0.3919). In contrast, low-intermediate learners made significantly more improvement in the production of different light verb collocations than

did advanced learners (χ2 = 1.064; df = 1; p = 0.3023).

Overall, there was no significant difference in the amount of increase in the

posttest between same and different light verb types (χ2 = 3.554; df = 1; p = 0.0594).

Advanced learners’ production showed a greater percentage of improvement in the different light verb type (χ2 = 0.264, df = 1; p = 0.6074). No such significant difference was observed for the low-intermediate learners’ (χ2 = 5.066; df = 1; p* = 0.0244).

However, the accuracy rate was still higher in the same light verb types than different verb types in the posttest.

To summarize, in the posttest, learners produced greater numbers of accurate collocations for all types of collocations. The differences were significant for all types except for low-intermediate learners’ collocations of two types: with L1 equivalents and same light verbs. There was no significant difference in the increase in accurate

220

collocations across the two proficiency groups overall. In general, more improvement

was found for the more difficult types of collocations (1. Infrequent; 2. Without L1

equivalent; 3. Light verbs; and 4. Different light verbs) than for the easier types. The

differences in the rate of improvement were significant in all the types except the

following: low-intermediate learners’ production of frequent vs. infrequent collocations,

advanced learners’ production of collocations with vs. without L1 equivalents, same vs.

different light verbs, and the overall performance in collocations involving same vs. light verbs. Table 6-1 below summarizes the subtypes where significantly greater

improvement was observed for more difficult types.

Low- Advanced Total intermediate 1. Frequency Infrequent type - √ √ 2. Existence of L1 equivalent Without L1 equivalent √ - √

3. Light vs. content verbs Content verb √ √ √

4. Same vs. different light V Different light verb √ - -

TABLE 6-1. Subtypes where significantly greater improvement was observed for more difficult types: Experiment 3.

Although greater improvement was observed in more difficult types of collocations, the

percentages of accurate collocations of all the difficult types are still lower than those for

easier types of collocations.

6.4. Experiment 4 (Post test for Experiment 2: The role of meaning in acquiring Verb + Noun collocations

Experiment 4 served as a posttest for Experiment 2. The same thirty collocations

tested in Experiment 2 were tested again (See Appendix-XVI for all the test sentences of

221

Experiment 4). As Experiment 3, we consider the efficacy of instruction and how the

four variables may have affected learners’ performance. First I discuss the overall results,

then I examine each of the four variables.

6.4.1. Results for Experiment 4

As in the previous section, the percentages of accurate collocations for each

collocation type in Experiments 2 (pretest) and 4 (posttest) are presented in Table XVIII-

1 (in Appendix XVIII) and Figures 6-4 and 6-5. Table 6-F and Figure 6-F display

percentages of accurate collocations of all the types of collocations. See Appendix XVII

for all the combination tested in this experiment. The criteria for accurate collocations

was the same as for Experiment 3; Learners’ collocations were regarded as accurate when native speakers’ average scores of both Grammaticality Judgments and Meaning

Conveyed (6-point-scale) were 5.0 or higher. As in Experiment 3, for each item I

compared each learner’s most confident answer in the pretest with their answer in the

posttest.

222

92.67% 100% 86.00% 89.33% 91.00% 90% 82.00% 74.00% 78.00% 77.33%72.67% 80% 73.67% 68.00% 73.33% 70% 60% Low-intermediate 50% Advanced 40% Total 30% 20% 10% 0% pretest posttest pretest posttest

A1 Literal A1 Abstract

Figure 6-4. Percentages of accurate collocations of types A1.

100% 89.33% 87.33% 85.33% 90% 82.67% 83.33% 80.00% 80% 68.00% 70% 59.34% 60% 50.67% Low-intermediate 50% Advanced 40% Total 24.00% 30% 21.67% 19.33% 20%

10%

0% pretest posttest pretest posttest

A2 Literal A2 Abstract

Figure 6-5. Percentages of accurate collocations of types A2.

223

100%

90%

80% 64.67% 70% 58.00% 60% 55.67% 51.67% 50% 46.67% 45.33% 35.33% 32.67% Low-intermediate 40% 31.00% 28.34% 26.67% Advanced 30% 24.00% Total 20%

10%

0% pretest posttest pretest posttest

A3 Literal A3 Abstract

Figure 6-6. Percentages of accurate collocations of types A3.

Pretest Posttest Improve ment Low-intermediate 444 583 (# = 30) 49.34% 64.78% 15.44% Advanced learner 517 667 (# = 30) 57.44% 74.12% 16.68% Total 961 1250 (# = 60) 53.39% 69.45% 16.06% TABLE 6-F. The summary of tokens and percentages of accurate collocations in Experiments 2 and 4.

224

80% 74.12% 69.45% 70% 64.78% 57.44% 60% 53.39% 49.34% 50% Low-intermediate 40% Advanced Total 30%

20%

10%

0% pretest posttest

Figure 6-F. Summary of percentages of accurate collocations in Experiments 1 and 3.

Higher percentages of accurate collocations were observed in the posttest across all the collocation types. The learners did improve after instruction and study at home. The overall increase in percentages of accurate productions from Experiment 2 to 4 was statistically significant for both proficiency groups (Low-intermediate: χ2 = 18.54; df = 1;

p* <.0001; Advanced: χ2 = 18.76; df = 1; p* <.0001; The total: χ2 = 37.52; df = 1; p*

<.0001).

Both groups made similar amounts of improvement; no significant difference was observed (χ2 = 0.042; df = 1; p = 0.8376).

Next, we examine the effects of each of the four variables: (1) Literal vs. abstract meaning; (2) The existence of L1 equivalents; (3) Light vs. content verbs; and (4) Same vs. Different light verbs.

The first variable is literalness of meaning. Table 6-G and Figure 6-G show the numbers of tokens and percentages of accurate literal vs. abstract collocations produced by learners in Experiments 2 and 4.

225

Literal Abstract

Pretest Posttest Improve Pretest Posttest Improve Ex. 2 Ex. 4 ment Ex. 2 Ex. 4 ment Low-intermediate 273 329 171 254 (# = 30) 60.67% 73.11% 12.44% 38.00% 56.44% 18.44% Advanced learner 312 367 205 300 (# = 30) 69.33% 81.56% 12.23% 45.55% 66.67% 21.12% Total 585 696 376 554 (# = 60) 65.00% 77.34% 12.34% 41.78% 61.56% 19.78% TABLE 6-G. Tokens and percentages of accurate productions in Experiments 2 and 4. Literal vs. abstract.

100% 90% 81.56% 77.34% 80% 69.33% 73.11% 70% 65.00% 66.67% 60.67% 61.56% 60% 56.44% 45.55% Low-intermediate 50% 38.00% 41.78% Advanced 40% Total 30% 20%

10% 0% pretest posttest pretest posttest

Literal Abstract

Figure 6-G. Percentages of accurate productions in Experiments 2 and 4. Literal vs. abstract.

Both proficiency groups attained higher percentages for the posttest on both literal and abstract collocations. The differences in the percentages of accurate collocations between the pretest and the posttest were significant (Low-intermediate, literal: χ2 = 5.02; df = 1;

p* = 0.0251; abstract: χ2 = 15.82; df = 1; p* <.0001; Advanced, literal: χ2 = 4.3; df = 1; *p

226

= 0.0381; abstract: χ2 = 17.5; df = 1; p* <.0001; Total, literal: χ2 = 9.44; df = 1; p* =

0.0021; abstract: χ2 = 33.68; df = 1; p* <.0001).

There was no significant difference between low-intermediate and advanced

learners in the amount of increase in the percentage of accurate collocations in the

posttest (Literal: χ2 = 0.046; df = 1; p = 0.8302; Abstract: χ2 = 0.012; df = 1; p = 0.9128).

Learners achieved greater improvement in their performance on abstract than on literal collocations. This difference was significant for results involving all the subjects

(χ2 = 6.015; df = 1; p* = 0.01418478). But the difference was not significant in the

separate proficiency groups (Low-intermediate: χ2 = 2.654; df = 1; p = 0.10323;

Advanced: χ2 = 3.377; df = 1; p = 0.0661. In spite of greater improvement in their

performance on abstract collocations, the accurate percentages for the posttest remained

higher for literal collocations.

The second analysis compares performance on collocations with vs. without L1

equivalents. Table 6-H and Figure 6-H display the percentages of accurate collocations produced by each proficiency group in their performance on these subsets of collocations.

With L1equivalent Without L1 equivalent

Pretest Posttest Improve Pretest Posttest Improve Ex. 2 Ex. 4 ment Ex. 2 Ex. 4 ment Low-intermediate 219 244 225 339 (# = 30) 73.00% 81.33% 8.33% 37.50% 56.50% 19.00% Advanced learner 245 250 272 417 (# = 30) 81.67% 83.34% 1.67% 45.33% 69.50% 24.17% Total 464 494 497 756 (# = 60) 77.34% 82.34% 5.00% 41.42% 63.00% 21.58% TABLE 6-H. Tokens and percentages of accurate collocations in Experiments 2 and 4: with vs. without L1 equivalents.

227

100% 83.34% 90% 81.67% 81.33% 82.34% 77.34% 80% 73.00% 69.50% 70% 63.00% 60% 56.50% 45.33% 50% 41.42% 37.50% Low-intermediate 40% Advanced 30% Total 20%

10% 0% pretest posttest pretest posttest

With L1 equivalent Without L1 equivalent

Figure 6-H. Percentages of accurate collocations in Experiments 2 and 4: with vs. without L1 equivalents.

There was no statistically difference between the pretest and the posttest in the percentages of accurate collocations with L1 equivalents (Low-intermediate: χ2 = 1.24; df

= 1; p = 0.2655; Advanced: χ2 = 0.04; df = 1; p = 0.8415; Total: χ2 = 0.88; df = 1; p =

0.3482). In contrast, the difference in learners’ accuracy on collocations without L1 equivalents was statistically significant in both proficiency groups (Low-intermediate: χ2

= 22.64; df = 1; p* <.0001; Advanced: χ2 = 30.1; df = 1; p* <.0001; Total: χ2 = 53.12; df

= 1; p* <.0001).

No significant difference was observed between low-intermediate and advanced

learners’ in the increase of accurate collocations (With L1 equivalents: χ2 = 0.461; df = 1;

p = 0.4972; Without L1 equivalents: χ2 = 0.022; df = 1; p = 0.8820).

Both proficiency groups achieved significantly greater improvement in accuracy

on their performance with collocations without L1 equivalents than for those with L1

equivalents (Low-intermediate: χ2 = 5.684; df = 1; p* = 0.01712; Advanced: χ2 = 11.751;

228

df = 1; p* = 0.00061; Total: χ2 = 16.991; df = 1; p* < .0000). Although the learners

improved more for collocations without L1 equivalents, they still produced higher

percentages of accurate collocations with L1 equivalents.

In my third analysis, I tested success with light verbs vs. content verbs. Table 6-I

and Figure 6-I show percentages of accurate collocations involving light vs. content verbs

in Experiments 2 and 4.

Light verb Content verb

Pretest Posttest Improve Pretest Posttest Improve Ex. 2 Ex. 4 ment Ex. 2 Ex. 4 ment Low-intermediate 368 445 76 138 (# = 30) 61.33% 74.17% 12.84% 25.34% 46.00% 20.66% Advanced learner 415 483 102 184 (# = 30) 69.17% 80.50% 11.33% 34.00% 61.34% 27.34% Total 783 928 178 322 (# = 60) 65.25% 77.34% 12.09% 29.67% 53.67% 24.00% TABLE 6-I. Tokens and percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs.

100% 90% 80.50% 77.34% 80% 69.17% 74.17% 65.25% 70% 61.33% 61.34% 60% 53.67% 50% 46.00% 34.00% Low-intermediate 40% 29.67% 25.34% Advanced 30% Total 20% 10% 0% pretest posttest pretest posttest

Light verbs Content verbs

Figure 6-I. Percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs.

229

Overall the learners attained significantly higher scores for the posttest for both

collocation types (light verbs vs. content verbs) (Total, light verbs: χ2 = 13.78; df = 1; p*

= 0.0002; content verbs: χ2 = 40.9; df = 1; p* <.0001). This result was obtained in most

subsets of their performance by both proficiency groups (Low-intermediate, content

verbs: χ2 = 17.38; df = 1; p* <.0001; Advanced, light verbs: χ2 = 5; df = 1; p* = 0.0253;

content verbs: χ2 = 22.94; df = 1; p* <.0001). However, for low-intermediate learners’

performance on the posttest, the increase in the percentage of accurate collocations was

not significant (χ2 = 2.7; df = 1; p = 0.1003)

There was no statistical difference in the increase of accuracy between the lower

and the advanced learners (Light verbs: χ2 = 0.155; df = 1; p = 0.6938; Content verbs: χ2

= 0.001; df = 1; p = 0.9748).

Both proficiency groups improved significantly more in their performance on content verb collocations than on light verb collocations (Low-intermediate: χ2 = 6.563; df = 1; p* = 0.0104; Advanced: χ2 = 9.814; df = 1; p* = 0.0017; Total: χ2 = 16.263; df = 1;

p* < .0000). Again, in spite of a notable increase in their performance on content verb

collocations, the observed results show even greater accuracy in their performance on

light verb collocations.

The fourth analysis compares performance on collocations with the same vs.

different light verbs. Table 6-J and Figure 6-J compare percentages of accurate

collocations in these subtypes.

230

Same light verb Different light verb

Pretest Posttest Improve Pretest Posttest Improve Ex. 2 Ex. 4 ment Ex. 2 Ex. 4 ment Low-intermediate 219 244 50 67 (# = 30) 73.00% 81.33% 8.33% 49.67% 67.00% 17.33% Advanced learner 245 250 57 78 (# = 30) 81.67% 83.34% 1.67% 56.67% 77.67% 21.00% Total 464 494 107 72 (# = 60) 77.34% 82.34% 5.00% 53.17% 72.34% 19.17% Figure 6-J. Tokens and percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs.

100% 90% 83.34% 81.67% 81.33% 82.34% 77.34% 77.67% 80% 73.00% 72.34% 70% 67.00% 56.67% 60% 49.67% 53.17% 50% Low-intermediate Advanced 40% Total 30% 20% 10% 0% pretest posttest pretest posttest

Same light verbs Dif ferent light verbs

Figure 6-J. Percentages of accurate collocations in Experiments 2 and 4. Light verbs vs. content verbs.

Overall, there was no significant difference in the increase of accurate combinations between the pretest and the posttest (Low-intermediate, same light verbs: χ2 = 1.24; df =

1; p = 0.2655; different light verbs: χ2 = 2.18; df = 1; p = 0.1398; Advanced, same light

verbs: χ2 = 0.04; df = 1; p = 0.8415; different light verbs: χ2 = 2.96; df = 1; p = 0.0853).

Pooling all the subjects, no significant difference was found in the rate of improvement in

the same light verb types (χ2 = 0.88; df = 1; p = 0.3482), while the improvement was

231

significant for the different light verb types (χ2 = 6.46; df = 1; p* = 0.011).

No significant difference was observed between the low-intermediate and the

advanced groups in the rate of increase in the accurate percentages in the posttest for both

types of combinations (Same light verbs: χ2 = 0.461; df = 1; p = 0.4972; Different light

verbs: χ2 = 0.007; df = 1; p = 0.9333).

When all the subjects’ data are analyzed together, improvement for the

performance on the different light verb type was statistically greater than for the same

light verb type (χ2 = 7.761; df = 1; p* = 0.0053). But the difference was not significant

when each proficiency group was analyzed separately, probably because the total number of tokens in this experiment was not large (Low-intermediate: χ2 = 0.783; df = 1; p =

0.3762; Advanced: χ2 = 2.248; df = 1; p = 0.1338). As we have seen for the other variables, learners’ accuracy in the posttest for the combinations involving same light verbs remained higher than for the combinations involving different light verbs although

they made greater improvement with different light verb types.

In summary, posttest improvements were observed for all the types of collocations.

The improvement was statistically significant in many of the subtypes except (1)

Collocations with L1 equivalents (both proficiency groups); (2) Light verbs types (low-

intermediate); (3) Same light verbs (both proficiency groups); and (4) Different light

verbs (both proficiency groups).

There was no significant difference in the amount of increase in accurate

collocations between the two proficiency groups for each of the subtypes. More

improvement was generally found in more difficult types of collocations (1. Abstract; 2.

Without L1 equivalent; 3. Light verbs; and 4. Different light verbs) than for easier types.

232

The differences in the amount of improvement were significant for all collocation types except low-intermediate learners’ production of literal vs. abstract collocations, and both low-intermediate and advanced learners’ production on collocations involving same light verbs vs. different light verbs. Although greater improvement was observed in more difficult types of collocations, the percentages of accurate collocations for all the difficult types remained lower than for the easier types of collocations. Table 6-2 summarizes the subtypes where significantly greater improvement was observed for more difficult types.

Low- Advanced Total intermediate 1. Literal vs. abstract Abstract type - - √ 2. Existence of L1 equivalent Without L1 equivalent √ √ √

3. Light vs. content verbs Content verb √ √ √

4. Same vs. different light V Different light verb - - √

TABLE 6-2. Subtypes where significantly greater improvement was observed for more difficult types: Experiment 4.

These results were quite similar to the results obtained in Experiments 1 and 3.

In the next section, we examine the study habits preceding the posttests

(Experiments 3 and 4) together.

6.5. Studying for the posttests

6.5.1. Materials learners used, amount of time they studied and their improvement

The learners spent between 10 minutes and 5 hours (average = 1 hour 24 min) studying collocations at home during the two weeks between the testing. Table 6-3 presents numbers of learners who did each exercise (reading the instruction handout, doing paper-based exercises and online flashcard exercises) and the average amount of

233

time spent for each exercise.

Advanced (30) Total (60) Low-intermediate (30) Type of # of Average # of Average # of Average exercises people Time people time people time Handout 30 43.00 min 28 39.33 min 58 41.17 min Paper Exercise 29 37.00 min 29 35.67 min 58 36.34 min Flashcards 6 4.17 min 9 8.83 min 15 6.5 min Average time 84.17 min 83.83 min 84.00 min TABLE 6-3. Numbers of learners who did each exercise and average amount of time spent for each exercise.

58 of 60 learners read the handout, 58 learners did the paper-based exercises, and 15 did

the online flashcard exercises. Fourteen of the learners who did the flashcard exercises

also did paper-based exercises and read the handout. One learner did only flashcard

exercises and read the handout. One learner read the handout and didn’t do any exercises.

One learner who didn’t read the handout did both paper-based and flashcard exercises.

The other learner who didn’t read the handout did only the paper-based exercise.

Figure 6-7 displays the percentages of accurate collocations in the pretest

(Experiments 1 and 2) and the posttest (Experiments 3 and 4).

234

80% 76.32% 69.93% 70% 63.53% 58.06% 60% 52.95% low-intermediate 47.83% 50% advanced

40% total

30%

20% pretest posttest

Figure 6-7. Percentages of accurate collocations in the pretest and the posttest.

As we saw in sections 6.3.1 and 6.4.1, both proficiency groups made a significant improvement in the posttest (χ2 = 240.69, df = 1, p* = 0), although there were some individual differences; learners improved between 53.33% and -10.56% (average improvement = 17.87%).

Figure 6-8 shows each learner’s improvement ordered by the amount of time they spent studying.

235

45%

30%

15% I m p r o v e m e n t n e m e o v r p m I

0% 90min 65min 50min 35min 300min 240min 180min 150min 120min 100min -15% Amount of time learners studied collocations

Figure 6-8. Learners’ improvement vs. amount of time they studied.

There was no observable relation between learners’ improvement and amount of time they studied. However, learners who used a wider range of materials in studying tended to make greater improvement. Figure 6-9 shows percentages of accurate collocations in the pretest and the posttest produced by learners who used one (n = 2), two (n = 45) or three (n = 13) kinds of materials.

80% 77.22% 68.10% 70% 62.09%

60% 54.17% 47.92% 47.54% 50% pretest

40% posttest

30%

20% 1 material (n=2) 2 materials (n=45) 3 materials (n=13)

Figure 6-9. Percentages of accurate collocations by learners who used 1, 2 or 3 materials.

236

The learners who used 1 or 2 materials improved 14.17% in the posttest, whereas the learners who used all 3 materials improved 29.68%. However, learners’ improvements may also have been influenced by other factors, such as motivation, amount of time spent, and so forth. Indeed the learners who used more materials usually spent more time studying than did those who used fewer materials. The learners who used only 1 material studied on average of 40 minutes, those who used 2 materials averaged 75.11 minutes, and those who used 3 materials averaged 113 minutes.

6.6. Other factors

In this section, we examine other factors related to learners’ improvement in the posttest. These are (1) ages of the subjects; (2) the subjects’ ratings of usefulness of the instructional materials prior to the posttest; and (3) the percentage of use of English in the subjects’ daily life.

The first factor is the age of the subjects. In this study, older subjects’ performance improved more than that of younger subjects. The subjects’ ages varied from 19 to 48 years, with the mean age being 23 years 7 months. Figure 6-10 displays learner’s improvement in the posttest ordered by age groups:

237

Figure 6-10. Learners’ improvement vs. age.

As the subjects’ age increased, there was more improvement. A possible explanation for

this result is that the procedure and the materials of my study were more helpful to the

subjects who had more years in school. Such things as fill-in-the-blank tests, lecture- style instructions, handouts which summarize main points, exercises based on multiple choice questions are very frequently used in many university classes. Most of the subjects in their late teens and early twenties were either undergraduate students or students studying English at the HELP or the NICE program, while the majority of the subjects in their late twenties or older were graduate students. The range of ages of the subjects in this study was relatively narrow. It would be interesting to collect data from a wider range of age groups to further examine the relationship between the learners’ improvement and their age.

Second, it was found that the subjects who rated the instruction materials (the handout, the exercises, and the flashcards) more highly tended to improve more. Before the posttest, the subjects were asked to rate whether each of the three materials seamed

238 useful to them, using a 4-point scale (4: very useful; 3: useful; 2: not very useful; 1: not useful at all). The average improvements in the posttest arranged by the overall ratings of the materials are given in Figure 6-11.

Figure 6-11. Evaluations of materials (4-point scale) vs. improvements.

Overall, all the subjects rated the materials quite highly. Almost all the subjects (58 subjects out of 60) rated the materials as either 4 (very useful) or 3 (useful); only two subjects rated the materials as 2 (not very useful); one subject rated 2 for the handout, and the other subject for the flashcards. The subjects’ overall average was 3.53 (Low- intermediate: 3.48; and Advanced: 3.58). Their scores for each of the materials were: the handout 3.53 (Low-intermediate: 3.50; Advanced: 3.57); the exercises 3.55 (Low- intermediate: 3.48; Advanced: 3.62); the flashcards 3.36 (Low-intermediate: 3.00;

Advanced: 3.56).

We have seen that the older subjects and the subjects who rated the materials more highly made greater improvement in the posttest. However, there was no clear relation between the age of the subjects and their ratings of the materials. Figure 6-12

239 displays the average ratings of the materials divided by the subjects’ age groups.

Figure 6-12. Subjects’ age and average ratings of the materials.

The subjects who were 30 years and older tended to rate the materials more highly than did other subjcts, but there was no observable relation among the other three age groups.

The final factor is how much English the subjects use in their daily life. It was expected that subjects who use English very frequently may have been more motivated and would improve more in this study. However, there was no observable relation between the percentages of English use and their improvement. Figure 6-13 shows the improvements in the posttest arranged by how much English they used in daily life (See

Appendix-XII for the questionnaire).

240

Figure 6-13. Improvement vs. the percentages of English use.

The subjects who use English 80% or more in their life made more improvement than

other groups, but for rest of the groups, there was no relationship found between the

percentages of English use and improvement. The percentages of English use was calculated based on a questionnaire each subject filled out before the pretest. The subjects filled in percentages of languages they use in their everyday life. In this study, the subjects, on average, used English 48.58% (Low-intermediate: 43.83%; Advanced:

53.33%), Japanese 50.17% (Low-intermediate: 55%; Advanced: 45.33%), and other languages (Korean, Hawaiian, Chinese and German) 1.25% (Low-intermediate: 1.17%;

Advanced: 1.33%).

6.7. General discussion of post tests

The results from Experiments 3 and 4 show the effects of instruction in acquiring

Verb + Noun collocations that are, on the whole, consistent with my predictions although some discrepancies were found.

241

Support for predictions.

Prediction 1 that instruction has a measurable role in acquisition was supported.

In both Experiments 3 and 4, there were overall improvements in the learners’

performance after explicit instruction. The learners produced more accurate collocations

for all types in the posttest. This suggests that the instruction in this study did help

learners acquire collocations more effectively. It helped learners understand overlapping

collocational restrictions between words with similar meanings and to correct their

mistaken over- or under-extensions. Those learners who used a wider range of materials

in studying tended to make greater improvements. Older subjects made more

improvement than younger subjects. There was no observable correlation between

improvement and the length of study or the amount of English use in daily life.

Regarding predictions 2-5, the same trend was observed: the improvements were

greater for difficult types than for easy ones (Easy types = 11.98%; Difficult types =

21.70%). Although the subjects made greater improvements in difficult types, overall

performance on easy types remained better than on difficult types (frequency: χ2 = 27.333, df = 1, p* < .000; meaning: χ2 = 11.342, df = 1, p* < .000; L1 equivalents: χ2 = 102.456,

df = 1, p* = 0; light vs content verbs: χ2 = 112.046, df = 1, p* = 0).

Prediction 2 that learners would demonstrate better productive knowledge of frequent collocations as compared to less frequent ones was supported. Although all the

collocations were presented at an equal frequency (i.e., all the collocations appeared once

in each material), the subjects attained higher scores for the more frequent types than

infrequent types. As observed in Experiments 1 and 2 (See 5.9.1), the average

242 percentages of acceptable (GJ & MC 4.5+) combinations in Experiment 4 (where all the collocations are neither frequent nor infrequent) generally fall between the scores for frequent and infrequent collocations although the advanced learners attained slightly higher scores for infrequent types than neither frequent nor infrequent types. Figure 6-14 shows the percentages of acceptable collocations of each type (frequent: the frequent collocations of Experiment 3; neither frequent nor infrequent: all the collocations of

Experiment 4; infrequent: the infrequent collocations of Experiment 3):

Figure 6-14. Acceptable collocations. Frequent, neither frequent nor infrequent and infrequent types.

This result suggests that although in the instruction all the collocations were presented at the same frequency, the learners’ difficulty still ranged on a cline from frequent collocations being easier to infrequent collocations being more difficult. However, the difference between frequent and infrequent types in Experiments 3 and 4 was not as great as in Experiments 1 and 2; the subjects made good improvement in infrequent types. The

243

average percentages of improvements (percentages of the accurate combinations in the

posttest minus percentages of accurate combinations in the pretest) in these three types

divided by frequency are shown in Figure 6-15:

Figure 6-15. Percentages of improvement.

There was a tendency for subjects to show more improvement for infrequent types. This difference was particularly evident among the advanced learners. The subjects made slightly greater improvement for mid-frequent collocations (Experiment 4) than for frequent types.

Prediction 3 that learners would demonstrate better productive knowledge of collocations which have literal meanings was also supported. Again, the subjects produced more acceptable combinations for literal types than for abstract types although all the collocations were presented in the same manner. Similar to the results from

Experiments 1 and 2, the percentages of acceptable collocations (GJ and MC 4.5+) in

Experiment 3 (all the collocations are relatively literal) fall between literal and abstract

244

types.

Figure 6-16. Acceptable collocations. Literal, relatively literal and abstract types.

The learners’ difficulty can be again described as a continuum; from literal collocations on the left being easier to abstract collocations on the right being more difficult. But again, it is necessary to collect more data to confirm this result because, in this study, the collocations in Experiment 1 were either frequent or infrequent while all the collocations in Experiment 2 are neither frequent nor infrequent.

The subjects made more improvements for abstract types than literal types. The improvement for relatively literal types was greater than for the literal type but less than the abstract type. Figure 6-17 summarizes the increase of percentages of acceptable collocations in the posttest:

245

Figure 6-17. Increase of acceptable collocations in the posttest. Literal, relatively literal and abstract types.

This result is similar to the increase of acceptable collocations ranged by frequency.

There are two possible reasons to explain why the subjects made more improvement in more difficult subtypes.

The first reason is that all the collocations (both easy and difficult types) were treated equally in the instruction materials. They all appeared once in each of the three materials, and roughly the same amount of emphasis was given to all the collocations. In many language classrooms, only a small number of extremely frequent collocations (i.e., the easy type) is presented. In contrast, it is possible that the instruction methods in this study enabled the subjects to direct more attention to the more difficult collocations.

The second reason is due to the differences in the accuracy in the pretest between the easy and difficult types. The subjects attained higher scores for all the easy types in the pretests, thus, there was more room for improvement in the difficult types. But as shown in Figure 6-17, the subjects still did better on easier collocations.

Prediction 4 that learners would perform better if there is a translational

246

equivalent in L1 was supported. The subjects did significantly better on collocations with

L1 equivalents than on those without, and greater improvement was observed in the collocations without L1 equivalents. As discussed in 5.9.1, the greater accuracy with L1 equivalents may reflect transfer from L1, but at the same time, the learners’ L1 can also cause negative transfer. There were a total of 110 erroneous forms (out of 3960 tokens)

in Experiments 3 and 4 which seem to be due to the interference from the subjects’ L1

Japanese. The percentages of negative transfer for each light verb are displayed in Figure

6-18:

Figure 6-18. Percentages of negative transfer for each Japanese light verb.

There is no data for have in this figure because the test did not contain questions which required translation of Japanese aru (have) to other verbs in English. The results in

Figure 6-18 were similar to those in Experiments 1 and 2 (see Figure 5-31). Among these eight verbs, the subjects committed relatively many interference errors for the same four verbs: ataeru (give), tsukuru (make), eru (get) and naru (become). As in Experiments 1

247

and 2, interference with suru (do) was very rare.

A large part of L1 transfer seems to be the result of idiosyncratic errors; 54.55%

of the erroneous forms were produced by fewer than 10 subjects (55.00% in Experiments

1 and 2). No erroneous forms were produced by more than half of the subjects in the

posttest (7 erroneous forms were produced by more than half of the subjects in the

pretest; see 5.9.2.2), but the following 5 errors were made by more than 10 of the 60 subjects: become victim (for fall victim) (25), subjects produced take court (for pay court)

(16), give damage (for cause damage) (15), become a sacrifice (for make a sacrifice) (12), and get a name (for have a name) (11).

Prediction 5 that light verbs are more difficult than content verbs was not supported in the posttest, either. Higher accuracy scores were observed for the light verbs than for the content verb types, and the subjects made greater improvement for the content verbs. However, again, this results does not lead to the conclusion that light verbs are easy. Errors with light verbs were quite frequent in the posttest as well. The errors due to misuse of light verbs consist of 31.67% among all the errors in the posttest. Figure

6-19 shows the numbers of misuse for each light verb.

248

Figure 6-19. Misuse of light verbs pretest Experiments 3 and 4.

Among the eight light verbs, the same four verbs (See Figure 5-32 in section 5.9.2.2):

make, have, take and get were misused more frequently than the other verbs. This result supports the pedagogical implication that teachers would do well to focus on these four light verbs.

Prediction 6 that advanced learners would perform significantly better than low-

intermediate learners was supported. Improved performance was observed at the higher

proficiency level; 52.97% of low-intermediate learners’ collocations and 69.89% of

advanced learners’ collocations were judged to be native-like. Both proficiency groups

made greater improvement in difficult types than in easy types. Figure 6-20 summarizes

the increases in acceptable combinations in the posttest.

249

Figure 6-20. Increase of accurate percentages in the posttest. Easy vs. difficult types.

Not much difference was found in the easy types between the low-intermediate and advanced learners’ improvement, whereas the advanced group’s performance for the difficult types improved somewhat more than that of low-intermediate learners.

250

CHAPTER 7

Summary and conclusion: findings, implications and future directions

7.1. Summary

Although it is widely acknowledged that collocations play an important part in

second language learning, especially at intermediate-advanced levels (Ellis, 2001;

Fillmore, 1976; Hakuta, 1974; Lewis, 1993; Pawley and Syder, 1983), learners'

difficulties with collocations have not yet been investigated in much detail. This study

aimed to empirically examine production of verb-noun collocations by Japanese learners

of English, and the effects of instruction which directed learners' attention to collocations

and to restrictions on their use.

Sixty-six Japanese students (35 intermediate, 31 advanced) took a fill-in-the-blank

test focusing on the production of English collocations. This was followed by a single

session of instruction, with a brief introduction to collocations and a discussion of

common mistakes and differences in the restrictions between English and Japanese usage.

The subjects then agreed to study collocations at home for two weeks. They used paper-

based exercises provided after the pretest and online flashcards. After the study period 60

subjects (30 intermediate, 30 advanced) came back to take a second fill-in-the blank test

which served as a posttest.

Statistical analyses show significant influences on learners' performance of various properties of the collocations tested: relative frequency, literal vs. abstract

meaning, the existence of L1 equivalents, light vs. content verbs, and same vs. different

light verbs. The following eleven error types were identified: (1) Inappropriate

251 paraphrases; (2) Interference of the native language (Japanese); (3) Misuse of light verbs

(take, give, have, etc); (4) Blending two collocations with similar meanings; (5)

Unnecessary articles, particles and prepositions inserted between verbs and nouns; (6)

Confusions between intransitive and transitive verbs; (7) Misunderstanding actor-patient relations of verbs; (8) Mistakes using derivationally related words (stop a stop for put a stop); (9) Creating collocations from compound nouns; (10) Use of words other than verbs; and (11) Phonological errors. Among these error types, it was found that paraphrases, misuse of light verbs and interference of the native language had the strongest impact on the difficulty of collocations for the learners. We can conclude that it is useful to make learners aware of possible interferences between collocations in their native and target languages.

Significant improvements were observed in the posttest (the average pretest score was 52.55%, while the post-test average was 70.32%). These findings highlight the possibility of improving second language learners' collocational knowledge so as to enhance their proficiency in a target language. It also shows that explicit instruction focusing on collocations indeed improves their collocational competence in the target language. Among the eight light verbs tested, it was found that the four verbs: have, get, make and take, were overused much more than the others. Thus teachers could focus on these troublesome light verbs in class.

7.2. Suggestions for future research

To investigate second language collocations further, I suggest the following research topics.

252

• Examine naturalistic data

The current experiment employed a fill-in-the-blank test where verbs were

replaced by brackets, but it is possible that L2 learners may use expressions other than V

+ N collocations in some of these contexts. After the experiment, some of the subjects

commented that they felt more comfortable using just a single word rather than a V + N

combination. For example, they preferred The accident happened this morning to The

accident took place this morning.

Moreover, a few advanced learners commented that when they use English, they usually do not translate from Japanese to English. Instead they think and conceptualize their idea in English. Although I used a translation task in my data collection, it is possible that this method does not adequately reflect the competence of some of the learners, in particular, the very advanced ones.

Collecting more naturalistic data, such as free narratives, dialogues and compositions will be helpful in understanding the actual performance of L2 learners.

• Examine other factors

My study mainly explored the five factors which most heavily influenced learners’ performance. However, the results suggest other factors which may also affect learners’ performance, such as the restrictedness of collocations, learners’ age, their country of residence, their motivation and so on. Not enough of this kind of data was collected in the present study. Issues which should be explored further include the relation between restrictedness of collocations and L1 transfer, and more investigation of

253

individual differences (age, learning styles, personality, etc).

• Improve teaching methods and instruction materials

My study shows that explicit instruction is effective in learning collocations, but

my teaching methods and materials could be improved better to meet learners’ needs,

learning styles, personality and so on. In this study, I used the same method and same materials for all the subjects, but the evaluations of my materials by the subjects indicated

that my method was more highly rated by older subjects than by younger ones. One

possible approach is to have L2 learners experience several instruction sessions, each of

which employs a different pedagogical method, followed by an individual or group

interview to get reactions about the instruction methods.

Another important topic is to develop materials and effective instruction methods

to teach the four problematic light verbs: have, get, make and take.

• Collect more data on light verbs

This study dealt with how Japanese collocations involving light verbs can be translated into English (i.e., using same vs. different light verbs or content verbs).

However, there are collocations where Japanese content verbs are translated into English light verbs. These were beyond the scope of my investigation, but it is an important topic to investigate learners’ performance with light verb collocations more fully.

• Examine articles, particles and prepositions

Investigating the use of articles, particles, and prepositions in collocations are also

254 important future research topics. My study indicated that unnecessary additions of articles, particles and prepositions often considerably lowered the scores for both

Grammaticality and Meaning Conveyed.

• Examine other constructions

The focus of the present study was on Verb + Noun collocations, but there are other constructions, such as Adjective + Noun, Adverb + Verb and Verb + Prepositional phrases yet to be studied.

• Examine cross-linguistic data

In the current study, the subjects were Japanese who were studying English, but it will be interesting to investigate L2 collocations cross-linguistically. I am especially interested in how overall frequency of formulaic language in a given language may influence second language acquisition. Tannen and Öztek (1979: 38) report crosslinguistic differences in the frequency of formulae in adult speech. Their research shows that formulae in Turkish and Greek, as in English, fall along a continuum with respect to how often they are used, and how obligatorily they are used. Turkish has many formulae which cluster at the obligatory end of the continuum, while formulae in English tend towards the optional end. Greek has fewer fixed formulae than Turkish but many more than English. It will be interesting to compare Turkish and Greek learners of

English with each other. The findings might suggest that Turkish learners raised in their mother language backgrounds with rich formula usage may have stronger sensitivity to the importance of formulaic language as compared with Greek learners whose language

255

environment does not place so much stress on formulaic language.

Other areas of interest for future work include psychological and socio-cultural issues, and the relationship between collocations and grammar development in a second language. Such research can contribute to the field of language acquisition both for academic professionals and language learners by deepening our understanding of learners’ performance and by improving instructional methods.

256

Appendix I: Test sentences for the preliminary survey

１．彼女は学校の帰り道で事故に遭った。 She had an accident on her way home from school.

２．行動を起こす前にもっと情報を集める必要がある。 We need to gather more information before we take action.

３．水泳をすれば食欲が出るよ。 Swimming will increase your appetite.

４．前チャンピオンは引退して１年後に再起をはかったが失敗した。 The former champion made an attempt to come back after a year’s retirement but failed.

５．事業に成功しようと思ったら、時には危険を冒さなければならない。 You sometimes have to take/ run a risk to succeed in an undertaking.

６．目的を達成するために最善を尽くしなさい。 You should do/ try your best to achieve your goal.

７．いらぬお節介をするな。 Mind your own business.

８．（トランプの）札を配る番ですよ。 It’s your turn to deal the cards.

９．ここにじゅうたんを敷きましょうか。 Shall we lay the carpet here?

１０．彼女はよく授業をさぼる。 She often cuts/ skips class.

１１．恐ろしい事故現場を見て、彼は青ざめた。 Seeing the horrible scene of the accident, he changed/ lost color.

１２．付き合っている仲間を見ればその人柄が分かります。 You can judge a man’s character by the company he keeps.

１３．確かにエリザベスは美しいが、シャロンにはかなわない。 True, Elizabeth is beautiful, but she does not bear/ stand comparison with Sharon.

１４．ご都合がよろしければ、あす事務所へうかがいます。 I’ll come over to your place if it meets your convenience.

１５．この単語の意味が分からなければ、辞書で調べなさい。

257

If you don’t know the meaning of this word, check/ consult a dictionary.

１６．君が映画に行こうが家で勉強しようが、私にはどうでもよい。 It makes no difference to me whether you go to the movies or study at home.

１７．卵を泡立ててください。 Please beat/ whisk the eggs.

１８．メアリーの赤いセーターが私の目に留まった。 Mary’s red sweater caught my eyes.

１９．切符を買ってくるから、その間子どもたちから目を離さないでね。 Please keep an eye on the children while I am buying tickets.

２０．「お願いがあるのですが。」「いいですよ、もし私にできることでしたら。」 “May I ask a favor of you?” “Yes, if it’s anything I can do.”

２１．どうか見つかりませんようにと念じて、ローズは息を殺していた。 Rose held her breath, hoping that they wouldn’t find her.

２２．日本の木造家屋は燃えやすい。 Japanese wooden houses catch fire easily.

２３．あなたの努力が実を結ぶといいですね。 I hope your efforts will bear fruit.

２４．住友教授の最終講義が終ると、学生全員が教授に拍手を送った。 All the students gave Professor Sumitomo a big hand when his final lecture ended.

２５．私は昨日パチンコで大当たりした。 I hit the jackpot on the pinball machine yesterday.

２６．公平に見て、ヘレンよりジュリーの方が料理上手だ。 To do his justice, Julie is a better cook than Helen.

２７．子どものころ、よくたこ揚げをしました。 I often flew a kite when I was a child.

２８．フレッドは君にうそをついたに違いない。 Fred must have told you a lie.

２９．ドナは２０歳のとき自殺した。 Donna ended/ took her life when she was 20 years old.

３０．暇なときには便りをくれよ。 Drop me a line when you have time.

３１．お医者さんがくれた薬を飲むのを忘れないようにね。

258

Don’t forget to take medicine the doctor gave you.

３２．子どもは悪い夢を見て泣き叫んだ。 The child had a nightmare and cried.

３３．医者はまず患者の脈を測った。 The doctor first took the patient’s pulse.

３４．トムがなぜ大学を辞めるのか理由が分かりません。 I see no reason shy Tom should quit college.

３５．マリアは声をひそめて私に秘密を打ち明けた。 Maria dropped/ lowered her voice to tell me the secret.

３６．あの恐ろしい経験は言葉では言い表せません。 I cannot put my terrible experience into words.

３７．最終電車に乗り遅れ、タクシーで帰宅せざるを得なかった。 I missed the last train and had to go home by taxi.

３８．この写真をすぐに現像して下さい。 Please develop this photo right away.

３９．マイクは絵を描くのが好きです。 Make likes to draw pictures.

４０．漁業交渉は先週急速に進展した。 The fish negotiations made rapid progress last week.

259

Appendix II: Data from the preliminary survey

GJ = Native speakers’ Grammaticality Judgments (5-point-scale) MC = Native speakers’ Meaning Conveyed (5-point-scale)

The combinations are listed from the highest to the lowest average scores of the GJ and the MC.

1. She had an accident on her way home from school. GJ MC GJ MC had an accident 5.0 5.0 ran into an accident 1.5 2.0 made an accident 2.6 3.0 was an accident 1.4 1.8 got an accident 2.5 3.0 happened an accident 1.0 1.4 met an accident 1.8 2.5

2. We need to gather more information before we take action. GJ MC GJ MC take action 5.0 5.0 act action 1.2 2.1 begin action 3.9 4.5 move action 1.5 1.8 start action 3.8 4.5 get action 1.6 1.2 make action 2.5 4.0 gather action 1.2 1.2 do action 2.7 3.7 will action 1.2 1.2

3. Swimming will increase your appetite. GJ MC GJ MC increase your appetite 5.0 5.0 gain your appetite 2.4 2.8 stimulate your appetite4.7 4.8 power your appetite 2.1 2.2 help your appetite 3.7 4.1 make your appetite 1.6 1.9 grow your appetite 3.5 4.0 get your appetite 1.2 1.6 up your appetite 3.2 4.1 give your appetite 1.3 1.3 bring your appetite 3.0 3.5 health your appetite 1.1 1.3

4. The former champion made an attempt to come back after a year’s retirement but failed. GJ MC GJ MC made an attempt 5.0 5.0 lost an attempt 2.0 1.3 tried an attempt 3.2 4.4 missed an attempt 1.8 1.1 had an attempt 3.5 3.2 stopped an attempt 1.5 1.3 took an attempt 2.5 3.7 failed an attempt 1.2 1.5 did an attempt 2.4 3.8 resigned an attempt 1.4 1.0 planned an attempt 3.2 2.9 again an attempt 1.0 1.3 got an attempt 2.5 2.1

5. You sometimes have to take/ run a risk to succeed in an undertaking. GJ MC GJ MC run a risk 5.0 5.0 make a risk 2.3 3.2 take a risk 5.0 5.0 get a risk 1.7 1.6 experience a risk 2.8 3.1 dangerous a risk 1.0 1.0 do a risk 2.3 3.3 success a risk 1.0 1.0

260 have a risk 2.4 2.3

6. You should do/ try your best to achieve your goal. GJ MC GJ MC do your best 5.0 5.0 be your best 3.2 3.8 try your best 5.0 5.0

7. Mind your own business. GJ MC GJ MC Mind your own business 5.0 5.0 Not your own business 2.1 1.9 Do your own business 3.1 3.4 This is your own business2.2 1.6 Concentrate your own business2.3 2.9 None your own business1.3 1.4 It’s your own business 2.8 2.0 Don’t your own business1.0 1.2 It’s not your own business 2.5 1.8

8. It’s your turn to deal the cards. GJ MC GJ MC deal the cards 5.0 5.0 get the cards 2.6 1.8 distribute the cards 4.2 5.0 put the cards 1.9 2.0 pass the cards 3.9 4.1 make the cards 1.6 1.8 shuffle the cards 4.3 3.1 take the cards 2.0 1.3 give the cards 2.6 4.2 shake the cards 1.3 1.4 provide the cards 2.5 3.1 deliver the cards 1.5 1.0 spread the cards 1.8 3.5 on the cards 1.4 1.0 play the cards 2.2 2.3

9. Shall we lay the carpet here? GJ MC GJ MC lay the carpet 5.0 5.0 seat the carpet 2.2 3.1 put the carpet 4.9 4.8 do the carpet 2.0 2.2 place the carpet 4.5 5.0 take the carpet 2.3 1.8 spread the carpet 3.8 4.2 make the carpet 2.0 1.7 apply the carpet 3.6 4.3 down the carpet 1.5 2.0 roll the carpet 3.5 3.1 get the carpet 1.6 1.2 have the carpet 3.1 3.4 pat the carpet 1.4 1.1

10. She often cuts/ skips class. GJ MC GJ MC cuts class 5.0 5.0 cancels class 2.8 1.8 skips class 5.0 5.0 disappears class 1.0 2.3 misses class 4.8 4.7 absence class 1.0 2.0 escapes class 2.9 3.8 loses class 1.2 1.1 absent class 1.6 3.3 off class 1.0 1.3 ignore class 2.5 2.5

11. Seeing the horrible scene of the accident, he changed/ lost color. GJ MC GJ MC changed color 5.0 5.0 became color 1.1 1.4 lost color 5.0 5.0 was color 1.1 1.4 turned color 3.3 4.1 felt color 1.2 1.1 paled color 1.5 2.4 blue color 1.0 1.1 got color 1.2 1.7 blued color 1.0 1.1

261

12. You can judge a man’s character by the company he keeps. GJ MC GJ MC the company he keeps 5.0 5.0 the company he sees 2.1 2.4 the company he has 4.6 4.7 the company he contacts1.8 2.7 the company he hang out with 2.5 5.0 the company he faces 1.5 1.6 the company he associates 2.5 4.5 the company he plays 1.4 1.3 the company he knows 3.2 3.5 the company he is 1.2 1.1 the company he accompanies 2.0 3.0 the company he stays 1.2 1.0 the company he brings 2.6 2.2 the company he character1.0 1.0 the company he belongs 2.0 2.7

13. True, Elizabeth is beautiful, but she does not bear/ stand comparison with Sharon. GJ MC GJ MC stand comparison 5.0 5.0 make comparison 1.7 2.3 reach comparison 2.6 3.1 in comparison 1.1 2.3 deserve comparison 2.6 2.8 be comparison 1.4 1.5 go comparison 1.5 3.5 as comparison 1.4 1.3 beat comparison 1.8 2.5 enough comparison 1.2 1.5 have comparison 2.0 2.3 matter comparison 1.4 1.3 take comparison 1.8 2.3 more comparison 1.2 1.3 win comparison 1.5 2.6

14. I’ll come over to your place if it meets your convenience. GJ MC GJ MC meets your convenience 5.0 5.0 has your convenience 1.4 2.3 fits your convenience 3.9 4.5 whenever your convenience 1.4 2.2 is your convenience 3.2 3.8 provides your convenience 1.7 1.7 matches your convenience 2.5 3.6 gets your convenience 1.3 1.6 becomes your convenience 1.5 4.0 possible your convenience 1.0 1.3 good your convenience 1.4 2.8 frees your convenience 1.0 1.1

15. If you don’t know the meaning of this word, check/ consult a dictionary. GJ MC GJ MC check a dictionary 5.0 5.0 look a dictionary 1.5 3.5 use a dictionary 5.0 4.5 have a dictionary 1.8 2.1 see a dictionary 4.3 4.7 take a dictionary 1.9 2.0 find a dictionary 4.5 3.9 keep a dictionary 1.8 1.2 look up a dictionary 2.2 3.3 draw a dictionary 1.2 1.1

16. It makes no difference to me whether you go to the movies or study at home. GJ MC GJ MC makes no difference 5.0 5.0 has no difference 2.9 2.9 is no difference 4.3 4.5 either no difference 1.1 1.3 there is no difference 4.0 4.6 whether no difference 1.0 1.2 means no difference 3.2 3.9 ideas no difference 1.0 1.0

17. Please beat/ whisk the eggs. GJ MC GJ MC whisk the eggs 5.0 5.0 shake the eggs 2.4 2.2 whip the eggs 4.5 4.2 foam the eggs 1.6 2.1 mix the eggs 4.0 4.0 shuffle the eggs 1.3 1.8

262 scramble the eggs 3.9 3.9 screw the eggs 1.0 1.0 stir the eggs 4.1 3.7 stand the eggs 1.0 1.0 bubble the eggs 1.5 3.5

18. Mary’s red sweater caught my eyes. GJ MC GJ MC caught my eyes 5.0 5.0 appeared my eyes 1.3 1.1 captured my eyes 4.2 4.3 looked my eyes 1.2 1.1 intrigued my eyes 3.4 4.2 stayed my eyes 1.2 1.1 got my eyes 2.9 3.2 came my eyes 1.2 1.0 froze my eyes 2.3 2.6 at my eyes 1.0 1.0 stopped my eyes 2.2 2.7 on my eyes 1.0 1.0 took my eyes 1.7 2.4

19. Please keep an eye on the children while I am buying tickets. GJ MC GJ MC keep an eye 5.0 5.0 look an eye 1.1 1.5 put an eye 2.5 3.8 take an eye 1.0 1.4 hold an eye 1.7 3.1 don’t an eye 1.0 1.2 watch an eye 1.6 3.2 lose an eye 1.1 1.0

20. “May I ask a favor of you?” “Yes, if it’s anything I can do.” GJ MC GJ MC ask a favor 5.0 5.0 help a favor 1.2 1.5 have a favor 3.6 4.0 give a favor 1.2 1.3 do a favor 2.5 1.5 make a favor 1.3 1.2

21. Rose held her breath, hoping that they wouldn’t find her. GJ MC GJ MC held her breath 5.0 5.0 kept her breath 2.7 3.3 calmed her breath 4.0 4.0 lost her breath 2.9 1.5 stopped her breath 3.7 3.8 killed her breath 1.5 1.8 lowered her breath 2.5 4.0 played her breath 1.1 1.2

22. Japanese wooden houses catch fire easily. GJ MC GJ MC catch fire 5.0 5.0 get fire burn fire 1.3 2.1 burn fire 2.7 2.7 hold fire 1.6 1.5 make fire 2.3 2.9 on fire 1.0 1.8 can fire 1.8 2.4 are fire 1.2 1.5 have fire 1.7 2.3 easy fire 1.1 1.4

23. I hope your efforts will bear fruit. GJ MC GJ MC bear fruit 5.0 5.0 get fruit 1.7 2.3 bring fruit 4.0 4.5 be fruit 1.1 1.3 make fruit 2.3 2.8 good fruit 1.0 1.1 gain fruit 2.2 2.7

24. All the students gave Professor Sumitomo a big hand when his final lecture ended. GJ MC GJ MC gave Prof. S big hand 5.0 5.0 sent Prof. S a big hand 1.3 2.2

263 applauded Prof. S a big hand 3.2 4.1 took off Prof. S a big hand 1.0 1.2 clapped Prof. S a big hand 1.7 2.9

25. I hit the jackpot on the pinball machine yesterday. GJ MC GJ MC won the jackpot 4.8 4.4 had the jackpot 2.6 3.1 got the jackpot 4.6 4.4 lucked the jackpot 1.5 2.3 made the jackpot 4.0 5.0 betted the jackpot 1.3 1.4 earned the jackpot 3.0 4.0

26. To do his justice, Julie is a better cook than Helen. GJ MC GJ MC see her justice 2.0 2.1 be her justice 1.2 1.3 make her justice 2.0 1.9 take her justice 1.3 1.2 equal her justice 1.8 2.0 think her justice 1.3 1.1 fair he justice 1.4 1.9 official her justice 1.0 1.3 judge her justice 1.3 1.6 look her justice 1.0 1.0 average her justice 1.1 1.4 as her justice 1.0 1.0

27. I often flew a kite when I was a child. GJ MC GJ MC flew a kite 5.0 5.0 do a kite 1.0 2.5 flight a kite 1.2 2.5 up a kite 1.1 1.9 played a kite 1.3 2.4 took a kite 1.3 1.6 rose a kite 1.3 2.3 floated a kite 1.0 1.5

28. Fred must have told you a lie. GJ MC GJ MC told you a lie 5.0 5.0 been you a lie 1.1 1.2 said you a lie 1.5 3.2 white you a lie 1.0 1.2 made you a lie 1.4 1.9 with you a lie 1.0 1.1 cheated you a lie 1.4 1.5

29. Donna ended/ took her life when she was 20 years old. GJ MC GJ MC ended her life 5.0 5.0 suicide her life 1.7 3.6 took her life 5.0 5.0 quit her life 1.9 2.6 lost her life 4.4 2.9 pau her life 2.0 2.2 finished her life 3.0 4.2 threw her life 1.3 1.6 killed her life 2.0 4.0 committed her life 1.2 1.5 closed her life 3.5 2.0

30. Drop me a line when you have time. GJ MC GJ MC Drop me a line 5.0 5.0 Connect me a line 1.3 1.8 Send me a line 4.0 4.5 Call me a line 1.3 1.4 Give me a line 3.6 3.2 Letter me a line 1.0 1.6 E-mail me a line 2.8 3.3 Draw me a line 1.3 1.0 Mail me a line 1.9 3.1 Hear me a line 1.2 1.1 Write me a line 2.3 2.7 Please me a line 1.0 1.0 Tell me a line 1.5 2.0 Keep me a line 1.0 1.0

264

31. Don’t forget to take medicine the doctor gave you. GJ MC GJ MC take medicine 5.0 5.0 get medicine 2.5 2.1 drink medicine 3.7 4.1

32. The child had a nightmare and cried. GJ MC GJ MC had a nightmare 5.0 5.0 screamed a nightmare 1.5 1.4 dreamed a nightmare 4.3 4.4 shouted a nightmare 1.5 1.3 recalled a nightmare 3.4 3.1 bad a nightmare 1.2 1.4 saw a nightmare 2.2 3.0 yelled a nightmare 1.2 1.4 bad-dreamed a nightmare 1.8 2.4

33. The doctor first took the patient’s pulse. GJ MC GJ MC took the patient’s pulse 5.0 5.0 measured the patient’s pulse 3.9 3.7 checked the patient’s pulse 5.0 4.8 counted the patient’s pulse 3.1 3.3 recorded the patient’s pulse 4.5 4.5 watched the patient’s pulse 2.2 1.8 tested the patient’s pulse 4.0 4.0 saw the patient’s pulse 1.5 2.0

34. I see no reason shy Tom should quit college. GJ MC GJ MC see no reason 5.0 5.0 don’t no reason 1.4 1.7 have no reason 3.1 3.6 am no reason 1.4 1.0 understand no reason 2.9 3.0

35. Maria dropped/ lowered her voice to tell me the secret. GJ MC GJ MC dropped her voice 5.0 5.0 raised her voice 1.8 1.3 turned down her voice 2.6 4.5 killed her voice 1.3 1.6 decreased her voice 2.9 4.2 kept her voice 1.3 1.2 held her voice 2.5 4.0 made her voice 1.2 1.3 whispered her voice 2.5 3.4 taught her voice 1.2 1.2 devoiced her voice 1.0 3.5 said her voice 1.1 1.0 littled her voice 1.8 2.5

36. I cannot put my terrible experience into words. GJ MC GJ MC put my terrible experience 5.0 5.0 tell my terrible experience 2.8 3.6 translate my terrible experience 4.5 5.0 say my terrible experience 2.4 3.7 express my terrible experience 4.2 3.9 make my terrible experience 2.4 2.8 describe my terrible experience3.4 3.9 teach my terrible experience 1.4 1.9 explain my terrible experience 2.9 3.8 stand my terrible experience 1.0 1.5

37. I missed the last train and had to go home by taxi. GJ MC GJ MC missed the last train 5.0 5.0 got the last train 1.4 1.0 lost the last train 1.5 2.4 took the last train 1.4 1.0 mistook the last train 1.4 1.6 must the last train 1.0 1.2

38. Please develop this photo right away. GJ MC GJ MC

265 develop this photo 5.0 5.0 improve this photo 2.0 2.0 print this photo 3.8 3.7 take this photo 1.8 1.8 make this photo 2.9 3.2 dry this photo 1.9 1.4 copy this photo 2.3 2.0 soon this photo 1.2 1.2

39. Make likes to draw pictures. GJ MC GJ MC draw pictures 5.0 5.0 write pictures 1.9 2.4 make pictures 3.9 3.9

40. The fish negotiations made rapid progress last week. GJ MC GJ MC made rapid progress 5.0 5.0 improved rapid progress 1.7 2.4 had rapid progress 3.9 4.0 proceeded rapid progress 1.5 2.1 got rapid progress 2.9 2.9 went rapid progress 1.2 1.6 developed rapid progress 2.3 3.1 fast rapid progress 1.1 1.4 progressed rapid progress 2.1 2.8

266

Appendix III: Test sentences for Experiment 1

A1 frequent, same light verb • 「傾向がある」 Female tennis players have a tendency to gain weight more than men.

• 「許可を与えた」 The photographer John William Joensen kindly gave permission for the images to be reposted here.

• 「商売をしている」 Everyone doing business in the City of Seattle must obtain a business license.

• 「メモを取る」 That columnist doesn’t take notes during interviews.

• 「手紙を受け取った」 I recently received a letter from a friend of mine in a developing country.

• 「仕事を得た」 After graduating from college Lane got a job as a computer programmer.

A1 infrequent, same light verb • 「新聞をとっている（購読している）」 Over 20% of the students are taking a newspaper each day.

• 「発言権がある」 Does the husband have a say in his wife's property?

• 「削減があった」 This year we have had a cut in case-management funding of $150000.

• 「掃除をする」 If you have severe asthma, have someone else do the cleaning.

• 「許可を与える」 The Swedes are hesitating to give clearance to build the bridge because of the strength of the opposition to the project.

• 「大評判となった」 Tamagotchi, electronic toy designed by Akihiro Yokoi, has become a sensation around world with orders this year of 70 million.

A2 frequent, different light verb • 「アポを取る」 Do you wish to make an appointment with the physician?

267

• 「昼食をとる」 I didn't have lunch today.

• 「決定した」 We finally made a decision this morning and purchased a Kodak digital camera.

• 「考慮する」 The project is taking account of recent developments in software and standards.

• 「利用する」 You will find it convenient to make use of one of the two banks on campus.

• 「証言する」 They did not know that Roberts was going to give evidence in the second trial.

A2 infrequent, different light verb • 「パスポートをとった」 Joshua got a passport in 1997, hoping to travel to the UK for a year.

• 「故障した」 Her Ferrari had a breakdown recently on a California highway.

• 「犠牲になった」 Christians say Jesus made a sacrifice when he died on the cross.

• 「相談する」 The rulers take counsel together in order to talk out their differences.

• 「得点した」 Jeffrey Taylor made a goal, winning the game for Newberry 1-0.

• 「発見する」 It is impossible to predict which kid will make a discovery that will change the world.

A3 frequent, content verb • 「裁判を受けた」 Skinheads, who killed an Armenian, stood trial in Moscow on July 26, 2004.

• 「犠牲になった」 In Manila seaweed farmers fell victim to a strong peso.

• 「自殺する」 She has never tried to commit suicide, but she constantly thinks about it.

• 「質問すること」 That is why asking questions about what happened on Haifa Street is very important.

• 「監視する」 Can you keep an eye on my suitcase while I am away?

268

• 「停止する」 Legislators are trying to put a stop to online gambling.

A3 infrequent, content verb • 「投票する」 Nearly 8.8 million eligible Canadians (39.1 per cent) didn't cast a ballot in the 2004 election.

• 「バランスをとる」 A factory manager must keep a balance between cost to produce quality of product and service to the customer.

• 「染みをとる」 When possible, it is best to try to remove a stain before it dries.

• 「米をつくる」 Farmers use the latest technology to grow rice.

• 「重視する」 Health care providers put emphasis on customer service.

• 「学校をつくった」 Mary’s dream is to found a school for young children and to develop a curriculum that blends art and science.

269

Appendix IV: Questionnaire before the pretest (Experiments 1 and 2)

あてはまるものに○をつけてください。(Please circle or fill in the response for each question)

1．性別 (Sex)

男性 (male) 女性 (female)

2．年齢 (Age)

歳 (years)

３．学校名 (name of school)

４．英語圏滞在期間 (Length of stay in English-speaking countries)

年 (years) ヶ月 (months)

５．何歳の時から英語の学習を始めましたか？ (At what age did you start studying English?)

歳 (years)

６．もし英語の資格試験を受験されたことがあれば、分かる範囲で記入してください。 (Scores for the following English exams and dates you took them)

TOEFL 点受験年月 (date) 年 (year) 月 (month) TOEIC 点受験年月 (date) 年 (year) 月 (month)

英検(Test in Practical English Proficiency conducted by the Society for Testing English Proficiency, Japan) 級 (grade) 受験年月 (date) 年 (year) 月 (month)

７．普段の生活で、英語・日本語・その他の言語の使用比率はどのくらいですか。（例：英語７０％、日本語３０％） The percentage of languages used in daily life. (Example: English 70%, Japanese 30%)

英語 (English) ％日本語 (Japanese) ％その他の言語 (Other languages) 語(name of language) ％

270

Appendix V: Data from Experiment 1

# of responses = number of responses produced by subjects (n = 66) GJ = Native speakers’ Grammaticality Judgments (6-point-scale) MC = Native speakers’ Meaning Conveyed (6-point-scale)

The combinations are listed from the highest to the lowest average scores of the GJ and the MC.

A1 frequent, same light verb Female tennis players have a tendency to gain weight more than men. # of responses GJ MC have a tendency 53 6.00 6.00 show a tendency 2 3.67 4.83 display a tendency 1 3.33 4.67 tend to have a tendency 1 2.50 5.00 carry a tendency 3 3.00 4.33 hold a tendency 2 2.83 3.50 experienced a tendency 1 2.50 3.67 learn toward a tendency 1 1.50 3.00 get a tendency 6 1.17 3.00 can see a tendency 1 1.33 2.33 take to a tendency 1 1.33 2.33 suggest a tendency 1 1.17 2.17 take a tendency 3 1.17 2.17 do a tendency 2 1.17 1.83 make a tendency 2 1.17 1.83 gain a tendency 1 1.33 1.50 keep a tendency 2 1.00 1.83 are a tendency 4 1.00 1.67 tend a tendency 1 1.17 1.50 tend to be a tendency 1 1.00 1.50 change a tendency 1 1.00 1.33 exist a tendency 1 1.00 1.33 incline a tendency 1 1.00 1.33 play a tendency 1 1.00 1.17 tend to a tendency 1 1.00 1.17 trend a tendency 1 1.00 1.17 look a tendency 1 1.00 1.00 NO ANSWER 1 – –

A1frequent, same light verb The photographer John William Joensen kindly gave permission for the images to be reposted here. # of responses GJ MC gave permission 57 6.00 6.00 granted permission 1 5.00 5.33 provided permission 6 2.67 4.67 issued permission 2 2.50 4.33 allowed permission 5 2.17 3.33 gave a permission 4 2.00 3.50

271

gave away permission 1 2.33 3.17 offered permission 2 1.00 4.17 admitted permission 1 1.83 3.33 passed permission 1 1.67 2.17 assigned permission 1 1.17 2.50 allowed to give permission 1 1.00 2.33 had permission 2 1.17 2.33 showed permission 1 1.17 2.33 got permission 5 1.00 2.00 received permission 1 1.00 2.00 made permission 7 1.00 1.83 accepted permission 3 1.00 1.17 took permission 2 1.00 1.17 asked permission 2 1.00 1.00

A1 frequent, same light verb Everyone doing business in the City of Seattle must obtain a business license. # of responses GJ MC doing business 32 6.00 6.00 conducting business 1 5.17 5.67 running a business 2 5.33 4.50 having a business 1 4.17 5.33 owning a business 1 4.00 4.50 in business 1 3.83 4.50 doing a business 1 3.33 4.67 managing business 2 3.00 3.50 starting business 1 2.50 3.83 running business 13 1.83 4.33 establishing business 1 2.50 3.50 handling business 1 2.67 3.17 having business 31 2.33 3.33 owning business 11 1.83 3.50 holding business 4 2.50 2.67 making a business 2 2.17 2.67 opening business 1 2.17 2.50 keeping business 5 2.00 2.50 working for their owning business 1 1.00 3.17 providing business 1 1.83 2.17 making business 14 1.50 2.50 owning company business 1 1.17 2.83 possessing business 1 1.83 2.00 controlling business 1 1.67 2.00 doing the business 1 1.33 2.33 getting business 5 1.33 2.33 with business 1 1.17 2.17 acting business 1 1.00 2.33 taking business 2 1.17 1.50 enjoying business 1 1.00 1.33 geeing business 1 1.00 1.00 NO ANSWER 2 – –

A1 frequent, same light verb

272

That columnist doesn’t take notes during interviews. # of responses GJ MC take notes 63 6.00 6.00 write down notes 8 6.00 6.00 write notes 25 5.67 5.83 make notes 25 4.83 5.67 jot notes 1 5.00 5.00 keep notes 5 4.83 5.17 scribble notes 1 3.83 5.00 put down notes 1 3.00 4.67 prepare notes 1 2.67 3.17 do notes 1 1.83 3.33 hold notes 1 1.67 2.83 have notes 13 1.50 2.33 retain notes 1 1.50 2.00 put down on notes 1 1.17 2.17 get notes 7 1.33 2.00 write down on notes 1 1.33 1.83 report notes 1 1.00 2.00 fix notes 1 1.33 1.50 receive notes 1 1.00 1.17

A1 frequent, same light verb I recently received a letter from a friend of mine in a developing country. # of responses GJ MC received a letter 64 6.00 6.00 got a letter 61 5.67 5.83 obtained a letter 1 4.33 5.33 was sent a letter 3 4.33 5.17 was handed a letter 1 2.50 4.17 accepted a letter 6 2.33 4.00 gained a letter 1 1.83 3.50 had a letter 21 2.00 3.00 kept a letter 1 1.67 1.83 took a letter 3 1.17 2.17 achieved a letter 1 1.00 1.83 was delighted a letter 1 1.17 1.33 mailed a letter 1 1.00 1.17 was taken a letter 1 1.00 1.17 sent a letter 1 1.00 1.00

A1 frequent, same light verb After graduating from college Lane got a job as a computer programmer. # of responses GJ MC got a job 65 6.00 6.00 acquired a job 4 5.50 5.83 attained a job 1 5.33 5.83 found a job 10 5.00 5.83 obtained a job 12 4.83 5.83 was hired for a job 1 5.00 5.17 accepted a job 1 4.83 4.50 received a job 8 4.83 4.50

273

took a job 7 4.33 5.00 started a job 1 4.17 4.33 gained a job 5 3.83 4.50 was offered a job 2 3.00 3.50 had a job 24 1.67 3.33 joined to a job 1 1.17 3.33 made a job 4 1.17 2.50 was hired a job 2 1.67 2.00 was accepted a job 1 1.33 2.33 received in a job 1 1.33 1.83 employed a job 1 1.17 1.83 set a job 1 1.00 1.83 positioned a job 1 1.17 1.50 offered a job 1 1.17 1.33 was accepted from a job 1 1.00 1.50

* * * * * * * * * * * * * * * *

A1 infrequent, same light verb Over 20% of the students are taking a newspaper each day. # of responses GJ MC taking a newspaper 30 6.00 6.00 getting a newspaper 10 4.50 5.67 buying a newspaper 29 2.50 3.50 purchasing a newspaper 10 2.50 3.50 subscribing to a newspaper 1 3.00 2.83 delivered a newspaper 1 2.67 2.83 looking at a newspaper 1 2.50 3.00 given a newspaper 1 2.50 2.83 paying for a newspaper 1 2.50 2.83 received a newspaper 2 2.33 3.00 going through a newspaper 1 2.00 3.00 constantly reading a newspaper 1 1.67 3.00 reading a newspaper 36 1.83 2.83 keeping a newspaper 1 1.83 2.67 paying attention to a newspaper 1 2.17 2.33 ordering a newspaper 6 1.67 2.50 having a newspaper 5 1.83 2.17 asking for a newspaper 1 1.67 2.00 skimming a newspaper 1 1.67 1.83 subscribing a newspaper 15 1.33 1.50 doing a newspaper 1 1.00 1.17 issuing a newspaper 1 1.00 1.17 keeping reading a newspaper 1 1.00 1.17 setting a newspaper 1 1.00 1.00

A1 infrequent, same light verb Does the husband have a say in his wife's property? # of responses GJ MC have a say 53 6.00 6.00 get a say 8 5.00 5.00 have a right to a say 1 3.50 4.00

274

have a right in a say 1 3.00 3.17 hold a say 4 2.50 3.50 allow to have a say 1 2.17 2.67 give a say 1 1.67 2.67 make a say 5 1.33 3.00 provide a say 1 1.33 2.83 allow a say 1 1.50 2.50 possess a say 3 1.00 3.00 take a say 2 1.50 2.50 own a say 3 1.33 2.50 qualify a say 1 1.00 2.50 have a right a say 2 1.17 2.00 have to a say 1 1.00 2.17 have an authority a say 1 1.17 1.67 keep a say 1 1.00 1.83 find a say 1 1.00 1.67 give it a say 1 1.00 1.67 say a say 1 1.00 1.67 have right a say 2 1.00 1.50 right a say 2 1.00 1.50 set right a say 1 1.00 1.50 take a speak a say 1 1.00 1.33 available a say 1 1.00 1.17

A1 infrequent, same light verb This year we have had a cut in case-management funding of $150000. # of responses GJ MC had a cut 16 6.00 6.00 taken a cut 3 5.17 5.33 faced a cut 2 5.00 5.33 experienced a cut 3 5.00 4.00 received a cut 5 4.33 4.50 undergone a cut 1 4.00 4.00 made a cut 25 3.50 3.50 got a cut 19 3.17 3.67 witnessed a cut 1 3.17 3.33 seen a cut 1 3.00 3.00 been forced a cut 1 2.50 3.00 created a cut 1 2.33 2.50 accounted a cut 1 2.00 2.50 resulted in a cut 1 1.67 2.67 proposed a cut 1 2.00 2.17 conducted a cut 1 1.83 2.17 dealt a cut 1 1.67 2.33 decided a cut 1 1.50 2.50 done a cut 4 1.67 2.33 accepted a cut 1 1.17 2.50 required a cut 1 1.33 1.67 gained a cut 2 1.17 1.67 evidenced a cut 1 1.00 1.83 been convinced of a cut 1 1.00 1.67 budget a cut 1 1.17 1.50 given a cut 2 1.00 1.67

275

happened a cut 1 1.00 1.50 let a cut 1 1.00 1.50 achieved a cut 1 1.00 1.33 been a cut 3 1.00 1.33 decreased a cut 4 1.00 1.33 increased a cut 1 1.00 1.33 become a cut 1 1.00 1.17 deleted a cut 1 1.00 1.00 dismissed a cut 1 1.00 1.00 got rid of a cut 1 1.00 1.00 prohibited a cut 1 1.00 1.00 reduced a cut 1 1.00 1.00 removed a cut 1 1.00 1.00 NO ANSWER 3 – –

A1 infrequent, same light verb If you have severe asthma, have someone else do the cleaning. # of responses GJ MC do the cleaning 54 6.00 6.00 take care of the cleaning 2 6.00 6.00 start the cleaning 1 5.67 6.00 deal with the cleaning 1 5.67 5.83 take over the cleaning 1 5.67 5.67 complete the cleaning 1 4.67 5.50 conduct the cleaning 1 3.83 4.50 try the cleaning 2 2.33 3.67 operate the cleaning 1 1.67 3.50 participate in the cleaning 1 1.67 3.50 perform the cleaning 2 1.67 3.50 take the cleaning 10 2.50 2.50 make the cleaning 18 1.17 3.33 need to the cleaning 1 1.17 3.33 achieve the cleaning 1 1.17 3.00 experience the cleaning 1 1.50 2.67 act the cleaning 1 1.00 2.50 work the cleaning 1 1.33 2.17 get the cleaning 5 1.33 1.83 have the cleaning 12 1.50 1.50 help the cleaning 2 1.33 1.50 keep the cleaning 4 1.17 1.50 clean up the cleaning 1 1.00 1.50 ask the cleaning 1 1.00 1.00 encourage the cleaning 1 1.00 1.00

A1 infrequent, same light verb The Swedes are hesitating to give clearance to build the bridge because of the strength of the opposition to the project. # of responses GJ MC give clearance 51 6.00 6.00 issue clearance 1 6.00 6.00 provide clearance 6 5.67 5.67 allow clearance 11 4.17 5.50

276

permit clearance 11 4.83 4.83 get clearance 4 4.50 4.83 supply clearance 1 3.33 4.33 be clearance 1 3.17 3.67 offer clearance 1 2.83 4.00 provide a clearance 1 3.17 3.50 admit clearance 2 3.33 3.33 give a clearance 2 3.17 3.33 give on clearance 1 3.17 3.33 give out clearance 1 3.17 3.33 pass a clearance 1 2.17 3.17 make clearance 10 1.67 2.67 pass forward clearance 1 1.83 2.50 approve clearance 2 1.67 2.33 get around clearance 1 1.83 2.17 give away clearance 1 1.50 1.83 have clearance 2 1.50 1.83 put a clearance 1 1.50 1.67 take clearance 2 1.50 1.67 make a clearance 1 1.17 1.67 accept clearance 2 1.17 1.50 assign clearance 1 1.17 1.50 declare clearance 1 1.00 1.50

A1 infrequent, same light verb Tamagotchi, electronic toy designed by Akihiro Yokoi, has become a sensation around world with orders this year of 70 million. # of responses GJ MC become a sensation 27 6.00 6.00 come to be a sensation 1 2.67 4.83 produced a sensation 2 3.33 3.67 created a sensation 6 2.83 3.67 achieved a sensation 1 2.17 3.83 been a sensation 14 2.50 3.50 caused a sensation 6 2.67 3.33 made a sensation 26 3.00 3.00 triggered a sensation 1 2.67 3.33 brought a sensation 2 2.83 3.00 marked a sensation 1 2.50 3.17 led a sensation 1 2.33 3.17 spread a sensation 1 2.67 2.50 reached a sensation 1 1.83 3.33 introduced a sensation 1 2.33 2.50 had a sensation 10 1.67 3.00 hit a sensation 1 1.17 3.17 splashed a sensation 1 2.00 2.17 moved a sensation 1 1.67 2.33 experienced a sensation 1 1.67 2.17 held a sensation 1 1.17 2.33 seen as a sensation 1 1.00 2.50 become famous a sensation 1 1.17 2.00 gained a sensation 1 1.17 1.83 big a sensation 1 1.17 1.67

277

given a sensation 6 1.17 1.67 got a sensation 13 1.17 1.50 popular a sensation 1 1.17 1.50 succeeded a sensation 1 1.17 1.50 taken a sensation 3 1.17 1.50 come about a sensation 1 1.17 1.33 NO ANSWER 1 – –

* * * * * * * * * * * * * * * *

A2 frequent, different light verb Do you wish to make an appointment with the physician? # of responses GJ MC make an appointment 64 6.00 6.00 book an appointment 1 6.00 6.00 schedule an appointment 1 6.00 6.00 arrange an appointment 4 6.00 5.83 call in for an appointment 1 4.67 5.67 get an appointment 32 4.67 5.67 ask for an appointment 2 4.33 5.33 reserve an appointment 6 4.17 5.00 create an appointment 2 3.67 4.50 have an appointment 28 3.17 4.17 take an appointment 19 3.17 3.50 ask an appointment 1 2.83 3.67 hold an appointment 3 2.83 3.33 fix an arrangement 1 2.33 3.67 receive an appointment 1 2.83 3.50 pursue an appointment 1 1.83 3.17 reserve for an appointment 1 2.17 2.17 keep an appointment 2 1.83 2.33 reservation an appointment 1 1.33 2.67 promise an appointment 1 1.67 2.17

A2 frequent, different light verb I didn't have lunch today. # of responses GJ MC have lunch 59 6.00 6.00 eat lunch 52 6.00 6.00 get lunch 27 4.33 5.33 have a lunch 1 4.33 5.17 go to lunch 2 4.00 4.83 consume lunch 2 3.50 5.17 get a lunch 1 3.50 4.83 go for lunch 1 3.00 4.67 do lunch 4 2.83 4.33 feed myself lunch 1 2.83 4.33 fix lunch 1 2.33 3.00 have time for lunch 1 1.83 2.67 take lunch 30 1.83 2.50 make lunch 2 1.50 2.33 munch lunch 1 1.00 2.83

278

receive lunch 1 1.17 2.17 hold lunch 1 1.33 2.00 want lunch 1 1.33 1.33 put lunch 1 1.00 1.33

A2 frequent, different light verb We finally made a decision this morning and purchased a Kodak digital camera. # of responses GJ MC made a decision 61 6.00 6.00 came to a decision 4 5.67 5.83 reached a decision 8 5.33 5.67 finalized a decision 1 3.17 5.17 came up with a decision 2 3.33 5.17 determined a decision 1 3.33 5.00 agreed on a decision 1 3.00 5.17 got to a decision 1 2.50 4.33 found a decision 3 2.50 3.83 came to have a decision 1 2.00 4.00 concluded with a decision 1 2.00 3.50 fixed a decision 1 2.33 3.17 had a decision 14 2.00 3.50 came a decision 1 1.50 3.67 got a decision 11 1.83 3.33 obtained a decision 1 2.00 3.00 chose a decision 1 1.50 3.33 picked a decision 1 2.00 2.67 decided a decision 9 1.17 3.33 reached to a decision 2 1.33 3.17 succeeded a decision 1 1.00 3.50 convinced myself to a decision 1 1.17 3.00 set a decision 1 1.00 2.83 took a decision 3 1.50 2.17 received a decision 1 1.17 2.50 concluded a decision 4 1.17 2.50 forced a decision 1 1.00 2.33 persuaded to a decision 1 1.00 1.33 NO ANSWER 1 – –

A2 frequent, different light verb The project is taking account of recent developments in software and standards. # of responses GJ MC taking account 16 6.00 6.00 making an account 1 3.17 3.67 taking in account 1 3.17 3.67 keeping account 2 3.00 3.83 considering account 17 3.00 3.50 taking into account 4 2.50 3.33 making account 12 2.17 3.50 taking into an account 1 2.67 3.00 taken account 1 2.17 3.00 considered account 8 1.83 3.00 including account 1 2.00 2.83

279

pondering account 1 1.67 2.83 getting account 3 1.83 2.17 taken into account 4 1.50 2.50 thinking account 6 1.17 2.67 due to the account 1 1.83 2.00 concerned about account 1 1.33 2.33 held account 1 1.50 2.00 on the account 1 1.50 2.00 accepted account 1 1.50 1.67 dealing account 1 1.00 2.00 giving account 1 1.17 1.83 decided account 1 1.50 1.50 putting account 2 1.33 1.67 made account 4 1.17 1.67 paying account 1 1.17 1.67 given an account 2 1.17 1.50 had account 1 1.17 1.33 having account 7 1.00 1.50 concerned account 2 1.17 1.17 got account 2 1.00 1.33 thought account 5 1.00 1.17 wondering account 1 1.00 1.17 NO ANSWER 3 – –

A2 frequent, different light verb You will find it convenient to make use of one of the two banks on campus. # of responses GJ MC make use 35 6.00 6.00 make a use 5 4.33 5.33 take opportunity of the use 1 3.33 3.67 take use 11 3.00 3.33 get use 5 1.83 4.17 have use 11 2.67 3.33 utilize use 1 2.33 3.67 acquire use 1 2.33 3.50 accept use 1 2.33 3.17 access use 1 1.67 3.33 take to use 1 2.33 2.67 be able to use 1 1.50 2.83 need use 1 1.83 2.33 make a most use 1 1.00 3.83 experience use 1 1.17 3.33 go use 2 1.00 2.83 be use 2 1.50 2.00 for use 1 1.00 2.17 manipulate use 2 1.17 1.67 try and use 1 1.17 1.67 to use 1 1.33 1.50 place use 1 1.00 1.67 do use 3 1.00 1.50 look use 1 1.00 1.50 how to use 1 1.00 1.17 NO ANSWER 3 – –

280

A2 frequent, different light verb They did not know that Roberts was going to give evidence in the second trial. # of responses GJ MC give evidence 11 6.00 6.00 introduce evidence 1 5.33 5.83 provide evidence 8 5.17 5.67 show evidence 19 5.00 5.33 give his evidence 1 5.00 5.00 reveal evidence 1 4.17 5.33 bring evidence 1 4.33 5.00 share evidence 1 4.17 4.67 show his evidence 1 4.17 4.50 have evidence 4 3.83 4.83 display evidence 1 3.83 4.50 report evidence 1 3.17 4.33 state evidence 4 3.33 4.17 declare evidence 2 3.17 4.17 illustrate evidence 1 3.17 3.67 explain evidence 1 2.83 3.33 testify evidence 1 2.17 3.83 mention evidence 2 2.33 3.50 prepare evidence 1 2.50 3.17 prove evidence 7 2.67 3.00 talk about evidence 2 2.67 3.00 state an evidence 1 2.50 3.00 propose evidence 1 2.17 3.00 get evidence 2 2.33 2.83 take evidence 3 2.33 2.83 support evidence 1 2.17 2.67 propose the evidence 1 2.00 2.83 indicate evidence 1 2.00 2.67 prove the evidence 1 2.00 2.17 represent evidence 1 1.50 2.67 tell evidence 10 1.17 3.00 tell the evidence 1 1.67 2.50 make clear evidence 1 1.50 2.50 bring up an evidence 1 1.67 2.17 say evidence 12 1.50 2.33 attest evidence 1 1.67 2.00 make evidence 8 1.17 2.33 speak out about the evidence 1 1.67 1.83 speak out evidence 1 1.50 1.67 speak evidence 5 1.33 1.83 stand up for evidence 1 1.33 1.83 make a show evidence 1 1.17 1.83 testimony evidence 2 1.17 1.83 talk evidence 5 1.17 1.67 speak up evidence 1 1.33 1.50 proof evidence 3 1.33 1.50 use for evidence 1 1.00 1.67 confess the evidence 1 1.17 1.33

281

make sure the evidence 1 1.00 1.33 say for evidence 1 1.00 1.17 say the evidence 1 1.00 1.17 be evidence 3 1.00 1.00 become evidence 2 1.00 1.00 go evidence 1 1.00 1.00 see evidence 1 1.00 1.00

* * * * * * * * * * * * * * * *

A2 infrequent, different light verb Joshua got a passport in 1997, hoping to travel to the UK for a year. # of responses GJ MC got a passport 57 6.00 6.00 acquired a passport 5 5.67 5.83 obtained a passport 13 5.00 5.83 was issued a passport 3 4.83 5.80 received a passport 20 4.50 5.00 was given a passport 1 3.50 5.00 picked up a passport 2 3.50 3.67 gained a passport 4 2.50 4.17 bought a passport 1 2.50 3.33 applied for a passport 1 2.00 2.50 retrieved a passport 1 2.00 2.50 had a passport 10 1.67 2.50 asked a passport 1 1.33 2.17 picked a passport 2 1.50 1.67 created a passport 2 1.67 1.17 reserved a passport 1 1.17 1.67 issued a passport 2 1.33 1.50 prepared a passport 5 1.17 1.50 took a passport 31 1.17 1.50 reached to a passport 1 1.17 1.17 made a passport 24 1.00 1.17 lost a passport 1 1.00 1.00

A2 infrequent, different light verb Her Ferrari had a breakdown recently on a California highway. # of responses GJ MC had a breakdown 28 6.00 6.00 experienced a breakdown 5 5.50 5.67 suffered a breakdown 1 4.83 5.16 got into a breakdown 1 3.17 3.50 went through a breakdown 1 2.33 3.33 went into a breakdown 1 1.67 3.00 received a breakdown 1 1.83 2.83 took a break down 4 1.50 2.50 was a breakdown 9 1.33 2.67 did a breakdown 2 2.00 1.83 got a breakdown 33 1.33 2.33 felt a breakdown 1 1.50 1.83 broke a breakdown 4 1.17 2.00 became a breakdown 12 1.33 1.83

282

caused a breakdown 3 1.17 1.33 troubled a breakdown 1 1.17 1.33 went to a breakdown 3 1.17 1.33 got accident a breakdown 1 1.17 1.17 was on a breakdown 1 1.00 1.33 happened a breakdown 1 1.00 1.17 made a breakdown 2 1.00 1.17 crushed a breakdown 1 1.00 1.00 didn’t a breakdown 1 1.00 1.00 doesn’t work a breakdown 1 1.00 1.00 was broken a breakdown 1 1.00 1.00 was put a breakdown 1 1.00 1.00 went a breakdown 3 1.00 1.00

A2 infrequent, different light verb Christians say Jesus made a sacrifice when he died on the cross. # of responses GJ MC made a sacrifice 7 6.00 6.00 gave a sacrifice 4 4.67 4.83 did a sacrifice 1 3.00 4.00 intended to be a sacrifice 1 2.67 3.17 got to be a sacrifice 1 2.67 3.00 was offered as a sacrifice 1 2.67 3.00 took a sacrifice 8 2.50 2.67 became a sacrifice 41 2.17 2.83 was a sacrifice 25 2.33 2.67 experienced a sacrifice 1 2.00 2.50 was introduced as a sacrifice 1 1.67 2.00 needed a sacrifice 1 1.67 1.67 put him as a sacrifice 1 1.17 1.50 had a sacrifice 9 1.17 1.33 got a sacrifice 11 1.17 1.17 into a sacrifice 1 1.17 1.17 used a sacrifice 1 1.00 1.17 received a sacrifice 5 1.00 1.00 sacrificed a sacrifice 1 1.00 1.00 NO ANSWER 2 – –

A2 infrequent, different light verb The rulers take counsel together in order to talk out their differences. # of responses GJ MC take counsel 12 6.00 6.00 hold counsel 5 5.50 5.50 call a counsel 1 4.33 4.50 attend counsel 1 4.17 4.50 have counsel 23 4.00 4.67 have a counsel 2 3.83 4.67 go to counsel 3 3.67 4.33 do a counsel 1 2.50 4.50 unite and counsel 1 3.00 4.00 seek counsel 2 3.33 2.50 hear counsel 1 2.67 3.00

283

gather counsel 1 3.00 2.50 make counsel 6 2.17 3.33 do counsel 4 1.50 3.67 get counsel 9 2.17 3.00 give counsel 2 2.33 2.67 see a counsel 2 2.17 2.83 ask for counsel 1 2.17 2.67 seek a counsel 1 2.17 2.50 receive counsel 3 2.17 2.17 consult counsel 3 1.83 2.17 designed counsel 1 1.83 2.17 put on counsel 1 1.83 2.00 make a counsel 6 1.50 2.33 argue counsel 1 1.33 2.17 see counsel 1 1.33 2.17 face counsel 1 1.50 2.00 go counsel 2 1.17 2.33 meet counsel 3 1.50 2.00 talk with counsel 2 1.33 2.17 sermon counsel 1 1.67 1.67 ask counsel 8 1.17 2.00 fix counsel 1 1.17 1.83 talk counsel 7 1.17 1.83 talk to counsel 3 1.17 1.83 speak counsel 2 1.17 1.67 listen counsel 1 1.00 1.50 council counsel 1 1.00 1.33 discuss counsel 1 1.00 1.17 converse counsel 1 1.00 1.00 discuss with counsel 1 1.00 1.00 tell counsel 1 1.00 1.00 NO ANSWER 1 – –

A2 infrequent, different light verb Jeffrey Taylor made a goal, winning the game for Newberry 1-0. # of responses GJ MC made a goal 42 6.00 6.00 scored a goal 11 5.17 5.67 got a goal 35 4.17 5.00 obtained a goal 3 3.67 4.67 earned a goal 3 3.50 4.33 achieved a goal 1 3.33 3.50 attained a goal 1 2.83 3.50 shot a goal 3 2.83 3.33 gained a goal 2 2.17 3.33 had a goal 13 2.17 3.17 won a goal 2 2.50 2.83 produced a goal 1 2.17 2.67 reached a goal 2 2.00 2.50 set a goal 1 1.83 2.17 caught a goal 1 1.67 2.17 did a goal 4 1.00 2.83 received a goal 2 1.33 2.00

284

owned a goal 1 1.33 1.50 pointed a goal 2 1.33 1.50 put a goal 2 1.17 1.33 took a goal 13 1.00 1.50 gave a goal 2 1.00 1.17 into a goal 1 1.00 1.17 got a point a goal 1 1.00 1.00 went a goal 1 1.00 1.00

A2 infrequent, different light verb It is impossible to predict which kid will make a discovery that will change the world. # of responses GJ MC make a discovery 30 6.00 6.00 have a discovery 17 4.67 4.83 find a discovery 32 3.83 5.00 reach a discovery 2 3.83 4.67 achieve a discovery 2 2.83 3.50 experience a discovery 2 2.83 3.33 face a discovery 2 2.33 3.00 succeed in a discovery 1 1.83 3.50 think of a discovery 1 2.00 3.17 be involved a discovery 1 2.33 2.83 discover a discovery 7 1.83 3.33 figure out a discovery 1 2.33 2.83 accomplish a discovery 2 1.67 3.33 create a discovery 1 1.67 2.67 notice a discovery 2 1.83 2.50 realize discovery 1 2.00 2.33 do a discovery 4 1.33 2.50 get a discovery 15 1.50 2.33 see a discovery 4 1.50 2.33 find out a discovery 3 1.50 2.17 manage a discovery 1 1.83 1.83 catch a discovery 1 1.33 2.17 reach to a discovery 1 1.50 2.00 give a discovery 1 1.33 1.83 confront a discovery 1 1.17 1.83 invent a discovery 1 1.17 1.67 feast a discovery 1 1.33 1.50 be enlightened by a discovery 1 1.17 1.50 search a discovery 1 1.00 1.00 success a discovery 1 1.00 1.00 take a discovery 6 1.00 1.00

* * * * * * * * * * * * * * * *

A3 frequent, content verb Skinheads, who killed an Armenian, stood trial in Moscow on July 26, 2004. # of responses GJ MC went to trial 2 5.33 5.83 went on trial 2 5.17 5.17 were taken to trial 1 5.00 5.17

285

underwent trial 1 4.33 4.67 went through trial 1 4.33 4.50 had a trial 4 3.83 4.67 were put on trial 2 4.00 4.00 were called to trial 1 4.00 3.83 went through a trial 1 3.67 4.00 went trial 1 2.33 5.17 faced the trial 1 3.17 4.17 had trial 24 3.33 4.00 stood in trial 1 3.50 3.83 were put in trial 1 3.17 3.83 attended trial 1 2.67 3.83 received a trial 1 2.50 3.67 were in trial 3 2.67 3.50 were called for trial 2 2.83 3.17 were put into trial 1 2.67 3.33 were in the trial 1 2.67 3.17 received trial 14 2.50 3.33 were judged trial 2 2.83 2.83 were on a trial 1 2.50 3.00 were put on the trial 1 2.33 3.00 attended the trial 1 2.33 2.67 got a trial 2 2.00 3.00 were in a trial 1 2.17 2.67 participated in a trial 1 1.50 2.83 were courted to trial 1 2.00 2.33 were held trial 1 1.50 2.83 were placed to trial 1 2.00 2.33 were put trial 2 2.00 2.33 were taken trial 3 2.00 2.33 accepted trial 2 1.50 2.67 experienced trial 1 1.67 2.50 attended a trial 1 1.83 2.17 suffered trial 1 1.50 2.50 undertook trial 1 1.50 2.50 were accused on the trial 1 1.67 2.33 were at the trial 1 1.67 2.33 did trial 2 1.33 2.50 got trial 13 1.33 2.50 were sued trial 2 1.50 1.50 took trial 23 1.17 1.50 met trial 1 1.33 1.50 got sued trial 1 1.00 1.50 joined trial 1 1.17 1.17 showed up trial 1 1.00 1.17 took a trial 1 1.00 1.17 gave trial 1 1.00 1.00 judged trial 1 1.00 1.00 made trial 1 1.00 1.00 prosecuted trial 1 1.00 1.00

A3 frequent, content verb In Manila seaweed farmers fell victim to a strong peso.

286

# of responses GJ MC became victim 42 4.83 5.67 were victim 17 4.50 4.67 became a victim 6 4.17 4.83 became the victim 1 3.50 4.50 were a victim 2 3.67 4.00 were the victim 2 3.67 4.00 went victim 1 4.50 3.00 suffered victim 1 2.67 3.67 turned victim 1 3.00 3.33 turned into a victim 1 2.33 3.17 were seen as a victim 1 2.17 2.83 took victim 5 2.33 2.50 got victim 10 2.00 2.67 were made into victim 1 2.17 2.33 turned to be a victim 1 1.67 2.67 turned into victim 1 1.83 2.50 were subject to victim 1 1.83 2.33 were suffered as victim 1 1.83 2.33 got to be victim 2 1.67 2.33 fell into victim 1 1.33 2.33 were given victim 1 1.67 1.83 forced to be victim 1 1.17 2.00 experienced victim 1 1.17 1.50 grew to be victim 1 1.17 1.50 had victim 2 1.17 1.50 harmed victim 1 1.17 1.50 received victim 2 1.17 1.50 succumbed to a victim 1 1.17 1.50 fell to a victim 1 1.17 1.33 gained victim 1 1.17 1.33 made victim 1 1.00 1.50 did victim 1 1.17 1.17 failed victim 1 1.17 1.17 sacrificed themselves as victim 1 1.00 1.33 sacrificed victim 3 1.00 1.17 NO ANSWER 3 – –

A3 frequent, content verb She has never tried to commit suicide, but she constantly thinks about it. # of responses GJ MC commit suicide 43 6.00 6.00 attempt suicide 2 5.17 6.00 carry out suicide 1 5.00 5.17 commit a suicide 4 5.00 5.17 execute suicide 1 4.00 5.17 undergo suicide 1 3.67 4.33 engage in suicide 1 3.33 4.33 try suicide 4 3.83 3.83 do suicide 24 3.67 3.83 do the suicide 1 2.67 3.83 conduct suicide 1 1.83 4.33

287

practice suicide 1 2.67 3.00 do a suicide 1 1.67 3.67 experience suicide 1 2.17 3.00 have a suicide 1 1.67 3.50 reach suicide 1 2.17 2.83 have suicide 5 1.33 2.67 force suicide 1 1.33 2.17 deal with suicide 1 1.83 1.33 go into suicide 1 1.50 2.00 decide suicide 2 1.50 1.83 die suicide 1 1.17 1.83 be killed suicide 1 1.00 1.83 get suicide 8 1.50 1.33 take suicide 7 1.17 1.50 be suicide 2 1.00 1.50 kill by herself suicide 1 1.00 1.33 make suicide 6 1.00 1.33 kill suicide 4 1.00 1.00 suicide suicide 1 1.00 1.00

A3 frequent, content verb That is why asking questions about what happened on Haifa Street is very important. # of responses GJ MC asking questions 60 6.00 6.00 bringing up questions 1 5.17 5.67 raising questions 7 5.00 5.83 posing questions 1 4.67 4.83 coming up with questions 1 3.83 4.00 having questions 21 3.50 3.83 requesting questions 1 5.00 2.17 discussing questions 1 2.83 3.50 getting questions 1 2.67 3.00 illustrating questions 1 2.50 2.83 inquiring questions 2 2.00 3.17 these questions 1 2.33 2.83 asking for questions 1 2.00 3.00 pondering questions 1 2.17 2.83 coming up questions 1 2.17 2.50 bringing questions 2 1.83 2.33 carrying questions 1 1.83 2.17 calling questions 1 1.67 2.17 media questions 1 1.67 1.83 saying questions 2 1.83 2.83 the questions 1 2.00 2.50 giving questions 5 2.17 2.17 taking questions 2 2.00 2.33 making questions 18 1.50 2.17 requiring to ask questions 1 1.17 2.33 rising questions 1 1.67 1.67 community questions 1 1.17 2.00 telling questions 2 1.17 1.33 tossing questions 1 1.17 1.17 doing questions 1 1.00 1.33

288

questioning questions 2 1.00 1.33 don’t you questions 1 1.00 1.00 people questions 1 1.00 1.00 we questions 1 1.00 1.00

A3 frequent, content verb Can you keep an eye on my suitcase while I am away? # of responses GJ MC keep an eye 44 6.00 6.00 leave an eye 2 3.50 4.33 put an eye 12 3.50 4.17 place an eye 4 3.17 4.17 hold an eye 3 2.67 3.67 have an eye 9 2.67 3.33 watch with an eye 1 2.17 3.33 check an eye 1 1.50 2.33 gaze an eye 1 1.17 2.50 keep on an eye 1 2.17 3.67 keep to an eye 1 2.67 3.17 keep up an eye 1 1.83 3.33 look an eye 6 1.17 2.33 look at an eye 1 1.33 2.17 make an eye 1 1.33 2.00 remain an eye 1 1.33 2.33 see an eye 7 1.17 2.17 watch an eye 16 1.33 2.33 get an eye 2 1.00 2.17 pay an eye 1 1.17 1.83 around an eye 1 1.00 1.83 give an eye 4 1.00 1.83 stay an eye 3 1.00 1.83 do an eye 1 1.00 1.67 take an eye 5 1.00 1.67 take care an eye 1 1.00 1.50 protect an eye 1 1.17 1.33 use an eye 1 1.17 1.33 send an eye 1 1.00 1.17 stare an eye 1 1.00 1.17

A3 frequent, content verb Legislators are trying to put a stop to online gambling. # of responses GJ MC put a stop 5 6.00 6.00 bring a stop 1 4.33 5.00 call for a stop 1 3.17 4.33 pass the bill for a stop 1 2.50 4.67 implement a stop 1 3.33 3.67 impose a stop 1 2.83 4.17 place a stop 2 2.50 4.00 force a stop 1 2.33 4.00 get a stop 8 2.67 3.83 set a stop 1 2.17 3.33

289

provide a stop 1 2.00 3.33 have a stop 10 1.33 3.83 produce a stop 1 1.33 3.83 carry out a stop 1 1.67 3.33 give a stop 4 2.33 2.33 do a stop 6 1.17 3.33 make a stop 46 1.50 2.83 suggest a stop 1 2.00 2.33 build a stop 1 1.17 2.83 exercise a stop 1 1.00 3.00 give it a stop 1 2.00 1.83 reach a stop 1 1.50 2.33 conduct a stop 1 1.00 2.67 allow a stop 2 1.00 2.33 keep a stop 2 1.83 1.50 face a stop 1 1.00 2.00 come to a stop 1 1.00 1.83 catch a stop 1 1.00 1.50 drop a stop 1 1.00 1.50 go to a stop 1 1.00 1.50 prevent a stop 1 1.50 1.00 set to a stop 1 1.00 1.50 take a stop 9 1.17 1.33 let a stop 1 1.00 1.33 break a stop 1 1.00 1.17 finish a stop 1 1.00 1.00 no work to a stop 1 1.00 1.00 prohibit a stop 1 1.00 1.00 rest a stop 1 1.00 1.00 stay a stop 1 1.00 1.00 stop a stop 1 1.00 1.00 NO ANSWER 2 – –

* * * * * * * * * * * * * * * *

A3 infrequent, content verb Nearly 8.8 million eligible Canadians (39.1 per cent) didn't cast a ballot in the 2004 election. # of responses GJ MC cast a ballot 3 6.00 6.00 place a ballot 1 4.33 5.17 turn in a ballot 1 4.50 5.00 enter a ballot 1 3.67 4.33 put in a ballot 1 3.33 4.67 send in a ballot 1 3.17 4.67 do a ballot 2 3.17 4.00 give a ballot 9 3.17 3.33 give away their opinions in a ballot 1 2.83 3.67 fill a ballot 1 2.83 3.33 post a ballot 1 2.67 3.17 make a ballot 8 1.50 2.83 write a ballot 1 1.67 2.50 drop in a ballot 1 1.83 2.33 put in opinion in a ballot 1 1.83 2.33

290

put a ballot 6 1.50 2.00 take a ballot 12 1.33 2.17 turn a ballot 1 1.33 2.17 pull a ballot 1 1.17 2.00 elect a ballot 7 1.50 1.50 have a ballot 8 1.50 1.50 throw a ballot 4 1.00 2.00 vote a ballot 36 1.00 2.00 boat a ballot 2 1.17 1.67 select a ballot 1 1.50 1.33 get a ballot 4 1.17 1.50 vote for a ballot 2 1.00 1.67 raise a ballot 1 1.17 1.33 go a ballot 3 1.00 1.17 go for a ballot 1 1.00 1.17 care a ballot 1 1.00 1.00 go to a ballot 3 1.00 1.00 join a ballot 1 1.00 1.00 make a vote a ballot 1 1.00 1.00 participate a ballot 2 1.00 1.00

A3 infrequent, content verb A factory manager must keep a balance between cost to produce quality of product and service to the customer. # of responses GJ MC keep a balance 43 6.00 6.00 maintain a balance 8 5.80 5.83 achieve a balance 2 4.67 5.00 keep up a balance 1 4.50 4.67 manage a balance 4 4.33 4.83 have a balance 23 4.17 4.67 secure a balance 1 4.00 4.50 hold a balance 2 3.83 4.50 obtain a balance 1 3.83 4.17 create a balance 2 3.33 3.83 control a balance 5 3.00 3.67 conserve a balance 1 3.00 3.17 reach a balance 1 2.67 3.33 get a balance 4 2.67 2.83 support a balance 1 2.17 3.33 do a balance 2 1.50 2.50 make a balance 13 1.83 3.17 see a balance 2 2.17 2.83 adjust a balance 1 2.33 2.50 take care of a balance 1 1.83 2.50 stay a balance 2 1.50 2.33 communicate a balance 1 1.83 1.83 be a balance 3 1.50 2.00 need a balance 1 1.50 2.00 perform a balance 1 1.50 2.00 remain a balance 1 1.17 2.17 save a balance 1 1.17 2.17 be careful about a balance 1 1.50 1.67

291

succeed a balance 1 1.00 2.17 put a balance 1 1.17 1.83 consider a balance 4 2.50 2.67 watch a balance 1 1.17 1.50 take a balance 22 1.00 1.33

A3 infrequent, content verb When possible, it is best to try to remove a stain before it dries. # of responses GJ MC remove a stain 33 6.00 6.00 get rid of a stain 17 5.17 5.67 clean a stain 5 5.17 5.50 clean off a stain 1 5.17 5.33 get out a stain 2 5.17 5.17 wash off a stain 3 4.83 5.17 eliminate a stain 4 4.50 5.17 get off a stain 4 4.17 4.67 erase a stain 7 3.67 4.83 take off a stain 20 3.67 4.83 wash a stain 5 3.83 4.50 take out a stain 4 3.83 4.33 bring out a stain 1 3.17 4.00 take care of a stain 1 2.67 3.83 reduce a stain 2 2.67 3.50 wipe a stain 1 2.67 3.33 take away a stain 8 2.67 3.17 delete a stain 2 2.00 3.50 destroy a stain 1 1.83 3.67 soak out a stain 1 1.83 3.33 lose a stain 1 1.83 2.83 decrease a stain 1 1.50 3.17 come out a stain 1 2.17 2.17 put off a stain 5 2.00 2.33 off a stain 1 1.50 2.67 absorb a stain 1 1.83 2.00 whiten a stain 1 1.50 2.33 prevent a stain 1 1.83 1.83 release a stain 1 1.33 2.33 drop a stain 2 1.17 2.00 pick a stain 1 1.00 1.83 get a stain 2 1.17 1.50 recover a stain 1 1.00 1.33 leave a stain 1 1.00 1.17 take a stain 3 1.00 1.17 reveal a stain 1 1.00 1.00

A3 infrequent, content verb Farmers use the latest technology to grow rice. # of responses GJ MC grow rice 24 6.00 6.00 farm rice 3 5.67 5.83 cultivate rice 7 5.33 5.67

292

harvest rice 7 4.67 5.00 crop rice 1 4.17 4.50 generate rice 1 3.83 5.50 produce rice 20 4.33 4.83 create rice 3 3.00 4.17 manufacture rice 3 3.00 3.83 plant rice 2 2.67 3.67 cultivate for rice 1 2.50 3.50 make rice 54 2.17 3.83 raise rice 7 2.00 3.67 have rice 1 1.50 3.50 grow up rice 1 1.67 3.17 prepare rice 3 2.00 2.50 form rice 1 1.17 2.17 supply rice 1 1.67 1.83 design rice 1 1.17 1.83 gain rice 1 1.17 1.83 invent rice 1 1.17 1.67 product rice 3 1.00 1.17 cook rice 56 1.00 1.00 try rice 1 1.00 1.00 vase rice 1 1.00 1.00

A3 infrequent, content verb Health care providers put emphasis on customer service. # of responses GJ MC put emphasis 27 6.00 6.00 put an emphasis 2 5.50 5.67 place emphasis 4 5.17 5.67 focus emphasis 1 3.67 5.17 have emphasis 15 3.17 4.17 push emphasis 1 3.33 4.00 give emphasis 3 3.00 4.00 maintain emphasis 1 2.83 3.50 concentrate emphasis 1 2.67 3.50 keep emphasis 7 2.33 3.67 mark emphasis 1 2.33 3.67 give an emphasis 2 2.33 3.50 make emphasis 16 2.00 3.67 force emphasis 1 2.17 3.17 focus on emphasis 1 1.67 3.33 emphasize emphasis 1 1.67 3.17 keep on emphasis 1 2.17 2.33 announce emphasis 1 2.00 2.33 put on emphasis 3 1.83 2.17 do emphasis 4 1.00 2.67 pay emphasis 1 1.50 2.17 get emphasis 3 1.33 2.33 important emphasis 1 1.33 2.33 take emphasis 3 1.67 1.83 think about emphasis 1 1.00 2.50 see emphasis 2 1.17 2.17 consider emphasis 4 1.33 2.00

293

take serious emphasis 1 1.00 2.17 think emphasis 4 1.00 2.17 concern emphasis 1 1.00 2.00 force to emphasis 1 1.33 1.50 take care of emphasis 1 1.00 1.83 watch emphasis 1 1.00 1.83 need emphasis 1 1.17 1.50 say emphasis 1 1.00 1.67 look at emphasis 1 1.17 1.33 look emphasis 2 1.17 1.33 strong emphasis 1 1.17 1.33 care emphasis 1 1.00 1.00

A3 infrequent, content verb Mary’s dream is to found a school for young children and to develop a curriculum that blends art and science. # of responses GJ MC found a school 11 6.00 6.00 establish a school 22 5.67 5.67 set up a school 1 4.50 5.00 start a school 2 5.50 5.50 open a school 4 4.67 5.17 create a school 11 4.67 4.67 organize a school 1 4.33 4.83 build a school 41 4.17 4.83 make a school 49 4.17 4.83 develop a school 1 4.33 4.50 design a school 1 3.67 4.00 institute a school 1 3.17 4.17 produce a school 3 2.50 4.33 construct a school 7 3.50 3.83 run a school 2 3.17 3.83 build up a school 1 2.50 3.50 have a school 4 2.67 3.00 foundation a school 1 2.00 3.00 own a school 1 1.83 3.00 find a school 1 1.67 2.33 get a school 2 1.83 2.17 consulate a school 1 1.83 1.83

294

Appendix VI: Tables from Experiment 1

Appendix VI-1: Grammaticality Judgments. Experiment 1. (Figures 4-1, 4-2, 4-3 and 4-4)

Type of collocations Low-intermediate Advanced (# = 31) Total (# = 66) High (4.5-6) (#= 35) Mid (2.67-4.33) Token Percentag Token Percentag Token Percentag Low (1-2.5) e e e A1 Frequent same High 254 60.62% 245 59.76% 499 60.19% light V Mid 9 2.15% 37 9.02% 46 5.59% Low 156 37.23% 128 31.22% 284 34.23% Infrequent High 143 38.44% 147 36.93% 290 37.69% Mid 49 13.17% 77 19.35% 126 16.26% Low 180 48.39% 174 43.72% 354 46.06% A2 Frequent differe High 194 42.92% 182 45.27% 376 44.10% nt light Mid 87 19.25% 89 22.14% 176 20.70% V Low 171 37.83% 131 32.59% 302 35.21% Infrequent High 114 24.68% 156 37.32% 270 31.00% Mid 68 14.72% 56 13.40% 124 14.06% Low 280 60.61% 206 49.28% 486 54.95% A3 Frequent Conten High 112 26.73% 123 33.61% 235 30.17% t Mid 77 18.38% 68 18.58% 145 18.48% V Low 230 54.90% 175 47.81% 405 51.36% Infrequent High 99 21.85% 153 33.33% 114 27.59% Mid 130 28.69% 146 31.80% 276 30.25% Low 224 49.45% 160 34.86% 384 42.16% Total (# = 66) High 916 35.55% 1006 41.01% 1922 38.28% Mid 420 16.30% 473 19.28% 893 17.79% Low 1241 48.16% 974 39.71% 2215 43.94% TABLE VI-1. Grammaticality Judgments. Experiment 1 (Figures 4-1, 4-2, 4-3 and 4-4)

295

Appendix VI-2: Grammaticality Judgments. Experiment 1 (Figures 4-5, 4-6, 4-7 and 4-8)

Type of Low-intermediate Advanced (# = 31) Total (# = 66) collocations (# = 35) High (4.5-6) Token Percentag Token Percentag Token Percentag Mid (2.67-4.33) e e e Low (1-2.5) 1. Frequent Frequen High 560 43.41% 550 46.69% 1110 45.05% cy Mid 173 13.41% 194 16.47% 367 14.94% Low 557 43.18% 434 36.84% 991 40.01% Infreq High 356 27.66% 456 35.76% 812 31.71% Mid 247 19.20% 279 21.88% 526 20.54% Low 684 53.15% 540 42.35% 1224 47.75% 2. With equi With vs. High 397 50.19% 392 48.51% 789 49.35% without Mid 58 7.33% 114 14.11% 172 10.72% L1 Low 336 42.48% 302 37.38% 638 39.93% equivale Without nt High 519 29.06% 614 37.33% 1133 33.20% Mid 362 20.27% 359 21.82% 721 21.05% Low 905 50.67% 672 40.85% 1577 45.76% 3. Light V Light High 705 41.35% 276 44.84% 1435 43.10% V+N Mid 213 12.49% 214 15.91% 472 14.20% vs. Low 787 46.16% 335 39.25% 1426 42.71% Content Content V V+N High 211 24.20% 276 33.45% 487 28.83% Mid 207 23.74% 214 25.94% 421 24.84% Low 454 52.06% 335 40.61% 789 46.34% 4. Same ligh Same High 397 50.19% 392 48.51% 789 49.35% light V Mid 58 7.33% 114 14.11% 172 10.72% vs. Low 336 42.48% 302 37.38% 638 39.93% Differen Different t light V High 308 33.70% 338 49.35% 646 41.53% Mid 155 16.96% 137 10.72% 292 13.84% Low 451 49.34% 335 39.93% 786 44.64% TABLE VI-2. Grammaticality Judgments. Experiment 1 (Figures 4-5, 4-6, 4-7 and 4-8)

296

Appendix VI-3. Meaning Conveyed. Experiment 1 (Figures 4-9, 4-10, 4-11 and 4-12)

Type of Low-intermediate Advanced (# = 31) Total (# = 66) collocations (#= 35) High (4.5-6) Token Percentag Token Percentag Token Percentag Mid (2.67-4.33) e e e Low (1-2.5) A1 Frequent same High 260 62.35% 279 68.22% 539 65.29% light V Mid 81 19.42% 70 17.11% 151 18.27% Low 76 18.23% 60 14.67% 136 16.45% Infrequent High 150 40.11% 153 38.44% 303 39.28% Mid 127 33.96% 151 37.94% 278 35.95% Low 97 25.94% 94 23.62% 191 24.78% A2 Frequent differe High 227 49.56% 223 55.06% 450 52.31% nt light Mid 141 30.79% 112 27.65% 253 29.22% V Low 90 19.65% 70 17.28% 160 18.47% Infrequent High 181 38.43% 197 47.58% 378 43.01% Mid 91 19.32% 91 21.98% 182 20.65% Low 199 42.25% 126 30.43% 325 36.34% A3 Frequent Conten High 113 27.29% 132 35.11% 245 31.20% t Mid 144 34.78% 134 34.57% 278 34.68% V Low 157 37.92% 114 30.32% 271 34.12% Infrequent High 195 43.82% 246 54.07% 441 48.95% Mid 115 25.84% 126 26.37% 241 26.11% Low 135 30.34% 89 19.56% 224 24.95% Total (# = 66) High 1126 43.66% 1230 49.86% 2356 46.76% Mid 699 27.10% 684 27.73% 1383 27.42% Low 754 29.24% 553 22.42% 1307 25.83% TABLE VI-3. Meaning Conveyed. Experiment 1 (Figures 4-9, 4-10, 4-11 and 4-12)

297

Appendix VI-4. Meaning Conveyed. Experiment 1 (Figures 4-13, 4-14, 4-15, and 4-16)

Type of collocations Low-intermediate Advanced (# = 31) Total (# = 66) High (4.5-6) (# = 35) Mid (2.67-4.33) Token Percentag Token Percentag Token Percentag Low (1-2.5) e e e 1. Frequent Frequen High 600 46.55% 634 53.10% 1234 49.83% cy Mid 366 28.39% 316 26.47% 682 27.43% Low 323 25.06% 244 20.44% 567 22.75% Infreque High 526 40.68% 596 47.04% 1122 43.86% Mid 333 25.85% 368 28.57% 701 27.21% Low 431 33.46% 309 24.39% 740 28.93% 2. With equi With vs. High 410 51.83% 432 53.53% 842 52.68% without Mid 208 26.30% 221 27.39% 429 26.85% L1 Low 173 21.87% 154 19.08% 327 20.48% equivale Without nt High 716 39.98% 798 48.25% 1514 44.12% Mid 491 27.49% 463 27.63% 954 27.56% Low 581 32.53% 399 32.53% 980 32.53% 3. Light V Light High 818 47.56% 852 52.40% 1670 49.98% V+N Mid 440 25.58% 424 26.08% 864 25.83% vs. Low 462 26.86% 350 21.53% 812 24.20% Content Content V V+N High 308 35.71% 378 45.27% 686 40.49% Mid 259 30.22% 260 30.42% 519 30.32% Low 292 34.07% 203 24.31% 495 29.19% 4. Same Same light 410 51.83% 432 53.53% 842 52.68% light V High 208 26.30% 221 27.39% 429 26.85% vs. Mid 173 21.87% 154 19.08% 327 20.48% Differen Low t light V Different High 408 43.92% 420 51.28% 828 47.60% Mid 232 24.97% 203 24.79% 435 24.88% Low 289 31.11% 196 23.93% 485 27.52% TABLE VI-4. Meaning Conveyed. Experiment 1 (Figures 4-13, 4-14, 4-15, and 4-16)

298

Appendix VI-5: Grammaticality Judgments by native speakers vs. learners’ confidence in their productions. Experiments 1 (Figures 4-17 and 4-18)

Type of collocations Low-intermediate learners Advanced learners

GJ by natives (# = 35) (# = 31) Grammatical (4.5-6) Confide Not Total Confide Not Total Ungrammatical (1-4.33) nt Confide nt Confide (7-10) nt (1-6) (7-10) nt (1-6) A1 Frequent same Grammatical 190 64 254 212 33 245 light 45.35% 15.27% 60.62% 51.71% 8.05% 59.76% verb Ungrammatic 73 92 165 107 58 165 al 17.42% 21.96% 39.38% 26.10% 14.14% 40.24% Infrequent Grammatical 62 81 143 92 55 147 16.67% 21.77% 38.44% 23.12% 13.82% 36.93% Ungrammatic 80 149 229 130 121 251 al 21.51% 40.05% 61.56% 32.66% 30.40% 63.07% A2 Frequent differen Grammatical 154 40 194 161 21 182 t light 34.07% 8.85% 42.92% 40.05% 5.22% 45.27% verb Ungrammatic 102 156 258 122 98 220 al 22.57% 34.51% 57.08% 30.35% 24.38% 54.73% Infrequent Grammatical 68 46 114 123 33 156 14.72% 9.96% 24.68% 29.43% 7.89% 37.32% Ungrammatic 156 192 348 131 131 262 al 33.77% 41.56% 75.32% 31.34% 31.34% 62.68% A3 Frequent content Grammatical 79 33 112 114 9 123 verb Ungrammatic 18.85% 7.88% 26.73% 31.15% 2.46% 33.61% al 125 182 307 145 98 243 29.83% 43.44% 73.27% 39.62% 26.78% 66.39% Infrequent Grammatical 46 53 99 124 29 153 10.15% 11.70% 21.85% 27.02% 6.31% 33.33% Ungrammatic 164 190 354 198 108 306 al 36.20% 41.95% 78.15% 43.14% 23.53% 66.67% Total Grammatical 599 317 916 826 180 1006 (# = 66) Ungrammatic 23.42% 13.02% 35.55% 33.67% 7.34% 41.01% al 700 961 1661 833 614 1447 26.83% 36.90% 64.45% 33.69% 25.03% 58.99% TABLE VI-5. Grammaticality Judgments by native speakers vs. learners’ confidence in their productions. Experiment 1 (Figures 4-17 and 4-18)

299

Appendix VI-6: Meaning Conveyed rated by native speakers vs. learners’ confidence in their productions. Experiment 1 (Figures 4-19 and 4-20)

Type of collocations Low-intermediate learners Advanced learners (# = 35) (# = 31) MC rated by natives Confide Not Total Confiden Not Total High (4.5-6) nt confident t confident Low/Mid (1-4.33) (7-10) (1-6) (7-10) (1-6) A1 Frequent same Meaning 190 70 260 236 43 279 light conveyed 45.56% 16.79% 62.35% 57.70% 10.52% 68.22%

verb Meaning NOT 73 84 157 83 47 130 conveyed 8.75% 10.075% 37.65% 20.29% 11.49% 31.78% Infrequent Meaning 64 86 150 100 53 153 conveyed 17.11% 23.00% 40.11% 25.12% 13.32% 38.44%

Meaning NOT 79 145 224 125 120 245 conveyed 21.13% 39.77% 59.89% 31.41% 30.15% 61.56% A2 Frequent different Meaning 171 56 227 193 30 223 light conveyed 37.34% 12.22% 49.56% 47.65% 7.41% 55.06%

verb Meaning NOT 86 145 231 93 89 182 conveyed 18.78% 31.66% 50.44% 22.97% 21.96% 44.94% Infrequent Meaning 97 84 181 143 54 197 conveyed 20.59% 17.84% 38.43% 34.54% 13.04% 47.58%

Meaning NOT 127 163 290 112 135 217 conveyed 26.97% 34.6% 61.57% 27.06% 25.35% 52.42% A3 Frequent content Meaning 77 36 113 122 10 132 verb conveyed 18.60% 8.70% 27.29% 32.11% 2.63% 34.74%

Meaning NOT 132 169 301 133 115 248 conveyed 31.88% 40.82% 72.71% 35.00% 30.26% 65.26% Infrequent Meaning 128 67 195 201 45 246 conveyed 28.76% 15.06% 43.82% 43.60% 9.76% 53.36%

Meaning NOT 97 153 250 125 90 215 conveyed 21.80% 34.38% 56.18% 27.11% 19.52% 46.64% Total Meaning 727 399 1126 995 235 1230 conveyed 28.19% 15.47% 43.66% 39.85% 9.41% 49.26%

Meaning NOT 594 859 1453 671 596 1267

conveyed 23.03% 33.31% 56.34% 26.87% 23.87% 50.74%

300

TABLE VI-6. Meaning Conveyed rated by native speakers vs. learners’ confidence in their productions. Experiment 1 (Figures 4-19 and 4-20)

Appendix VI-7: Learners’ confidence, familiarity, and differences between confidence and familiarity. Experiment 1 (Figures 4-21, 4-22, 4-23, 4-24, 4-25, 4-26)

Low-intermediate Advanced Total Conf Fa Diff Conf Fa Diff Conf Fa Diff m m m Frequenc Frequent 6.64 6.13 0.51 7.95 7.38 0.57 7.30 6.76 0.54 y Infrequent 5.96 5.20 0.76 7.02 6.42 0.60 6.49 5.81 0.68

L1 With L1 6.21 5.51 0.70 7.41 6.89 0.52 6.81 6.20 0.61 equivalen Without 6.35 5.77 0.58 7.53 6.90 0.63 6.94 6.34 0.60 t L1 Light vs. Light V 6.27 5.62 0.65 7.36 6.83 0.53 6.82 6.23 0.59 content V Content V 6.37 5.81 0.56 7.74 7.04 0.70 7.06 6.43 0.63 Same vs. Same 6.21 5.51 0.70 7.41 6.89 0.52 6.81 6.20 0.61 different Different 6.34 5.73 0.61 7.31 6.77 0.54 6.83 6.25 0.58 light V Easy types 6.33 5.69 0.64 7.53 6.99 0.54 6.93 6.34 0.59 Total Difficult 6.26 5.63 0.63 7.40 6.78 0.62 6.83 6.20 0.63 Total 6.30 5.66 0.64 7.47 6.88 0.59 6.88 6.26 0.62 TABLE VI-7. Learners’ confidence, familiarity, and differences between confidence and familiarity. (Figures 4-21, 4-22, 4-23, 4-24, 4-25, 4-26)

301

Appendix VII: Test sentences for Experiment 2

A1 literal, same light verb • 「プロジェクトをした」 Recently Ken did a project on Australian aboriginal education.

• 「招待を受けた」 Juliet received an invitation to attend a Middle Eastern Studies conference at Harvard.

• 「先生になること」 Becoming a teacher is a life-long learning process.

• 「授業があった」 At eight Stella had a class in General Biology.

• 「チケットを得た」 Bob got a ticket for the concert tonight.

A1 abstract, same light verb • 「承諾を得る」 We need to get the nod from five of the seven members selected from the committee.

• 「チャンスを得た」 Albuquerque’s low-wage workers got a break. They finally got a boost in wages.

• 「雰囲気がある」 Old theaters in Ireland always have an air of nostalgia.

• 「才能があった」 Johnny Mercer had an ear for country music and was able to pick out any tune he wanted by ear.

• 「態度・立場をとっている」 Churches are taking a stand against the spread of HIV.

A2 literal, different light verb • 「散歩をした」 Campbell took a walk on the beach last Saturday.

• 「試験を受けること」 If you are thinking of taking an exam, make sure that you have enough time for home study.

• 「議論する」

302

I don't want to have an argument with you or anyone else about this.

• 「コピーをとった」 Jean made a copy of the article and gave it to her mother.

• 「入浴する」 Do not take a bath when suffering from cold fever.

A2 abstract, different light verb • 「馴れ馴れしくした」 “A Beautiful Mind” scriptwriters took liberties with the facts in Nasar’s biography.

• 「急落した」 The temperature took a dive last night.

• 「大儲けした」 David has made a killing in Network Marketing.

• 「評判を得た」 The cars manufactured by Saab Automobiles (a company name) had a name for safety and crash-resistance.

• 「首になった」 The shop assistant got the sack because he swore at/ to a customer.

A3 literal, content verb • 「手術を受ける」 Sharon was scheduled to undergo an operation on Friday morning.

• 「注文した」 Have you ever placed an order online with your credit card?

• 「被害を与えた」 The tornado caused damage to 43 homes before finally dissipating to Columbia.

• 「ガンになる」 This scientific discovery could save 3 to 4 million people a year from developing cancer.

• 「注文を受けた」 Johnny Rivers accepted an offer to do a concert for the reopening of the Mall of Memphis.

A3 abstract, content verb • 「協定する」 It is unclear whether Russian officials will strike a bargain with the Government of Lithuania.

• 「機嫌を取った」 Politicians of every stripe -- from ward heelers to the President of the United States – paid court to the archbishop of New York.

303

• 「目撃した」 On September 13th we bore witness to an event that should serve as a turning point in the history of the Middle East.

• 「近道した」 Joan cut corners to get there on time.

• 「結婚した」 Tanya Haden and her longtime boyfriend, Carey Hart, tied the knot over the weekend in Las Vegas.

304

Appendix VIII: Data from Experiment 2

# of responses = number of responses produced by subjects (n = 66) GJ = Native speakers’ Grammaticality Judgments (6-point-scale) MC = Native speakers’ Meaning Conveyed (6-point-scale)

The combinations are listed from the highest to the lowest average scores of the GJ and the MC.

A1 literal, same light verb Recently Ken did a project on Australian aboriginal education. # of responses GJ MC did a project 42 6.00 6.00 completed a project 4 5.33 5.33 finished a project 7 5.17 5.17 carried out a project 1 5.33 4.83 conducted a project 4 4.83 4.83 accomplished a project 4 4.33 5.17 made a project 29 4.00 4.50 produced a project 3 4.00 4.50 had a project 26 3.67 4.00 created a project 3 3.50 4.00 managed a project 1 3.67 3.83 organized a project 1 3.83 3.83 worked a project 2 3.33 4.00 worked on a project 6 3.33 3.83 performed a project 1 3.33 3.67 handled a project 1 2.83 3.83 was in a project 2 3.00 3.50 began a project 1 2.50 3.83 ran a project 3 2.67 3.33 presented a project 4 2.67 3.17 underwent a project 1 2.50 3.33 started a project 3 2.50 3.17 planned a project 1 2.83 2.83 initiated a project 1 2.67 2.83 experienced a project 1 2.67 2.67 went through a project 1 2.17 2.83 studied a project 2 2.17 2.67 gave a project 2 2.33 2.50 held a project 12 2.00 2.83 implemented a project 1 2.17 2.50 got a project 3 2.00 2.50 took a project 4 2.00 2.50 attended a project 1 1.50 2.50 supported a project 1 2.00 2.00 tried a project 1 1.17 2.50 promote a project 1 1.33 2.00

A1 literal, same light verb Juliet received an invitation to attend a Middle Eastern Studies conference at Harvard. # of responses GJ MC

305

received an invitation 50 6.00 6.00 got an invitation 55 5.83 6.00 was given an invitation 1 5.83 5.83 was mailed an invitation 1 4.17 5.00 was handed an invitation 1 3.67 4.67 was sent an invitation 1 3.33 5.00 accepted an invitation 3 4.17 4.00 gained an invitation 1 2.83 3.67 took an invitation 3 2.83 3.17 was offered an invitation 1 3.00 2.83 had an invitation 22 1.83 2.50 was taken an invitation 1 2.00 2.33 held an invitation 1 1.50 1.83 experienced an invitation 1 1.00 2.00 provided an invitation 1 1.17 1.50 asked an invitation 2 1.00 1.17 gave an invitation 1 1.00 1.17

A1 literal, same light verb Becoming a teacher is a life-long learning process. # of responses GJ MC Becoming a teacher 67 6.00 6.00 Being a teacher 52 3.67 3.83 Process to become a teacher 1 2.50 3.83 Working as a teacher 2 2.67 3.33 Turning out to be a teacher 1 1.83 3.33 Teaching as a teacher 1 1.33 3.33 Changing to a teacher 1 1.17 2.50 Turning out a teacher 1 1.33 2.00 Turning to a teacher 1 1.33 1.67 Coming a teacher 1 1.00 1.67 I wish a teacher 1 1.00 1.00

A1 literal, same light verb At eight Stella had a class in General Biology. # of responses GJ MC had a class 64 6.00 6.00 was in a class 4 5.17 5.67 attended a class 15 5.00 5.50 took a class 39 5.00 4.83 went to a class 7 4.33 4.83 joined a class 2 3.33 3.83 attended to a class 1 2.33 3.00 did a class 2 1.83 3.00 experienced a class 2 1.33 2.83 got a class 5 1.67 2.00 received a class 1 1.33 1.83 held a class 1 1.50 2.00 conducted a class 1 1.17 1.67 gave a class 1 1.33 1.50 finished a class 1 1.00 1.67 went a class 8 1.00 1.67

306

sit a class 1 1.00 1.50 existed a class 1 1.00 1.17 there was a class 1 1.00 1.00

A1 literal, same light verb Bob got a ticket for the concert tonight. # of responses GJ MC got a ticket 64 6.00 6.00 acquired a ticket 2 4.67 5.50 purchased a ticket 5 4.67 4.50 received a ticket 23 4.17 4.50 obtained a ticket 13 4.17 4.17 bought a ticket 20 3.67 4.33 was given a ticket 1 3.00 4.33 gained a ticket 2 3.17 3.83 had a ticket 15 2.83 3.33 held a ticket 1 2.50 3.33 took a ticket 10 2.50 2.50 won a ticket 1 2.17 2.83 found a ticket 1 1.67 3.00 reserved a ticket 4 1.83 2.17 made a ticket 3 1.00 1.33 offered a ticket 1 1.00 1.00

* * * * * * * * * * * * * * * *

A1 abstract, same light verb We need to get the nod from five of the seven members selected from the committee. # of responses GJ MC got the nod 57 6.00 6.00 obtain the nod 6 3.83 5.33 be given the nod 2 3.83 5.17 receive the nod 15 4.33 4.00 see the nod 1 3.17 4.17 acquire the nod 5 2.00 4.33 be issued the nod 1 2.17 4.00 have the nod 24 1.83 3.67 gain the nod 1 1.33 3.00 achieve the nod 1 1.33 2.67 be allowed the nod 1 1.17 2.67 gather the nod 1 1.00 2.00 grab the nod 1 1.00 1.83 hold the nod 1 1.00 1.67 accept the nod 2 1.00 1.33 put the nod 1 1.00 1.33 ask the nod 1 1.00 1.17 be accepted the nod 1 1.00 1.17 make the nod 6 1.00 1.17 permit the nod 1 1.00 1.17 take the nod 8 1.00 1.17 count the nod 1 1.00 1.00

307

A1 abstract, same light verb Albuquerque’s low-wage workers got a break. They finally got a boost in wages. # of responses GJ MC got a break 52 6.00 6.00 received a break 10 5.50 5.67 were given a break 1 5.33 5.67 obtained a break 2 3.50 5.00 experienced a break 1 3.67 3.67 had a break 24 2.83 4.00 acquired a break 1 2.50 3.17 gained a break 2 1.83 2.83 succeeded a break 2 2.00 2.33 found a break 1 1.83 2.00 accepted a break 1 1.50 2.00 won a break 1 1.67 1.83 hit a break 1 1.50 1.83 deserved a break 1 1.17 1.83 made a break 11 1.17 1.67 got chance a break 1 1.00 1.83 gave a break 2 1.00 1.67 needed a break 1 1.00 1.50 became a break 1 1.00 1.17 burst into a break 1 1.00 1.17 took a break 15 1.00 1.00 NO ANSWER 1 – –

A1 abstract, same light verb Old theaters in Ireland always have an air of nostalgia. # of responses GJ MC have an air 54 6.00 6.00 carry an air 2 3.67 4.67 possess an air 2 3.67 4.00 depict an air 1 3.50 3.83 give an air 7 3.00 3.67 emit an air 1 3.00 3.50 maintain an air 1 3.00 3.00 produce an air 1 2.83 3.17 exude an air 1 3.33 2.33 retain an air 1 2.33 3.33 create an air 3 2.50 3.00 hold an air 2 2.50 3.00 spread an air 1 2.33 2.00 offer an air 1 1.83 2.17 initiate an air 1 1.67 2.17 are an air 1 1.50 2.33 send an air 1 1.50 2.33 surround an air 1 1.83 1.83 wear an air 1 1.17 2.17 are in an air 1 1.33 2.00 feel an air 4 1.50 1.67 float an air 1 1.17 1.83 resemble an air 1 1.17 1.83

308

take an air 3 1.33 1.67 pack an air 1 1.17 1.67 get an air 3 1.33 1.50 keep an air 4 1.33 1.50 make an air 9 1.33 1.50 exist an air 2 1.17 1.50 blow an air 1 1.17 1.33 include an air 1 1.00 1.33 pretend an air 1 1.00 1.33 full an air 1 1.00 1.17 remember an air 1 1.00 1.17 remind an air 1 1.00 1.17 are smelled an air 1 1.00 1.00 in an air 1 1.00 1.00 think an air 1 1.00 1.00

A1 abstract, same light verb Johnny Mercer had an ear for country music and was able to pick out any tune he wanted by ear. # of responses GJ MC had an ear 61 6.00 6.00 was blessed with an ear 1 3.83 4.17 always showed an ear 1 3.17 4.00 always had an ear 1 3.33 3.67 was born with an ear 1 3.33 3.50 possessed an ear 3 2.67 3.33 was given an ear 1 2.50 3.17 held an ear 1 2.50 3.00 owned an ear 1 2.33 2.33 kept an ear 3 2.17 2.33 was an ear 1 1.67 2.33 was gifted an ear 1 1.50 2.33 inhibited an ear 1 1.67 2.17 was born an ear 1 1.50 2.17 got an ear 9 1.33 2.33 made an ear 1 1.67 1.67 existed an ear 1 1.00 2.00 had ability an ear 1 1.00 2.00 offered an ear 1 1.50 1.50 put an ear 1 1.17 1.50 used an air 1 1.00 1.67 took an air 2 1.00 1.50 earned an ear 1 1.00 1.33 cut an ear 1 1.00 1.00 talented an ear 3 1.00 1.00 NO ANSWER 3 – –

A1 abstract, same light verb Churches are taking a stand against the spread of HIV. # of responses GJ MC taking a stand 38 6.00 6.00 making a stand 8 5.00 5.50

309

on a stand 3 5.00 2.83 supporting a stand 1 2.33 3.50 holding a stand 3 2.17 3.33 creating a stand 1 2.17 3.17 showing a stand 3 1.50 3.67 continuing a stand 1 1.67 2.33 in a stand 3 2.17 2.67 keeping a stand 14 2.00 2.67 having a stand 21 1.83 2.83 positing a stand 1 1.83 2.67 doing a stand 1 1.83 2.50 giving a stand 5 1.00 2.83 faring a stand 1 1.50 2.00 getting a stand 5 1.17 2.17 retaining a stand 1 1.33 2.00 stating a stand 1 1.00 2.33 appealing a stand 1 1.00 2.17 at a stand 2 1.00 2.17 becoming a stand 1 1.00 2.17 providing a stand 1 1.50 1.67 remaining a stand 1 1.00 2.00 behavior a stand 1 1.00 1.50 being a stand 1 1.00 1.50

* * * * * * * * * * * * * * * *

A2 literal, different light verb Campbell took a walk on the beach last Saturday. # of responses GJ MC took a walk 52 6.00 6.00 went for a walk 6 5.83 5.83 went on a walk 2 5.83 5.83 went out on a walk 1 5.33 5.50 had a walk 21 2.67 4.33 did a walk 11 2.33 4.17 went a walk 7 1.83 2.83 made a walk 6 1.17 3.00 held a walk 1 1.17 2.67 got a walk 1 1.00 1.83 around a walk 2 1.00 1.50 gave a walk 2 1.00 1.50

A2 literal, different light verb If you are thinking of taking an exam, make sure that you have enough time for home study. # of responses GJ MC taking an exam 65 6.00 6.00 doing an exam 3 1.83 5.00 going through an exam 1 3.33 4.17 participating in an exam 1 3.17 3.50 having an exam 27 3.33 3.33 going to an exam 1 2.50 3.50 passing an exam 1 2.33 3.33

310

attending an exam 1 2.33 2.50 scheduling an exam 1 2.33 2.33 receiving an exam 1 2.00 2.50 trying for an exam 2 2.17 2.33 getting an exam 8 2.00 2.33 applying for an exam 1 1.83 2.33 challenging an exam 1 1.83 2.33 making an exam 2 1.83 2.17 sitting an exam 1 1.83 2.33 given an exam 1 2.00 1.67 seeing an exam 1 1.50 2.00 waving an exam 1 1.00 2.00 trying an exam 2 1.00 1.67 biting in for an exam 1 1.00 1.00

A2 literal, different light verb I don't want to have an argument with you or anyone else about this. # of responses GJ MC have an argument 39 6.00 6.00 engage in an argument 2 4.00 4.83 go into an argument 1 4.50 4.00 take part in an argument 1 4.17 4.33 create an argument 3 3.50 3.83 go through an argument 1 3.00 4.00 raise an argument 1 3.00 4.17 experience an argument 3 3.17 3.50 initiate an argument 1 2.50 4.17 hold an argument 1 2.50 3.83 begin an argument 1 2.67 2.83 conduct an argument 1 1.83 3.50 start an argument 3 2.50 2.67 deal an argument 1 1.33 3.83 make an argument 35 2.17 2.67 participate an argument 1 2.17 2.67 perform an argument 1 1.83 2.67 produce an argument 1 1.50 2.83 fight an argument 1 1.67 2.17 build an argument 1 1.50 2.33 discuss an argument 10 1.17 2.50 dispute an argument 1 1.17 2.50 take an argument 5 1.50 2.17 argue an argument 1 1.00 2.50 conflict an argument 1 1.33 2.17 do an argument 9 1.00 2.50 get an argument 7 1.00 2.50 say an argument 1 1.17 2.17 talk an argument 4 1.00 2.33 bring an argument 1 1.00 2.17 give an argument 1 1.17 1.83 go an argument 1 1.00 2.00 debate an argument 5 1.17 1.67 query an argument 1 1.00 1.83 keep an argument 1 1.17 1.50

311

A2 literal, different light verb Jean made a copy of the article and gave it to her mother. # of responses GJ MC made a copy 57 6.00 6.00 created a copy 4 5.83 5.83 printed a copy 1 5.67 5.83 prepared a copy 1 5.50 5.67 produced a copy 1 5.83 5.00 did a copy 5 3.17 3.67 got a copy 16 2.50 3.83 obtained a copy 1 2.67 3.50 had a copy 16 1.67 2.17 took a copy 28 1.17 1.67 held a copy 1 1.00 1.67 kept a copy 1 1.00 1.33 used a copy 1 1.00 1.33 received a copy 1 1.00 1.00

A2 literal, different light verb Do not take a bath when suffering from cold fever. # of responses GJ MC take a bath 66 6.00 6.00 have a bath 23 4.67 5.67 put yourself into a bath 1 4.17 4.67 go in a bath 1 3.67 4.67 go into a bath 5 3.17 4.17 get in a bath 1 2.67 4.33 get a bath 7 3.17 3.00 enter a bath 2 2.67 3.17 soak into a bath 1 2.50 3.33 use a bath 3 2.33 3.33 go to a bath 2 2.00 3.17 do a bath 3 2.33 2.83 try a bath 1 1.83 2.67 bathe a bath 1 1.67 2.50 into a bath 1 2.17 1.67 go a bath 1 1.33 2.00 give a bath 1 1.67 1.50

* * * * * * * * * * * * * * * *

A2 abstract, different light verb “A Beautiful Mind” scriptwriters took liberties with the facts in Nasar’s biography. # of responses GJ MC took liberties 10 6.00 6.00 made liberties 12 2.83 3.33 did liberties 8 2.67 3.17 had liberties 13 2.67 3.17 offered liberties 1 2.83 3.00 demonstrated liberties 1 2.83 2.83 showed liberties 3 2.33 3.33

312

gave liberties 3 2.00 3.17 kept liberties 1 1.50 2.33 performed liberties 1 2.33 2.50 filed liberties 1 2.00 2.33 used to liberties 1 2.00 2.33 got liberties 4 1.83 2.33 stole liberties 1 1.83 2.17 too friendly liberties 1 2.17 1.83 were liberties 2 1.83 2.17 touched liberties 1 1.67 2.17 acted liberties 1 1.33 2.50 took away liberties 1 1.83 2.00 asked liberties 2 1.67 2.00 familiar to liberties 1 1.83 1.83 played liberties 3 1.67 1.83 tied liberties 1 1.83 1.67 rid of liberties 1 1.50 2.00 were friendly liberties 1 1.33 2.00 became liberties 4 1.50 1.67 skinshipped liberties 1 1.67 1.50 no respect liberties 1 1.33 1.83 had no liberties 1 1.67 1.33 behaved liberties 1 1.00 1.67 violated liberties 1 1.17 1.50 over-rewrote liberties 1 1.50 1.00 looked liberties 1 1.00 1.33 NO ANSWER 10 – –

A2 abstract, different light verb The temperature took a dive last night. # of responses GJ MC took a dive 7 6.00 6.00 made a dive 14 4.50 5.00 suddenly made a dive 1 4.17 5.17 experienced a dive 2 3.67 5.00 did a dive 4 3.83 4.50 fell into a dive 1 2.17 3.50 had a dive 15 2.33 3.17 was thought to have a dive 1 2.50 3.00 went down like a dive 1 2.50 3.00 showed a dive 1 2.17 2.50 dropped a dive 8 1.17 2.50 dropped down a dive 2 1.50 2.00 was seen to have a dive 1 1.17 2.33 fell down a dive 1 1.17 2.17 down a dive 5 1.33 2.00 fell a dive 4 1.17 1.83 followed a dive 1 1.17 1.83 shaped a dive 1 1.17 1.67 went a dive 8 1.17 1.67 went down a dive 5 1.17 1.67 declined a dive 3 1.33 1.50 got down a dive 2 1.17 1.50

313

got a dive 12 1.17 1.33 decreased a dive 9 1.17 1.17 forwarded a dive 1 1.17 1.17 was a dive 1 1.17 1.17 became a dive 4 1.00 1.17 became down a dive 1 1.00 1.17 degreased a dive 1 1.00 1.00 turned off a dive 1 1.00 1.00 was made a dive 1 1.00 1.00 NO ANSWER 1 – –

A2 abstract, different light verb David has made a killing in Network Marketing. # of responses GJ MC made a killing 26 6.00 6.00 earned a killing 4 5.33 5.33 gained a killing 2 3.33 4.00 got a killing 24 2.67 3.67 achieved a killing 1 2.50 3.67 done a killing 2 2.50 3.17 experienced a killing 1 2.33 3.33 had a killing 11 2.50 2.67 won a killing 2 2.50 2.67 enjoyed a killing 1 2.00 2.83 made money a killing 1 2.00 2.67 become a killing 2 1.83 2.83 been a killing 1 1.83 2.50 took a killing 1 1.67 2.50 received a killing 2 1.50 2.50 gave a killing 2 1.67 2.17 had an opportunity in a killing 1 1.67 2.17 finished a killing 1 1.50 2.00 big money a killing 1 1.33 1.67 put a killing 1 1.50 1.50 lost a killing 1 1.17 1.17 NO ANSWER 10 – –

A2 abstract, different light verb The cars manufactured by Saab Automobiles (a company name) had a name for safety and crash-resistance. # of responses GJ MC had a name 14 6.00 6.00 are known to have a name 1 5.33 5.50 earned a name 3 4.83 5.17 gained a name 7 4.83 5.00 were given a name 2 4.83 5.00 obtained a name 3 3.67 4.50 acquired a name 2 3.17 4.67 received a name 13 3.50 4.33 became a name 2 3.50 4.00 got a name 42 3.33 4.00 were awarded a name 1 3.50 3.83

314

won a name 1 3.50 3.67 reputed a name 1 2.83 4.33 are given a name 1 2.67 3.67 made a name 14 2.67 3.67 took a name 5 2.50 3.83 created a name 1 2.83 3.50 were presented a name 1 2.67 2.67 were prized a name 1 2.33 2.83 were offered a name 1 2.33 2.33 promoted a name 1 1.83 2.67 popularized a name 1 2.17 2.17 raised a name 1 2.17 2.17 rose a name 1 2.00 2.00 spread a name 1 1.83 2.17 honored a name 1 1.67 2.17 got reputation a name 1 1.50 2.17 provided a name 1 1.50 2.00 sprouted a name 1 1.33 2.00 brought a name 1 1.33 1.83 gave a name 3 1.33 1.50 were allowed a name 1 1.33 1.50

A2 abstract, different light verb The shop assistant got the sack because he swore at/ to a customer. # of responses GJ MC got the sack 33 6.00 6.00 was given the sack 1 5.00 5.17 received the sack 4 3.67 5.00 met the sack 1 2.50 2.83 had the sack 6 2.00 2.67 was got the sack 1 1.17 3.33 packed the sack 2 2.00 2.00 forced to be the sack 1 1.83 2.00 took the sack 4 1.33 2.33 fired the sack 10 1.50 1.83 made the sack 5 1.50 1.83 was the sack 3 1.50 1.83 hired the sack 1 1.67 1.50 was fired the sack 4 1.50 1.67 broke the sack 1 1.67 1.17 cut the sack 4 1.33 1.50 was resigned the sack 1 1.50 1.33 was tired the sack 1 1.50 1.33 was kicked out the sack 1 1.00 1.67 was cut the sack 1 1.33 1.33 lost the sack 1 1.33 1.17 lost work the sack 1 1.33 1.17 dumped the sack 1 1.17 1.17 became the sack 5 1.33 1.00 gained the sack 1 1.33 1.00 came to the sack 1 1.00 1.17 NO ANSWER 3 – –

315

* * * * * * * * * * * * * * * *

A3 literal, content verb Sharon was scheduled to undergo an operation on Friday morning. # of responses GJ MC undergo an operation 2 6.00 6.00 go through an operation 2 5.00 5.33 get an operation 26 5.00 5.17 have an operation 44 5.00 5.17 receive an operation 13 4.50 5.00 be given an operation 3 4.33 4.67 do an operation 5 2.67 2.67 attend an operation 1 2.00 3.00 experienced an operation 1 1.50 2.67 be performed an operation 1 1.67 2.33 undertake an operation 1 2.00 2.00 take an operation 37 1.50 2.33 perform an operation 1 1.83 1.83 hold an operation 1 1.17 2.17 be executed an operation 1 1.33 2.00 get surgery an operation 1 1.17 2.00 ground an operation 1 1.00 1.67 make an operation 4 1.00 1.67 start an operation 1 1.00 1.50 surge an operation 1 1.17 1.17

A3 literal, content verb Have you ever placed an order online with your credit card? # of responses GJ MC placed an order 4 6.00 6.00 made an order 50 4.83 4.83 requested an order 2 3.17 4.00 done an order 13 2.50 3.67 put an order 2 1.83 3.17 got an order 3 2.17 2.50 reserved an order 1 1.83 2.67 given an order 7 2.00 2.17 had an order 16 1.83 2.33 asked for an order 1 1.83 2.17 called an order 1 1.33 2.50 taken an order 19 1.83 2.00 held an item 1 1.17 2.17 asked an order 4 1.33 2.00 chosen an order 1 1.50 1.83 picked an order 1 1.00 2.33 tried an order 2 1.17 2.00 kept an item 1 1.33 1.67 ordered an order 1 1.33 1.67 been taken an order 1 1.17 1.67 already an order 1 1.33 1.50 been an order 2 1.33 1.33

316

A3 literal, content verb The tornado caused damage to 43 homes before finally dissipating to Columbia. # of responses GJ MC caused damage 11 6.00 6.00 did damage 2 5.33 5.83 made damage 23 4.67 5.00 was responsible for the damage 1 4.67 4.67 left damage 4 4.00 4.50 brought damage 5 3.33 4.33 created damage 4 3.17 4.33 gave damage 41 3.00 4.17 produced damage 2 3.00 4.00 induced damage 1 2.67 3.83 made a damage 1 2.33 4.00 provided damage 1 2.33 3.67 put damage 1 2.50 3.50 placed damage 1 2.67 3.17 produced a damage 1 2.00 3.33 forced damage 1 2.00 3.00 posed a damage 1 1.67 3.17 created a damage 1 1.83 3.00 left a damage 1 1.83 3.00 gave a damage 1 1.67 3.00 added damage 1 1.50 2.83 gave accident damage 1 1.83 2.50 affected damage 1 1.33 2.50 led to damage 3 1.50 2.33 attacked damage 1 1.17 2.33 got damage 4 1.33 2.17 took damage 2 1.67 1.83 heavily damage 1 1.50 1.67 had damage 8 1.33 1.83 which damage 1 1.50 1.50 hit damage 2 1.33 1.50 that damage 1 1.33 1.50 was a cause of damage of a damage 1 1.33 1.33 happened damage 1 1.17 1.33

A3 literal, content verb This scientific discovery could save 3 to 4 million people a year from developing cancer. # of responses GJ MC developing cancer 1 6.00 6.00 getting cancer 37 5.67 5.67 becoming victim of cancer 1 4.17 5.50 having cancer 26 4.83 4.83 contracting cancer 1 4.67 4.83 suffering from cancer 4 4.33 5.17 experiencing cancer 2 4.00 4.50 obtaining cancer 1 3.33 3.83 being diagnosed cancer 1 2.50 4.50 suffering cancer 1 3.33 3.67 receiving cancer 4 3.00 3.33

317

holding cancer 1 2.83 3.00 catching cancer 1 2.50 3.17 carrying cancer 2 2.67 2.83 growing cancer 1 2.33 2.83 dealing cancer 1 1.83 2.67 giving cancer 1 2.50 2.00 becoming cancer 26 1.33 2.83 causing cancer 1 1.50 2.00 taking cancer 5 1.67 1.83 booming cancer 1 1.50 1.83 going into cancer 1 1.50 1.83 being cancer 17 1.33 1.67 NO ANSWER 1 – –

A3 literal, content verb Johnny Rivers accepted an offer to do a concert for the reopening of the Mall of Memphis. # of responses GJ MC accepted an offer 3 6.00 6.00 took an offer 19 4.17 4.67 got an offer 51 3.67 4.50 obtained an offer 2 2.67 3.50 was presented with an offer 1 2.83 2.83 had an offer 26 2.33 3.17 was made an offer 2 2.50 2.83 received an offer 47 2.17 2.67 was given an offer 3 2.17 2.67 was provided an offer 1 2.17 2.50 gained an offer 1 2.17 2.00 listened an offer 1 1.17 2.17 was asked for an offer 2 1.67 1.83 was obtained an offer 1 1.33 2.17 ordered an offer 1 1.33 1.67 put an offer 1 1.17 1.83 made an offer 6 1.67 1.17 concerned an offer 1 1.33 1.50 gave an offer 1 1.17 1.50

* * * * * * * * * * * * * * * *

A3 abstract, content verb It is unclear whether Russian officials will strike a bargain with the Government of Lithuania. 3 of responses GJ MC make a bargain 32 6.00 6.00 negotiate a bargain 2 4.83 5.00 try for a bargain 1 3.50 4.17 accept a bargain 2 3.50 4.00 consider a bargain 1 3.33 3.83 do a bargain 4 2.83 4.00 deal a bargain 2 2.83 3.83 go into a bargain 1 2.67 3.83 commit a bargain 1 2.67 3.17

318

agree a bargain 4 2.50 3.17 have a bargain 30 2.67 3.00 sign a bargain 1 2.67 3.00 go through a bargain 1 2.33 3.17 hold a bargain 2 2.67 2.83 conclude a bargain 1 2.00 3.00 cooperate a bargain 1 2.00 3.00 offer a bargain 1 2.00 3.00 build a bargain 1 1.50 3.00 receive a bargain 1 2.00 2.17 meet a bargain 1 1.83 2.33 tie a bargain 1 1.83 2.33 committee a bargain 1 1.67 2.33 gain a bargain 1 1.83 2.17 get a bargain 6 1.83 2.17 try a bargain 1 1.33 2.67 connect a bargain 1 1.50 2.33 take a bargain 13 1.33 2.50 think of a bargain 1 1.33 2.17 go a bargain 1 1.17 2.17 commute a bargain 1 1.00 2.17 give a bargain 4 1.33 1.83 keep a bargain 5 1.33 1.50 be a bargain 2 1.00 1.67 placed a bargain 1 1.17 1.50 help a bargain 1 1.17 1.33 corporate a bargain 2 1.00 1.33 relate a bargain 1 1.00 1.33 NO ANSWER 2 – –

A3 abstract, content verb Politicians of every stripe -- from ward heelers to the President of the United States -- paid court to the archbishop of New York. # of responses GJ MC paid court 1 6.00 6.00 gave court 8 3.83 4.33 made court 16 2.67 4.00 held court 1 3.00 2.83 took court 22 2.67 2.67 offered court 1 2.33 3.33 kept court 1 2.33 2.83 entertained court 1 2.33 2.67 had court 12 2.33 2.67 showed a court 1 2.50 2.50 got court 9 2.00 2.50 earned court 1 2.00 2.17 took a court 2 1.83 2.33 put court 1 1.67 2.33 created court 1 1.17 2.17 apple-polished court 1 1.33 1.83 asked court 1 1.17 1.33 reevaluated court 1 1.33 1.17 NO ANSWER 11 – –

319

A3 abstract, content verb On September 13th we bore witness to an event that should serve as a turning point in the history of the Middle East. # of responses GJ MC bore witness 1 6.00 6.00 were witness 16 4.67 5.00 were the witness 2 3.83 5.00 became a witness 1 3.33 4.17 came to be the witness 1 2.83 3.67 became witness 13 3.00 3.33 paid witness 1 2.33 3.00 eyed witness 2 2.17 3.00 made a witness 1 2.00 3.17 took witness 7 2.33 2.83 experienced witness 1 1.83 3.17 made witness 7 2.17 2.67 watched witness 10 1.83 2.83 got to witness 1 1.67 2.83 happened to witness 1 2.00 2.50 got witness 9 1.67 2.67 saw witness 18 2.00 2.17 gave witness 1 1.83 2.33 did witness 1 1.33 2.67 were given witness 1 1.67 2.33 had witness 16 1.50 2.17 saw as witness 1 1.50 1.67 stared witness 1 1.33 1.83 had a witness 2 1.50 1.50 looked at witness 1 1.00 1.83 saw a witness 1 1.00 1.83 hatched witness 1 1.00 1.67 looked witness 3 1.00 1.67 witnessed witness 1 1.17 1.50 behaved witness 1 1.00 1.50 checked witness 1 1.00 1.50 found witness 3 1.00 1.33

A3 abstract, content verb Joan cut corners to get there on time. # of responses GJ MC cut corners 24 6.00 6.00 took a short cut at the corners 1 3.50 4.50 cut the corners 1 3.33 4.00 short cut corners 7 2.00 3.67 skipped corners 2 2.00 3.50 shortened corners 3 1.50 2.67 made a short cut corners 1 1.33 2.83 crossed corners 2 1.17 2.50 passed corners 1 1.17 2.17 broke corners 1 1.17 2.00 walked corners 5 1.50 1.67

320

took corners 19 1.33 1.50 took cut corners 1 1.33 1.50 turned corners 3 1.50 1.17 used corners 1 1.33 1.33 chose corners 3 1.17 1.33 across corners 1 1.17 1.17 did corners 4 1.17 1.17 found corners 1 1.17 1.17 got corners 6 1.17 1.17 went corners 6 1.17 1.17 enjoyed corners 1 1.00 1.17 had corners 6 1.17 1.00 made corners 12 1.17 1.00 made short corners 1 1.17 1.00 shipped corners 1 1.17 1.00 went to corners 1 1.00 1.17

A3 abstract, content verb Tanya Haden and her longtime boyfriend, Carey Hart, tied the knot over the weekend in Las Vegas. # of responses GJ MC tied the knot 10 6.00 6.00 made the knot 16 1.50 2.33 did the knot 1 2.17 1.50 took the knot 9 1.67 2.00 agreed the knot 1 1.33 2.17 placed the knot 1 1.67 1.83 created the knot 2 1.67 1.50 marriaged the knot 2 1.17 2.00 married the knot 14 1.50 1.67 experienced the knot 1 1.67 1.33 tightened the knot 1 1.33 1.50 got the knot 20 1.17 1.50 promised the knot 1 1.17 1.50 pulled the knot 1 1.17 1.50 held the knot 1 1.33 1.33 got married the knot 6 1.00 1.50 opened the knot 1 1.00 1.50 became the knot 1 1.17 1.17 had the knot 20 1.17 1.17 involved the knot 1 1.17 1.17 gone the knot 1 1.17 1.17 meant the knot 1 1.17 1.17 put the knot 1 1.00 1.33 gave the knot 4 1.17 1.00 received the knot 2 1.00 1.17 NO ANSWER 4 – –

321

Appendix IX: Tables from Experiment 2

Appendix IX-1: Grammaticality Judgments. Experiment 2 (Figures 5-1, 5-2, 5-3 and 5-4)

Type of Low-intermediate Advanced (# = 31) Total (# = 66) collocations (#= 35) High (4.5-6) Token Percentag Token Percentag Token Percentag Mid (2.67-4.33) e e e Low (1-2.5) A1 Literal same High 215 55.56% 204 52.31% 419 53.94% light V Mid 118 30.49% 137 35.13% 255 32.81% Low 54 13.95% 49 12.56% 103 13.26% Abstract High 145 46.33% 135 42.45% 280 44.39% Mid 29 9.27% 48 15.09% 77 12.18% Low 139 44.41% 135 42.45% 274 43.43% A2 Literal differe High 156 49.37% 162 51.27% 318 50.32% nt light Mid 35 11.08% 46 14.56% 81 12.82% V Low 125 39.56% 108 34.18% 233 36.87% Abstract High 53 19.63% 68 27.42% 121 23.53% Mid 85 31.48% 71 28.63% 156 30.06% Low 132 48.89% 109 43.95% 241 46.42% A3 Literal Conten High 117 32.50% 123 37.39% 240 34.95% t Mid 75 20.83% 71 21.58% 146 21.21% V Low 168 46.67% 135 41.03% 303 43.85% Abstarct High 39 12.34% 45 16.73% 84 14.54% Mid 59 18.67% 50 18.59% 109 18.63% Low 218 68.99% 174 64.68% 392 66.84% Total (# = High 725 36.95% 737 39.41% 1462 38.18% 66) Mid 401 20.44% 423 22.62% 824 21.53% Low 836 42.61% 710 37.97% 1546 40.29% TABLE IX-1. Grammaticality Judgments. Experiment 2 (Figures 5-1, 5-2, 5-3 and 5-4)

322

Appendix IX-2: Grammaticality Judgments. Experiment 2 (Figures 5-5, 5-6, 5-7 and 5-8)

Type of collocations Low-intermediate Advanced (# = 31) Total (# = 66) High (4.5-6) (# = 35) Mid (2.67-4.33) Token Percentag Token Percentag Token Percentag Low (1-2.5) e e e 1. Literal Literal High 488 45.91% 489 47.25% 977 46.58% vs. Mid 228 21.45% 254 24.54% 482 23.00% Abstract Low 347 32.64% 292 28.21% 639 30.43% meaning Abstract High 237 26.36% 248 29.70% 485 28.03% Mid 173 19.24% 169 20.24% 342 19.74% Low 489 54.40% 418 50.06% 907 52.23% 2. With equi With vs. High 360 51.43% 339 47.88% 699 49.66% without Mid 147 21.00% 185 26.13% 332 23.57% L1 Low 193 27.57% 184 25.99% 907 26.78% equivale Without nt High 365 28.92% 398 34.25% 763 31.59% Mid 254 20.13% 238 20.48% 492 20.31% Low 643 50.95% 526 45.27% 1169 48.11% 3. Light V Light High 569 44.25% 569 44.73% 1138 44.49% V+N Mid 267 20.76% 302 23.74% 569 11.13% vs. Low 450 34.99% 401 31.53% 851 33.26% Content Content V V+N High 156 23.08% 168 28.09% 324 25.59% Mid 134 19.82% 121 20.23% 255 20.03% Low 386 57.10% 309 51.67% 695 54.39% 4. Same lig Same High 360 51.43% 339 47.88% 699 49.66% light V Mid 147 21.00% 185 26.13% 332 23.57% vs. Low 193 27.57% 184 25.99% 377 26.78% Differen Different t light V High 209 35.67% 230 40.78% 439 38.23% Mid 120 20.48% 117 20.74% 237 20.61% Low 257 43.86% 217 38.48% 474 41.17% TABLE IX-2: Grammaticality Judgments. Experiment 2 (Figures 5-5, 5-6, 5-7 and 5-8)

323

Appendix IX-3: Meaning Conveyed. Experiment 2 (Figures 5-9, 5-10, 5-11, 5-12)

Type of Low-intermediate Advanced (# = 31) Total (# = 66) collocations (#= 35) High (4.5-6) Mid (2.67-4.33) Token Percentag Token Percentag Token Percentag e e e Low (1-2.5) A1 Literal same High 252 65.45% 192 63.25% 444 64.35% light V Mid 86 22.34% 80 27.82% 166 25.08% Low 47 12.21% 17 8.92% 64 10.57% Abstract High 146 47.87% 145 44.48% 291 46.18% Mid 73 23.93% 99 30.37% 172 27.15% Low 86 28.20% 82 25.15% 168 26.68% A2 Literal differe High 159 50.96% 167 54.05% 326 52.51% nt light Mid 93 29.81% 81 26.21% 174 28.01% V Low 60 19.23% 61 19.74% 121 19.49% Abstract High 56 20.66% 82 33.06% 138 26.86% Mid 104 38.38% 94 37.90% 198 38.14% Low 111 40.96% 72 29.03% 183 35.00% A3 Literal Conten High 173 45.29% 163 47.38% 336 46.34% t Mid 105 27.49% 106 30.81% 211 29.15% V Low 104 27.23% 75 21.80% 179 24.52% Abstarct High 41 12.97% 49 17.88% 90 15.43% Mid 98 31.01% 91 33.21% 189 32.11% Low 177 56.01% 134 48.91% 311 52.46% Total (# = High 827 41.96% 798 44.58% 1625 43.27% 66) Mid 559 28.36% 551 30.78% 1110 29.57% Low 585 29.68% 441 24.64% 1026 27.16% TABLE IX-3. Meaning Conveyed. Experiment 2 (Figures 5-9, 5-10, 5-11, 5-12)

324

Appendix IX-4: Meaning Conveyed. Experiment 2 (Figures 5-13, 5-14, 5-15, 5-16)

Type of Low-intermediate Advanced (# = 31) Total (# = 66) collocations (# = 35) High (4.5-6) Token Percentag Token Percentag Token Percentag Mid (2.67-4.33) e e e Low (1-2.5) 1. Literal Literal High 584 54.12% 522 55.41% 1106 54.77% vs. Mid 284 26.32% 267 28.34% 551 27.33% Abstract Low 211 19.56% 153 16.24% 364 17.90% meaning Abstract High 243 27.24% 276 32.55% 519 29.90% Mid 275 30.83% 284 33.49% 559 32.16% Low 374 41.93% 288 33.96% 662 37.95% 2. With equi With vs. High 398 57.68% 337 54.80% 735 56.24% without Mid 159 23.04% 179 29.11% 338 26.08% L1 Low 133 19.28% 99 16.10% 993 17.69% equivale Without nt High 429 33.52% 461 39.23% 890 36.38% Mid 400 31.23% 372 31.66% 772 31.45% Low 452 35.28% 342 29.11% 486 23.00% 3. Light V Light High 613 48.15% 586 50.00% 1199 49.08% V+N Mid 356 27.97% 354 30.20% 710 29.09% vs. Low 304 23.88% 232 19.80% 536 21.84% Content Content V V+N High 214 30.66% 212 34.30% 426 32.48% Mid 203 29.08% 197 31.88% 400 30.48% Low 281 40.26% 209 33.82% 490 37.04% 4. Same lig Same High 398 57.68% 337 54.80% 735 56.24% light V Mid 159 23.04% 179 29.11% 338 26.08% vs. Low 133 19.28% 99 16.10% 232 17.69% Differen Different t light V High 215 36.88% 249 44.70% 464 40.79% Mid 197 33.79% 175 31.42% 372 32.61% Low 171 29.33% 133 23.88% 304 26.61% TABLE IX-4: Meaning Conveyed. Experiment 2 (Figures 5-13, 5-14, 5-15, 5-16)

325

Appendix IX-5: Grammaticality Judgments by native speakers vs. learners’ confidence in their productions. Experiment 2 (Figures 5-17 and 5-18)

Type of collocations Low-intermediate learners Advanced learners Grammaticality by (# = 35) (# = 31) natives: Confide Not Total Confide Not Total Grammatical: High (4.5-6) nt confident nt confident Ungrammatical: Mid/Low (7-10) (1-6) (7-10) (1-6) (1-4.33) A1 Literal same Grammatical 163 52 215 178 26 204 light 42.12% 13.44% 55.56% 45.64% 6.67% 52.31% verb Ungrammati 103 69 172 131 55 186 44.44% 47.69% cal 26.62% 18.73% 33.59% 14.10% Abstract Grammatical 56 89 145 80 55 135 17.89% 28.43% 46.33% 25.16% 17.30% 42.45% Ungrammati 50 118 168 71 112 183 cal 15.97% 37.69% 53.67% 22.33% 35.22% 57.55% A2 Literal different Grammatical 127 29 156 139 23 162 light 40.19% 9.18% 49.37% 43.99% 7.28% 51.27% Ungrammati verb 79 81 160 110 44 154 cal 25.00% 25.63% 50.63% 34.81% 13.93% 48.73% Abstract Grammatical 15 38 53 26 42 68 5.56% 12.03% 19.63% 9.63% 15.56% 27.42% Ungrammati 60 157 217 64 116 180 cal 22.22% 49.68% 80.37% 23.70% 42.96% 72.58% A3 Literal content Grammatical 78 39 117 101 22 123 verb 21.67% 10.83% 32.50% 30.70% 6.69% 37.39% Ungrammati 122 121 243 150 56 206 cal 33.89% 23.61% 67.50% 45.59% 17.02% 62.61% Abstract Grammatical 8 31 39 26 19 45 2.53% 9.81% 12.34% 9.67% 7.06% 16.73% Ungrammati 73 204 277 71 153 224 cal 23.10% 64.56% 87.66% 26.39% 56.88% 83.27% Total Grammatical 447 278 725 550 187 737 (# = 66) 22.78% 14.17% 36.95% 29.41% 10.00% 39.41% Ungrammati 487 750 1237 597 536 1133 cal 24.82% 38.23% 63.05% 31.93% 28.66% 60.59%

TABLE IX-5: Grammaticality Judgments by native speakers vs. learners’ confidence in their productions. Experiment 2

326

(Figures 5-17 and 5-18)

Appendix IX-6: Meaning Conveyed rated by native speakers vs. learners’ confidence in their productions. Experiment 2 (Figures 5-19 and 5-20)

Type of collocations Low-intermediate learners Advanced learners (# = 35) (# = 31) MC: High (4.5-6) Confid Not Total Confide Not Total Mid (2.67-4.33) ent Confide nt confide Low (1-2.5) (7-10) nt (1-6) (7-10) nt( 1-6) A1 Literal same Meaning 186 66 252 192 49 241 48.31% 17.14% 65.45% 50.39% 12.86% 63.25% light conveyed

verb Meaning NOT 78 55 133 97 43 140 conveyed 20.26% 14.28% 34.55% 25.46% 11.18% 36.75% Abstract Meaning 57 89 146 86 59 145 conveyed 18.69% 29.18% 47.87% 26.38% 18.10% 44.48%

Meaning NOT 46 113 159 74 107 181 conveyed 15.09% 37.05% 52.13% 22.70% 32.82% 55.52% A2 Literal different Meaning 129 30 159 150 17 167 41.35% 9.62% 50.96% 48.54% 5.50% 54.05% light conveyed

verb Meaning NOT 78 75 153 97 45 142 conveyed 25.00% 24.04% 49.04% 31.39% 14.57% 45.95% Abstract Meaning 17 39 56 32 50 82 conveyed 6.27% 14.39% 20.66% 12.90% 20.16% 33.06%

Meaning NOT 59 156 215 45 121 166 conveyed 21.77% 57.57% 79.34% 18.14% 48.79% 66.94% A3 Literal content Meaning 98 75 173 135 28 163 25.65% 19.63% 45.29% 39.24% 8.14% 47.38% verb conveyed

Meaning NOT 99 110 209 116 65 181 conveyed 25.92% 28.79% 54.71% 33.72% 18.89% 52.62% Abstract Meaning 10 31 41 29 20 49 conveyed 3.16% 9.81% 12.77% 10.58% 7.30% 17.88%

Meaning NOT 77 203 280 71 154 225 conveyed 22.78% 64.24% 87.23% 25.91% 56.21% 82.12% Total Meaning 497 330 827 624 223 847 (# = 66) conveyed 25.15% 16.70% 41.85% 33.16% 11.85% 45.01%

Meaning NOT 437 712 1149 500 535 1035 conveyed 22.12% 36.03% 58.15% 26.57% 28.43% 54.99%

327

TABLE IX-6: Meaning Conveyed rated by native speakers vs. learners’ confidence in their productions. Experiment 2 (Figures 5-19 and 5-20)

Appendix IX-7: Learners’ confidence, familiarity, and differences between confidence and familiarity. Experiment 2 (Figures 5-21, 5-22, 5-23, 5-24, 5-25 and 5-26)

Low-intermediate Advanced Total Conf Fam Diff Conf Fam Diff Conf Fam Diff Literal Literal 6.99 6.63 0.36 8.21 7.84 0.37 7.60 7.24 0.36 vs. Abstract 4.76 3.87 0.89 5.71 4.96 0.75 5.24 4.42 0.82 abstract L1 equiv With L1 6.35 5.69 0.66 7.22 6.70 0.52 6.79 6.20 0.59 Without L1 5.64 5.03 0.61 6.83 6.25 0.58 6.24 5.64 0.60 Light vs. Light V 6.11 5.47 0.64 6.97 6.49 0.48 6.54 5.98 0.56 content V Content V 5.40 4.82 0.58 6.94 6.22 0.72 6.17 5.52 0.65 Same vs. Same 6.35 5.69 0.66 7.22 6.70 0.52 6.79 6.20 0.59 different Different 5.88 4.82 1.06 6.72 6.22 0.50 6.30 5.52 0.78 light V Easy types 6.45 5.87 0.58 7.41 6.93 0.48 6.93 6.40 0.53 Total Difficult 5.42 4.64 0.78 6.55 5.91 0.64 5.99 5.28 0.71 Total 5.88 5.25 0.63 6.96 6.40 0.56 6.42 5.83 0.59 TABLE IX-7: Learners’ confidence, familiarity, and differences between confidence and familiarity. Experiment 2 (Figures 5-21, 5-22, 5-23, 5-24, 5-25 and 5-26)

328

Appendix X: Instruction handout

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

Appendix XI: Take-home exercises

1.「チケットを得た」 I ______a ticket for the 12:05 train.

1. got 2. took 3. did 4. made

2.「招待を受けた」 I ______an invitation to participate in a survey.

1. reached 2. received 3. caught 4. took

3.「先生になった」 He asked me why I ______a teacher.

1. became 2. got 3. turned 4. was

4.「プロジェクトをしている」 I am currently ______a project on diet and the obesity.

1. making 2. playing 3. holding 4. doing

5.「授業がある」 Do you ______a class today?

1. receive 2. have 3. take 4. go

6.「才能がある」 I wish I ______an ear for foreign languages.

1. obtained 2. were 3. had 4. endowed

7.「雰囲気があった」 The salesmen in the room looked older, better dressed and ______an air of power and authority.

1. owned 2. felt

361

3. had 4. obtained

8.「チャンスを得た」 Finally we ______a break late in the game and when time was about to run out, we took a 2 point lead.

1. hit 2. made 3. gave 4. got

9.「態度・立場をとる」 Hunters ______a stand against fishing law.

1. make 2. put 3. get 4. take

10.「承認を得た」 Richard ______the nod (from his wife) for a new bike.

1. took 2. obtained 3. made 4. got

11.「散歩をした」 Fred ______a walk in the park.

1. got 2. took 3. did 4. made

12.「コピーをとる」 ______a copy of your passport and visa and keep the copies in a separate location from your originals.

1. Make 2. Give 3. Take 4. Print

13.「議論した」 Kim ______an argument with her boss and now she’s suspended.

1. did 2. fought 3. had 4. made

14.「試験を受けた」 Michelle ______an exam for her economic course.

1. took 2. received 3. got

362

4. did

15.「入浴をする」 How often do you ______a bath?

1. get 2. do 3. take 4. give

16.「急落した」 The job market has ______a dive.

1. dropped 2. done 3. made 4. taken

17.「馴れ馴れしくしている」 I felt like he was ______liberties with me.

1. taking 2. having 3. doing 4. getting

18.「評判を得た」 McClinton ______a name for his musical mix: American.

1. took 2. gave 3. had 4. found

19.「大儲けした」 Traders ______a killing in gold market.

1. got 2. made 3. took 4. did

20.「首になった」 Why did the typist ______the sack?

1. became 2. got 3. took 4. had

21.「被害を与える」 These drugs may ______damage to liver and kidney.

1. suffer 2. cause 3. make 4. give

22.「注文した」

363

Paul just ______an order for a new iMac.

1. placed 2. did 3. made 4. took

23.「注文を受ける」 Marcello decided to ______an offer to work at the university's writing center.

1. accept 2. approve 3. receive 4. retain

24.「手術を受けた」 I've never ______an operation in my life.

1. taken 2. received 3. undergone 4. accepted

25.「ガンになった」 More than 10 million people ______cancer in 2000.

1. inflicted 2. developed 3. diagnosed 4. became

26.「近道した」 Cathy ______corners by driving on the grass and dirt.

1. shortened 2. cut 3. bend 4. curved

27.「結婚した」 Eminem has ______the knot with girlfriend Kim Mathers.

1. taken 2. traced 3. tried 4. tied

28.「目撃した」 Five University students ______witness to the impact of armed conflict on civilian populations.

1. had 2. took 3. bore 4. became

29.「機嫌をとる」 He ______court to her, but she rejects him.

364

1. pays 2. plays 3. prays 4. pleads

30.「協定する」 The Chancellor decided to ______a bargain with the NATO countries.

1. stand 2. trick 3. stretch 4. strike

31.「手紙を受け取った」 Did you ______a letter from him?

1. take 2. receive 3. catch 4. accept

32.「大評判となった」 The song was a great hit and ______a sensation.

1. became 2. came 3. turned 4. reached

33.「証言した」 Victoria ______evidence at a public hearing.

1. took 2. said 3. made 4. gave

34.「発見する」 Juston Seyfert is about to ______a discovery that will change his life.

1. take 2. make 3. create 4. do

35.「監視する」 Can you ______an eye on my bag, please?

1. do 2. get 3. keep 4. place

36.「投票する」 Over 2000 voters will ______a ballot in the election.

1. take 2. drop

365

3. cast 4. do

37.「メモをとる」 The medical students ______notes during lectures.

1. take 2. write 3. record 4. get

38.「許可を与える」 Air traffic specialists ______clearance to pilots for landing and takeoff.

1. accord 2. give 3. provide 4. make

39.「利用する」 Many students ______use of the computer as a tutorial aid.

1. have 2. do 3. get 4. make

40.「故障した」 The Red Cross vehicle ______a breakdown.

1. made 2. did 3. had 4. received

41.「質問した」 He ______questions and I answered them.

1. inquired 2. took 3. asked 4. did

42.「バランスをとる」 It’s important to ______a balance between work and play.

1. hold 2. keep 3. take 4. make

43.「仕事を得る」 It is getting more and more difficult to ______a job in Singapore.

1. gain 2. acquire 3. earn 4. get

366

44.「掃除をした」 Jones ______the cleaning.

1. did 2. made 3. took 4. conducted

45.「考慮している」 The companies are ______account of the requirements on management systems.

1. taking 2. giving 3. making 4. doing

46.「得点した」 Auburn ______a goal after 5 min of the game.

1. attained 2. did 3. gained 4. made

47.「自殺した」 He ______suicide in December at the age of 58.

1. did 2. committed 3. died 4. offended

48.「重視する」 The community needs to ______emphasis on unity.

1. make 2. locate 3. put 4. do

49.「商売した」 We've never ______business in the US before.

1. practiced 2. made 3. performed 4. done

50.「削減があった」 Last year we ______a cut in our budget.

1. had 2. took 3. made 4. became

51.「決定する」 The Constitutional Court will ______a decision in a few hours.

367

1. get 2. do 3. have 4. make

52.「相談した」 The king ______counsel with his lords as to the choice of the new minister.

1. held 2. made 3. did 4. took

53.「犠牲になる」 Don't ______victim to online credit card fraud.

1. fall 2. become 3. find 4. suffer

54.「米をつくる」 Why would farmers ______rice in the desert?

1. manufacture 2. make 3. grow 4. create

55.「許可を与えた」 Toby Hemenway, author of Gaia's Garden, ______permission to use comments from his book.

1. gave 2. allowed 3. presented 4. provided

56.「発言権がある」 Everybody ______a say in the making of the budget.

1. gets 2. has 3. gains 4. takes

57.「昼食をとった」 Roy ______lunch with his Director at Blowfish.

1. took 2. caught 3. had 4. got

58.「犠牲になる」 Have you ever had to ______a sacrifice for someone else?

1. get 2. make

368

3. have 4. become

59.「停止する」 Let's ______a stop to bullying at work.

1. make 2. do 3. lay 4. put

60.「染みをとる」 How can I ______a stain from carpet quickly?

1. remove 2. fix 3. take 4. abolish

61.「傾向がある」 Do you think Filipinos ______a tendency to be late?

1. give 2. have 3. present 4. possess

62.「新聞を取る」 I don't ______a newspaper, it is too expensive!

1. have 2. receive 3. take 4. make

63.「アポを取った」 A student has ______an appointment with his advisor.

1. took 2. reserved 3. made 4. received

64.「パスポートを取る」 Where can I find out how to ______a passport?

1. get 2. grab 3. take 4. issue

65.「裁判を受ける」 Jeffrey Dennis will ______trial for killing his wife.

1. sit 2. go 3. receive 4. stand

369

66.「学校をつくった」 In 1554 Jesuits ______a school to convert the Indians in the Sierra del Mar.

1. constituted 2. founded 3. got 4. made

ANSWERS:

1. 1. get a ticket 「チケットを得る」 2. 2. receive an invitation 「招待を受ける」 3. 1. become a teacher 「先生になる」 4. 4. do a project 「プロジェクトをする」 5. 2. have a class 「授業がある」 6. 3. have an ear 「才能がある」 7. 3. have an air 「雰囲気がある」 8. 4. get a break 「チャンスを得る」 9. 4. take a stand 「態度・立場をとる」 10. 4. get the nod 「承認を得る」 11. 2. take a walk 「散歩をする」 12. 1. make a copy 「コピーをとる」 13. 3. have an argument 「議論する」 14. 1. take an exam 「試験を受ける」 15. 3. take a bath 「入浴をする」 16. 4. take a dive 「急落する」 17. 1. take liberties 「馴れ馴れしくする」 18. 3. have a name 「評判を得る」 19. 2. make a killing 「大儲けする」 20. 2. get the sack 「首になる」 21. 2. cause damage 「被害を与える」 22. 1. place an order 「注文する」 23. 1. accept an offer 「注文を受ける」 24. 3. undergo an operation 「手術を受ける」 25. 2. develop cancer 「ガンになる」 26. 2. cut corners 「近道する」 27. 4. tie the knot 「結婚する」 28. 3. bear witness 「目撃する」 29. 1. pay court 「機嫌をとる」 30. 4. strike a bargain 「協定する」 31. 2. receive a letter 「手紙を受け取る」 32. 1. become a sensation 「大評判となる」 33. 4. give evidence 「証言する」 34. 2. make a discovery 「発見する」 35. 3. keep an eye 「監視する」 36. 3. cast a ballot 「投票する」 37. 1. take notes 「メモをとる」 38. 2. give clearance 「許可を与える」

370

39. 4. make use of 「利用する」 40. 3. have a breakdown 「故障する」 41. 3. ask questions 「質問する」 42. 2. keep a balance 「バランスをとる」 43. 4. get a job 「仕事を得る」 44. 1. do the cleaning 「掃除をする」 45. 1. take account 「考慮する」 46. 4. make a goal 「得点する」 47. 2. commit suicide 「自殺する」 48. 3. put emphasis 「重視する」 49. 4. do business 「商売する」 50. 1. have a cut 「削減がある」 51. 4. make a decision 「決定する」 52. 4. take counsel 「相談する」 53. 1. fall victim 「犠牲になる」 54. 3. grow rice 「米をつくる」 55. 1. give permission 「許可を与える」 56. 2. have a say 「発言権がある」 57. 3. have lunch 「昼食をとる」 58. 2. make a sacrifice 「犠牲になる」 59. 4. put a stop 「停止する」 60. 1. remove a stain 「染みをとる」 61. 2. have a tendency 「傾向がある」 62. 3. take a newspaper 「新聞を取る」 63. 3. make an appointment 「アポを取る」 64. 1. get a passport 「パスポートを取る」 65. 4. stand trial 「裁判を受ける」 66. 2. found a school 「学校をつくる」

371

XII: Questionnaire before the posttest (Experiments 3 and 4)

1．前回配布しましたハンドアウトを読んだり、練習問題を解いたり、WordChamp のオンラインフラッシュカードを使ったりしてみましたか？どのくらいやってみましたか？ (How much time did you spend in studying collocations using the handout, the exercise and the flashcards on the WordChamp?)

ハンドアウトを読む。(the handout) 時間(hours) 分(minutes)

練習問題をやる。(the exercises) 時間(hours) 分(minutes)

WordChamp の online flashcards を使う。(the online flashcards) 時間(hours) 分(minutes)

２．ハンドアウト、練習問題、WordChamp は役に立ちましたか？ (Were the handout, the exercises and the flashcards useful?)

ハンドアウト (the handout) 練習問題 (the exercise) WordChamp flashcards

・とても役に立った・とても役に立った・とても役に立った (very useful) (very useful) (very useful)

・役に立った・役に立った・役に立った (useful) (useful) (useful)

・あまり役に立たなかった・あまり役に立たなかった・あまり役に立たなかった (not very useful) (not very useful) (not very useful)

・ほとんど役に立たなかった・ほとんど役に立たなかった・ほとんど役に立たなかった (not useful) (not useful) (not useful)

・使っていないので分からない・使っていないので分からない・使っていないので分からない (didn’t use) (didn’t use) (didn’t use)

3．ハンドアウト、練習問題、WordChamp を使ってみた感想をお願いします。 (Any comment on the handout, the exercises and/or the WordChamp flashcards)

ハンドアウト (the handout)

練習問題 (the exercises)

WordChamp flashcards

372

Appendix XIII: Test sentences for Experiment 3

A1 frequent, same light verb • 「傾向がある」 Parents today have a tendency to view their children's behavior through psychological filters.

• 「許可を与える Absolutely no one can gain knowledge of your health information unless you give permission to do so.」

• 「商売する」 Tom is thinking of doing business in Italy.

• 「メモを取る」 You take notes so you will remember what you read or hear.

• 「手紙を受け取った」 Today, I received a letter dated May 13, 2004.

• 「仕事を得る」 Gaddis wants to get a job as the marketing director at a management-consulting firm in Atlanta.

A1 infrequent, same light verb • 「新聞をとらない（購読しない）」 Erica doesn’t take a newspaper or watch television.

• 「発言権がある」 You should have a say in how your pension is invested.

• 「削減があった」 We have had a cut in funding from the federal government.

• 「清掃をする」 What time should I do the cleaning?

• 「許可を与える」 The government refused to give clearance to prepared Swedish military rescue planes to aid in the rescue efforts from the tsunami disaster in Asia.

• 「大評判となった」 The discovery of Yakut diamonds in 1950s has become a sensation in the world.

A2 frequent, different light verb • 「アポを取った」 I have made an appointment to get my hair cut.

• 「昼食をとる」

373

Let's have lunch together.

• 「決定した」 Kate made a decision today to submit one of her stories to The New Yorker.

• 「考慮する」 Governments need to take account of potential long term structural developments and risks.

• 「利用する」 The most sophisticated work in music, film, and video often makes use of computer technology.

• 「証言する」 The case against Mr. Wanoghu was dropped in August 2004 after two key witnesses refused to give evidence.

A2 infrequent, different light verb • 「パスポートを取った」 I’ve got a passport, but are there any other entry requirements that I need to be aware of?

• 「故障した」 The machine has already had a breakdown.

• 「犠牲になる」 There are all sorts of other ways in which people can make a sacrifice for others.

• 「相談した」 Apollon and Poseidon took counsel to destroy the wall the Danaans had built.

• 「得点した」 Aaron Travis made a goal after receiving a pass from Kelly Cox.

• 「発見した」 The researchers at Dana-Farber Cancer Institute have made a discovery that could help solve a mystery in cancer biology.

A3 frequent, content verb • 「裁判を受ける」 A Chilean judge ordered Augusto Pinochet to stand trial on kidnapping charges.

• 「犠牲となった」 When you have fell victim to a crime you must notify the police as soon as possible.

• 「自殺する」 Why do men commit suicide more often than women do?

• 「質問した」 Nick asked questions regarding system restore in Windows XP.

374

• 「監視する」 Make sure you keep an eye on your belongings.

• 「停止する」 Ohio must put a stop to capital punishment.

A3 infrequent, content verb • 「投票する」 You have the right to cast a ballot if you are a registered voter.

• 「バランスをとる」 It takes a fair amount of discipline to keep a balance between work life and home life when your office is in your house.

• 「染みを取る」 Please help me remove a stain from a bath tub.

• 「米をつくる」 The geography of China provides an ideal environment for growing rice.

• 「重視している」 The university’s new MBA program puts emphasis on ethics.

• 「学校をつくった」 Howe founded a school for the deaf using the oral method in the late 1860's.

375

Appendix XIV: Data from Experiment 3

# of responses = number of responses produced by subjects (n = 60) GJ = Native speakers’ Grammaticality Judgments (6-point-scale) MC = Native speakers’ Meaning Conveyed (6-point-scale)

The combinations are listed from the highest to the lowest average scores of the GJ and the MC.

A1 frequent, same light verb Parents today have a tendency to view their children's behavior through psychological filters. # resp GJ MC # resp GJ MC have a tendency 55 6.00 6.00 do a tendency 1 1.17 1.83 got a tendency 1 1.17 3.00 make a tendency 1 1.17 1.83 take a tendency 1 1.33 2.33 NO ANSWER 1 – –

A1 frequent, same light verb Absolutely no one can gain knowledge of your health information unless you give permission to do so. # resp GJ MC # resp GJ MC give permission 41 6.00 6.00 have permission 1 1.17 2.33 allow permission 1 2.17 3.33 get permission 10 1.00 2.00 give a permission 1 2.33 3.17 make permission 1 1.00 1.83 admit permission 1 1.83 3.33 gain permission 1 1.00 1.17 license permission 1 2.00 3.00 take permission 2 1.00 1.17

A1 frequent, same light verb Tom is thinking of doing business in Italy. # resp GJ MC # resp GJ MC doing business 44 6.00 6.00 having business 5 2.33 3.33 starting business 2 2.50 3.83 owning business 1 1.83 3.50 running business 4 1.83 4.33 making business 2 2.17 2.67 giving business 1 1.17 1.83 opening business 1 2.17 2.50

A1 frequent, same light verb You take notes so you will remember what you read or hear. # resp GJ MC # resp GJ MC take notes 58 6.00 6.00 make notes 2 4.83 5.67

A1 frequent, same light verb Today, I received a letter dated May 13, 2004. # resp GJ MC # resp GJ MC received a letter56 6.00 6.00 got a letter 4 5.67 5.83

A1 frequent, same light verb Gaddis wants to get a job as the marketing director at a management-consulting firm in Atlanta. # resp GJ MC # resp GJ MC get a job 55 6.00 6.00 earn a job 1 4.17 5.17 find a job 1 5.00 5.83 have a job 1 1.67 3.33 obtain a job 1 4.83 5.83

376

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A1 infrequent, same light verb Erica doesn’t take a newspaper or watch television. # resp GJ MC # resp GJ MC take a newspaper 34 6.00 6.00 receive a newspaper 3 2.33 3.00 get a newspaper 5 4.50 5.67 have a newspaper 1 1.83 2.17 buy a newspaper 3 2.50 3.50 read a newspaper 7 1.83 2.83 purchase a newspaper 3 2.50 3.50 subscribe a newspaper 4 1.33 1.50

A1 infrequent, same light verb You should have a say in how your pension is invested. # resp GJ MC # resp GJ MC have a say 56 6.00 6.00 make a say 2 1.33 3.00 get a say 1 5.00 5.00 take a say 1 1.50 2.50

A1 infrequent, same light verb We have had a cut in funding from the federal government. # resp GJ MC # resp GJ MC had a cut 30 6.00 6.00 got a cut 4 3.17 3.67 took a cut 5 5.17 5.33 cost a cut 1 1.00 1.33 experienced a cut 1 5.00 4.00 down a cut 1 1.00 1.33 made a cut 17 3.50 3.50 worked a cut 1 1.00 1.33

A1 infrequent, same light verb What time should I do the cleaning? # resp GJ MC # resp GJ MC do the cleaning 50 6.00 6.00 clean the cleaning 1 1.00 1.50 make the cleaning 6 1.17 3.33 give the cleaning 1 1.00 1.50 have the cleaning 1 1.50 1.50 NO ANSWER 1 – –

A1 infrequent, same light verb The government refused to give clearance to prepared Swedish military rescue planes to aid in the rescue efforts from the tsunami disaster in Asia. # resp GJ MC # resp GJ MC give clearance 49 6.00 6.00 take clearance 1 1.50 1.67 offer clearance 1 2.83 4.00 drop clearance 1 1.17 1.33 admit clearance 2 3.33 3.33 NO ANSWER 1 – – make clearance 5 1.67 2.67

A1 infrequent, same light verb The discovery of Yakut diamonds in 1950s has become a sensation in the world. # resp GJ MC # resp GJ MC become a sensation 22 6.00 6.00 done a sensation 1 1.50 2.17 made a sensation 15 3.00 3.00 given a sensation 1 1.17 1.67 created a sensation 1 2.83 3.67 got a sensation 1 1.17 1.50 been a sensation 2 2.50 3.50 paid a sensation 1 1.17 1.50 brought a sensation 1 2.83 3.00 taken a sensation 5 1.17 1.50 had a sensation 6 1.67 3.00 NO ANSWER 3 – – hit a sensation 1 1.17 3.17

377

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A2 frequent, different light verb I have made an appointment to get my hair cut. # resp GJ MC # resp GJ MC made an appointment 53 6.00 6.00 taken an appointment 4 3.17 3.50 got an appointment 1 4.67 5.67 stopped an appointment 1 1.00 1.17 had an appointment 1 3.17 4.17

A2 frequent, different light verb Let's have lunch together. # resp GJ MC # resp GJ MC have lunch 54 6.00 6.00 have a lunch 1 4.33 5.17 eat lunch 3 6.00 6.00 make lunch 1 1.50 2.33 get lunch 1 4.33 5.33

A2 frequent, different light verb Kate made a decision today to submit one of her stories to The New Yorker. # resp GJ MC # resp GJ MC made a decision57 6.00 6.00 got a decision 1 1.83 3.33 had a decision 2 2.00 3.50 trial a decision 1 1.00 1.17

A2 frequent, different light verb Governments need to take account of potential long term structural developments and risks. # resp GJ MC # resp GJ MC take account 33 6.00 6.00 think account 1 1.17 2.67 take an account 1 3.33 4.33 do account 1 1.33 1.67 take into account 3 2.50 3.33 pay account 1 1.17 1.67 make account 12 2.17 3.50 have account 2 1.00 1.50 consider account 6 1.83 3.00

A2 frequent, different light verb The most sophisticated work in music, film, and video often makes use of computer technology. # resp GJ MC # resp GJ MC makes use 47 6.00 6.00 has use 2 2.67 3.33 makes a use 1 4.33 5.33 does use 1 1.00 1.50 takes use 3 3.00 3.33 NO ANSWER 3 – – gets use 3 1.83 4.17

A2 frequent, different light verb The case against Mr Wanoghu was dropped in August 2004 after two key witnesses refused to give evidence. # resp GJ MC # resp GJ MC give evidence 30 6.00 6.00 tell evidence 2 1.17 3.00 provide evidence 1 5.17 5.67 say evidence 1 1.50 2.33 show evidence 2 5.00 5.33 make evidence 7 1.17 2.33 bare evidence 2 4.00 5.00 speak evidence 1 1.33 1.83 have evidence 4 3.83 4.83 do evidence 1 1.17 1.83 prove evidence 2 2.67 3.00 stand evidence 3 1.00 1.00 propose evidence 1 2.17 3.00 NO ANSWER 2 – – take evidence 1 2.33 2.83

378

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A2 infrequent, different light verb I’ve got a passport, but are there any other entry requirements that I need to be aware of? # resp GJ MC # resp GJ MC got a passport 43 6.00 6.00 issued a passport 1 1.33 1.50 acquired a passport 1 5.50 5.83 taken a passport 7 1.17 1.50 obtained a passport 2 5.00 5.83 made a passport 1 1.00 1.17 received a passport 3 4.50 5.00 NO ANSWER 1 – – had a passport 1 1.67 2.50

A2 infrequent, different light verb The machine has already had a breakdown. # resp GJ MC # resp GJ MC had a breakdown 30 6.00 6.00 given a breakdown 1 1.33 2.00 experienced a breakdown 1 5.50 5.67 become a breakdown 3 1.33 1.83 gone through a breakdown 1 2.33 3.33 made a breakdown 7 1.00 1.17 undertook a breakdown 1 1.67 2.50 doesn’t a breakdown 1 1.00 1.00 been a breakdown 2 1.33 2.67 gone a breakdown 1 1.00 1.00 taken a breakdown 2 1.50 2.50 NO ANSWER 2 – – got a breakdown 8 1.33 2.33

A2 infrequent, different light verb There are all sorts of other ways in which people can make a sacrifice for others. # resp GJ MC # resp GJ MC make a sacrifice 26 6.00 6.00 undertake a sacrifice 1 2.00 1.83 give a sacrifice 3 4.67 4.83 fall a sacrifice 1 1.17 1.67 do a sacrifice 2 3.00 4.00 cause a sacrifice 1 1.17 1.17 take a sacrifice 2 2.50 2.67 get a sacrifice 1 1.17 1.17 be a sacrifice 4 2.33 2.67 have a sacrifice 3 1.00 1.33 become a sacrifice 12 2.17 2.83 NO ANSWER 3 – – suffer a sacrifice 1 2.33 2.67

A2 infrequent, different light verb Apollon and Poseidon took counsel to destroy the wall the Danaans had built. # resp GJ MC # resp GJ MC took counsel 27 6.00 6.00 got counsel 1 2.17 3.00 held counsel 1 5.50 5.50 met counsel 1 1.50 2.00 had counsel 14 4.00 4.67 asked counsel 3 1.17 2.00 made counsel 8 2.17 3.33 told counsel 3 1.00 1.00 did counsel 1 1.50 3.67 NO ANSWER 1 – –

A2 infrequent, different light verb Aaron Travis made a goal after receiving a pass from Kelly Cox. # resp GJ MC # resp GJ MC made a goal 41 6.00 6.00 had a goal 1 3.33 3.50 scored a goal 2 5.17 5.67 hit a goal 1 3.33 3.50 got a goal 11 4.17 5.00 did a goal 1 1.00 2.83 obtained a goal 1 3.67 4.67 took a goal 1 1.00 1.50 achieved a goal 1 3.33 3.50

379

A2 infrequent, different light verb The researchers at Dana-Farber Cancer Institute have made a discovery that could help solve a mystery in cancer biology. # resp GJ MC # resp GJ MC made a discovery 44 6.00 6.00 did a discovery 1 1.33 2.50 had a discovery 7 4.67 4.83 got a discovery 1 1.50 2.33 found a discovery 5 3.83 5.00 took a discovery 2 1.00 1.00

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A3 frequent, content verb A Chilean judge ordered Augusto Pinochet to stand trial on kidnapping charges. # resp GJ MC # resp GJ MC stand trial 20 6.00 6.00 take a trial 1 1.00 1.17 have a trial 1 3.33 4.67 make trial 1 1.00 1.00 have trial 6 3.33 4.00 put trail 1 1.00 1.00 receive trial 9 2.50 3.33 show trial 1 1.00 1.00 stand the trial 1 2.17 2.33 set trial 1 1.00 1.00 get trial 2 1.33 2.50 understand trial 1 1.00 1.00 take trial 14 1.17 1.50 NO ANSWER 1 – –

A3 frequent, content verb When you have fell victim to a crime you must notify the police as soon as possible. # resp GJ MC # resp GJ MC fell victim 19 6.00 6.00 turned to be victim 1 1.67 2.67 became victim 25 4.83 5.67 had victim 2 1.17 1.50 been victim 1 4.50 4.67 made victim 5 1.00 1.50 stood victim 1 2.67 4.00 done victim 1 1.17 1.17 taken victim 1 2.33 2.50 put victim 1 1.00 1.17 got victim 2 2.00 2.67 NO ANSWER 1 – –

A3 frequent, content verb Why do men commit suicide more often than women do? # resp GJ MC # resp GJ MC commit suicide 51 6.00 6.00 die suicide 1 1.17 1.83 do suicide 2 3.67 3.83 make suicide 2 1.00 1.33 go suicide 1 1.50 1.83 take suicide 3 1.17 1.50

A3 frequent, content verb Nick asked questions regarding system restore in Windows XP. # resp GJ MC # resp GJ MC asked questions46 6.00 6.00 had questions 6 3.50 3.83 raised questions 1 5.00 5.83 made questions 7 1.50 2.17

A3 frequent, content verb Make sure you keep an eye on your belongings. # resp GJ MC # resp GJ MC keep an eye 52 6.00 6.00 have an eye 5 2.67 3.33 put an eye 2 3.50 4.17 see an eye 1 1.17 2.17

A3 frequent, content verb Ohio must put a stop to capital punishment.

380

# resp GJ MC # resp GJ MC put a stop 27 6.00 6.00 make a stop 12 1.50 2.83 call a stop 1 2.67 4.17 keep a stop 1 1.83 1.50 get a stop 2 2.67 3.83 take a stop 6 1.17 1.33 place a stop 1 2.50 4.00 break a stop 1 1.00 1.17 have a stop 1 1.33 3.83 stop a stop 1 1.00 1.00 do a stop 5 1.17 3.33 NO ANSWER 1 – – give a stop 1 2.33 2.33

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A3 infrequent, content verb You have the right to cast a ballot if you are a registered voter. # resp GJ MC # resp GJ MC cast a ballot 22 6.00 6.00 elect a ballot 1 1.50 1.50 place a ballot 1 4.33 5.17 have a ballot 2 1.50 1.50 drop a ballot 3 3.33 3.67 vote a ballot 4 1.00 2.00 give a ballot 5 3.17 3.33 fall a ballot 1 1.00 1.50 make a ballot 6 1.50 2.83 court a ballot 1 1.00 1.00 put a ballot 1 1.50 2.00 talk a ballot 1 1.00 1.00 take a ballot 9 1.33 2.17 NO ANSWER 2 – – bare a ballot 1 1.00 2.00

A3 infrequent, content verb It takes a fair amount of discipline to keep a balance between work life and home life when your office is in your house. # resp GJ MC # resp GJ MC keep a balance 40 6.00 6.00 make a balance 5 1.83 3.17 maintain a balance 2 5.80 5.83 set a balance 1 2.17 2.50 have a balance 2 4.17 4.67 take a balance 10 1.00 1.33

A3 infrequent, content verb Please help me remove a stain from a bath tub. # resp GJ MC # resp GJ MC remove a stain 50 6.00 6.00 wipe a stain 1 2.67 3.33 get rid of a stain 1 5.17 5.67 do a stain 2 1.17 2.00 take off a stain 2 3.67 4.83 get a stain 1 1.17 1.50 reduce a stain 1 2.67 3.50 take a stain 2 1.00 1.17

A3 infrequent, content verb The geography of China provides an ideal environment for growing rice. # resp GJ MC # resp GJ MC growing rice 43 6.00 6.00 making rice 7 2.17 3.83 farming rice 1 5.67 5.83 bearing rice 1 1.00 1.17 harvesting rice 2 4.67 5.00 cooking rice 1 1.00 1.00 producing rice 3 4.33 4.83 NO ANSWER 1 – – manufacturing rice 1 3.00 3.83

A3 infrequent, content verb The university’s new MBA program puts emphasis on ethics. # resp GJ MC # resp GJ MC puts emphasis 29 6.00 6.00 makes emphasis 8 2.00 3.67

381 puts an emphasis 1 5.50 5.67 does emphasis 4 1.00 2.67 places emphasis 2 5.17 5.67 gets emphasis 1 1.33 2.33 focuses emphasis 1 3.67 5.17 takes emphasis 4 1.67 1.83 has emphasis 9 3.17 4.17 thinks emphasis 1 1.00 2.17

A3 infrequent, content verb Howe founded a school for the deaf using the oral method in the late 1860's. # resp GJ MC # resp GJ MC founded a school 36 6.00 6.00 made a school 2 4.17 4.83 established a school 9 5.67 5.67 NO ANSWER 1 – – built a school 12 4.17 4.83

382

Appendix XV. Table from Experiment 3

Appendix XV-1: Percentages of accurate collocations of each type in Experiments 1 and 3. (Figures 6-1, 6-2 and 6-3)

Low- Advanced learner Total (# = 60) intermediate (# = 30) learners (# = 30) Pretest Posttest Pretest Posttest Pretest Posttest (Exp.1) (Exp. 3) (Exp. 1) (Exp. 3) (Exp. 1) (Exp. 3) A 1 Frequent 83.89 % 87.22 % 86.67 % 91.11 % 85.28 % 89.17 % Same Infrequent light v 53.33 % 61.67 % 51.11 % 75.00 % 52.22 % 68.33 % A2 Frequent 46.67% 68.33 % 71.67 % 86.11 % 59.17 % 77.22 % Different Infrequent 32.22 % 48.89 % 51.11 % 71.67 % 41.67 % 60.28 % light V A 3 Frequent 31.67 % 51.67 % 44.44 % 68.33 % 38.05 % 60.00 % Content V Infrequent 31.67 % 57.22 % 46.11 % 76.67 % 38.89 % 66.94 % TABLE XV-1: Percentages of accurate collocations of each type in Experiments 1 and 3. (Figures 6-1, 6-2 and 6-3)

383

Appendix XVI: Test sentences for Experiment 4

A1 literal, same light verb • 「プロジェクトをする」 We want to do a project on the effects of volcanoes on the environment.

• 「招待を受けた」 Charles received an invitation to the wedding of Laura Drum and Andy Miller.

• 「先生になった」 Esther became a teacher of the guitar at age 40.

• 「授業がある」 Ihad a class on Fridays from 9:00-11:00.

• 「チケットを得る」 It’s hard to get a ticket for the Football World Cup 2006.

A1 abstract, same light verb • 「承認を得た」 Lindsey began making plans as soon as she got the nod from the executive committee.

• 「チャンスを得た」 In 1999 Amy got a break when she landed a major role on the series ‘Angel’.

• 「雰囲気がある」 She has an air of confidence and joy in what she does.

• 「才能がない」 I'm not Scottish and therefore don't have an ear for traditional Scottish dance music.

• 「態度・立場をとった」 About 40 thousand people gathered and took a stand against injustice.

A2 literal, different light verb • 「散歩をした」 Yesterday I took a walk by the river.

• 「試験を受ける」 Please bring an identification card, when you are going to take an exam.

• 「議論していた」 Graham and I were having an argument about American imperialism.

• 「コピーを取った」 Isabel made a copy of the document and shows it to Claire.

• 「入浴をした」

384

After dinner Susan took a bath and brushed her teeth.

A2 abstract, different light verb • 「馴れ馴れしくしている」 She is taking liberties with her friend by always borrowing her car.

• 「急落した」 Sales of sport-utility vehicle took a dive in September.

• 「大儲けしている」 Not everyone is making a killing in real estate.

• 「評判を得た」 Wageningen University has a name for innovative and often path-breaking research in the field of life sciences and natural resources.

• 「首になった」 Suresh got the sack for sleeping on the job.

A3 literal, content verb • 「手術を受けた」 Saddam Hussein underwent an operation to repair a hernia about 10 days ago.

• 「注文した」 I placed an order on the website, but haven't received my goods.

• 「被害を与えている」 Acid rain has been causing damage to forests, lakes and marshes.

• 「ガンになった」 Many of the workers developed cancer because they were exposed to radiation, beryllium or silica in the course of doing their jobs.

• 「注文を受ける」 My sellers didn't accept an offer that I thought was terrific.

A3 abstract, content verb • 「協定する」 The policymaker wanted to strike a bargain with school districts and provide more funding for improved performance.

• 「機嫌をとった」 They all paid court to Nicholas, flattered him, and, knowingly or unknowingly, helped inflate his ego.

• 「目撃した」 The second half of the twentieth century bore witness to significant decline in mortality among children in the developing world.

• 「近道した」

385

We cut corners to get to the destination on time.

• 「結婚する」 Solomon Maluka, a bachelor in his late 20s, says he will tie the knot when the time is right.

386

Appendix XVII: Data from Experiment 4

# of responses = number of responses produced by subjects (n = 60) GJ = Native speakers’ Grammaticality Judgments (6-point-scale) MC = Native speakers’ Meaning Conveyed (6-point-scale)

The combinations are listed from the highest to the lowest average scores of the GJ and the MC.

A1 literal, same light verb We want to do a project on the effects of volcanoes on the environment # resp GJ MC # resp GJ MC do a project 48 6.00 6.00 establish a project 1 2.83 3.33 make a project 2 4.00 4.50 present a project 1 2.67 3.17 participate in a project 1 4.17 4.33 take a project 1 2.00 2.50 have a project 5 3.67 4.00 proceed a project 1 1.50 1.67

A1 literal, same light verb Charles received an invitation to the wedding of Laura Drum and Andy Miller. # resp GJ MC # resp GJ MC received an invitation 46 6.00 6.00 had an invitation 3 1.83 2.50 got an invitation 8 5.83 6.00 made an invitation 1 1.00 1.33 took an invitation 2 2.83 3.17

A1 literal, same light verb Esther became a teacher of the guitar at age 40. # resp GJ MC # resp GJ MC became a teacher 57 6.00 6.00 did a teacher 2 1.00 1.33 was a teacher 1 3.67 3.83

A1 literal, same light verb I had a class on Fridays from 9:00-11:00. # resp GJ MC # resp GJ MC had a class 58 6.00 6.00 took a class 2 5.00 4.83

A1 literal, same light verb It’s hard to get a ticket for the Football World Cup 2006. # resp GJ MC # resp GJ MC get a ticket 54 6.00 6.00 earn a ticket 1 2.00 3.33 receive a ticket 1 4.17 4.50 take a ticket 1 2.50 2.50 obtain a ticket 2 4.17 4.17 find a ticket 1 1.67 3.00

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A1 abstract, same light verb Lindsey began making plans as soon as she got the nod from the executive committee. # resp GJ MC # resp GJ MC got the nod 36 6.00 6.00 gave the nod 3 1.00 1.17 received the nod 2 4.33 4.00 made the nod 2 1.00 1.17

387 had the nod 5 1.83 3.67 took the nod 8 1.00 1.17 heard the nod 1 1.00 2.17 tied the nod 1 1.00 1.00 kept the nod 1 1.00 1.50 trial the nod 1 1.00 1.00

A1 abstract, same light verb In 1999 Amy got a break when she landed a major role on the series ‘Angel’. # resp GJ MC # resp GJ MC got a break 34 6.00 6.00 made a break 7 1.17 1.67 had a break 8 2.83 4.00 gave a break 2 1.00 1.67 earned a break 1 3.00 3.67 placed a break 1 1.00 1.17 gained a break 1 1.83 2.83 took a break 6 1.00 1.00

A1 abstract, same light verb She has an air of confidence and joy in what she does. # resp GJ MC # resp GJ MC has an air 52 6.00 6.00 takes an air 1 1.33 1.67 creates an air 1 2.50 3.00 gets an air 2 1.33 1.50 shows an air 1 1.67 2.33 keeps an air 1 1.33 1.50 is an air 1 1.50 2.33 makes an air 1 1.33 1.50

A1 abstract, same light verb I'm not Scottish and therefore don't have an ear for traditional Scottish dance music. # resp GJ MC # resp GJ MC have an ear 56 6.00 6.00 get an ear 1 1.33 2.33 keep an ear 2 2.17 2.33 take an ear 1 1.00 1.50

A1 abstract, same light verb About 40 thousand people gathered and took a stand against injustice. # resp GJ MC # resp GJ MC took a stand 34 6.00 6.00 did a stand 2 1.83 2.50 made a stand 9 5.00 5.50 got a stand 3 1.17 2.17 kept a stand 3 2.00 2.67 stood a stand 1 1.00 2.33 had a stand 7 1.83 2.83 NO ANSWER 1 – –

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A2 literal, different light verb Yesterday I took a walk by the river. # resp GJ MC # resp GJ MC took a walk 54 6.00 6.00 trial a walk 1 1.00 1.17 had a walk 5 2.67 4.33

A2 literal, different light verb Please bring an identification card, when you are going to take an exam. # resp GJ MC # resp GJ MC take an exam 59 6.00 6.00 have an exam 1 3.33 3.33

A2 literal, different light verb Graham and I were having an argument about American imperialism. # resp GJ MC # resp GJ MC having an argument 33 6.00 6.00 taking an argument 1 1.50 2.17

388 making an argument 18 2.17 2.67 doing an argument 5 1.00 2.50 discussing an argument 3 1.17 2.50

A2 literal, different light verb Isabel made a copy of the document and shows it to Claire. # resp GJ MC # resp GJ MC made a copy 51 6.00 6.00 had a copy 2 1.67 2.17 did a copy 1 3.17 3.67 took a copy 6 1.17 1.67

A2 literal, different light verb After dinner Susan took a bath and brushed her teeth. # resp GJ MC # resp GJ MC took a bath 59 6.00 6.00 had a bath 1 4.67 5.67

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A2 abstract, different light verb She is taking liberties with her friend by always borrowing her car. # resp GJ MC # resp GJ MC taking liberties 30 6.00 6.00 being liberties 2 1.83 2.17 making liberties 8 2.83 3.33 acting liberties 1 1.33 2.50 doing liberties 2 2.67 3.17 keeping liberties 3 1.50 2.33 having liberties 4 2.67 3.17 becoming liberties 1 1.50 1.67 showing liberties 1 2.33 3.33 receiving liberties 1 1.17 1.50 using liberties 1 2.00 2.33 NO ANSWER 6 – –

A2 abstract, different light verb Sales of sport-utility vehicle took a dive in September. # resp GJ MC # resp GJ MC took a dive 25 6.00 6.00 down a dive 1 1.33 2.00 made a dive 7 4.50 5.00 fell a dive 1 1.17 1.83 did a dive 2 3.83 4.50 lost a dive 1 1.17 1.67 had a dive 8 2.33 3.17 went a dive 1 1.17 1.67 showed a dive 1 2.17 2.50 got a dive 3 1.17 1.33 dropped a dive 6 1.17 2.50 kept a dive 1 1.17 1.33 received a dive 1 1.33 2.17 NO ANSWER 2 – –

A2 abstract, different light verb Not everyone is making a killing in real estate. # resp GJ MC # resp GJ MC making a killing 44 6.00 6.00 giving a killing 1 1.67 2.17 earning a killing 1 5.33 5.33 keeping a killing 1 1.67 1.67 getting a killing 7 2.67 3.67 NO ANSWER 1 – – having a killing 5 2.50 2.67

A2 abstract, different light verb Wageningen University has a name for innovative and often path-breaking research in the field of life sciences and natural resources. # resp GJ MC # resp GJ MC has a name 36 6.00 6.00 made a name 2 2.67 3.67 gained a name 2 4.83 5.00 took a name 2 2.50 3.83

389 received a name 5 3.50 4.33 gave a name 1 1.33 1.50 got a name 11 3.33 4.00

A2 abstract, different light verb Suresh got the sack for sleeping on the job. # resp GJ MC # resp GJ MC got the sack 42 6.00 6.00 made the sack 1 1.50 1.83 received the sack 1 3.67 5.00 put the sack 1 1.50 1.67 had the sack 4 2.00 2.67 gave the sack 1 1.33 1.83 packed the sack 1 2.00 2.00 was tired the sack 1 1.50 1.33 took the sack 6 1.33 2.33 did the sack 2 1.00 1.00 fired the sack 1 1.50 1.83

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A3 literal, content verb Saddam Hussein underwent an operation to repair a hernia about 10 days ago. # resp GJ MC # resp GJ MC underwent an operation 28 6.00 6.00 undertook an operation 1 2.00 2.00 got an operation 1 5.00 5.17 took an operation 10 1.50 2.33 had an operation 12 5.00 5.17 ground an operation 1 1.00 1.67 received an operation 4 4.50 5.00 underdid an operation 1 1.17 1.17 obtained an operation 1 2.17 3.00 undergot an operation 1 1.00 1.17

A3 literal, content verb I placed an order on the website, but haven't received my goods. # resp GJ MC # resp GJ MC placed an order 22 6.00 6.00 took an order 4 1.83 2.00 made an order 25 4.83 4.83 ordered an order 1 1.33 1.67 put an order 1 1.83 3.17 told an order 1 1.17 1.83 gave an order 2 2.00 2.17 offered an order 1 1.00 1.00 had an order 3 1.83 2.33

A3 literal, content verb Acid rain has been causing damage to forests, lakes and marshes. # resp GJ MC # resp GJ MC causing damage 35 6.00 6.00 giving damage 15 3.00 4.17 doing damage 2 5.33 5.83 getting damage 2 1.33 2.17 making damage 2 4.67 5.00 having damage 2 1.33 1.83 creating damage 1 3.17 4.33 receiving damage 1 1.33 1.83

A3 literal, content verb Many of the workers developed cancer because they were exposed to radiation, beryllium or silica in the course of doing their jobs. # resp GJ MC # resp GJ MC developed cancer 34 6.00 6.00 became cancer 7 1.33 2.83 got cancer 12 5.67 5.67 NO ANSWER 2 – – had cancer 5 4.83 4.83

A3 literal, content verb My sellers didn't accept an offer that I thought was terrific.

390

# resp GJ MC # resp GJ MC accept an offer 24 6.00 6.00 place an offer 5 1.67 1.67 take an offer 8 4.17 4.67 offer an offer 1 1.33 1.67 get an offer 5 3.67 4.50 order an offer 2 1.33 1.67 have an offer 2 2.33 3.17 put an offer 1 1.17 1.83 receive an offer 8 2.17 2.67 make an offer 4 1.67 1.17

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

A3 abstract, content verb The policymaker wanted to strike a bargain with school districts and provide more funding for improved performance. # resp GJ MC # resp GJ MC strike a bargain 13 6.00 6.00 take a bargain 5 1.33 2.50 make a bargain 16 6.00 6.00 give a bargain 1 1.33 1.83 do a bargain 1 2.83 4.00 stand a bargain 4 1.00 2.00 set a bargain 1 3.00 3.50 keep a bargain 1 1.33 1.50 have a bargain 9 2.67 3.00 trial a bargain 1 1.00 1.17 get a bargain 2 1.83 2.17 NO ANSWER 5 – – cooperate a bargain 1 1.83 2.00

A3 abstract, content verb They all paid court to Nicholas, flattered him, and, knowingly or unknowingly, helped inflate his ego. # resp GJ MC # resp GJ MC paid court 19 6.00 6.00 put court 3 1.67 2.33 gave court 1 3.83 4.33 brought court 1 1.33 2.50 made court 7 2.67 4.00 stayed court 1 1.50 1.67 took court 16 2.67 2.67 balance court 1 1.33 1.50 had court 6 2.33 2.67 played court 1 1.33 1.33 got court 1 2.00 2.50 asked court 1 1.17 1.33 did court 1 2.00 2.17 NO ANSWER 1 – –

A3 abstract, content verb The second half of the twentieth century bore witness to significant decline in mortality among children in the developing world. # resp GJ MC # resp GJ MC bore witness 20 6.00 6.00 saw witness 1 2.00 2.17 became witness 5 3.00 3.33 gave witness 2 1.83 2.33 paid witness 1 2.33 3.00 did witness 1 1.33 2.67 took witness 3 2.33 2.83 kept witness 1 1.33 2.17 got witness 3 1.67 2.83 had witness 12 1.50 1.50 made witness 4 2.17 2.67 NO ANSWER 5 – – watched witness 2 1.83 2.83

A3 abstract, content verb We cut corners to get to the destination on time. # resp GJ MC # resp GJ MC cut corners 49 6.00 6.00 avoided corners 1 1.17 1.17 short cut corners 1 2.00 3.67 got corners 1 1.17 1.17 walked corners 1 1.50 1.67 short corners 1 1.00 1.17 took corners 3 1.33 1.50 made corners 3 1.17 1.00

391

A3 abstract, content verb Solomon Maluka, a bachelor in his late 20s, says he will tie the knot when the time is right. # resp GJ MC # resp GJ MC tie the knot 38 6.00 6.00 get the knot 6 1.17 1.50 make the knot 5 1.50 2.33 have the knot 1 1.17 1.17 take the knot 3 1.67 2.00 keep the knot 1 1.00 1.17 marry the knot 1 1.50 1.67 receive the knot 1 1.00 1.17 tighten the knot 1 1.33 1.50 NO ANSWER 2 – – be the knot 1 1.50 1.17

392

Appendix XVIII Table from Experiment 4

Appendix XVIII-1: Percentages of accurate collocations of each type in Experiments 2 and 4. (Figures 6-4, 6-5, 6-6)

Low-intermediate Advanced learner Total (60) learners (# = 30) (30) Pretest Posttest Pretest Posttest Pretest Posttest Exp. 2 Exp. 4 Exp. 2 Exp. 4 Exp. 2 Exp. 4 A 1 Literal 78.00 % 89.33 % 86.00 % 92.67 % 82.00 % 91.00 % Same Abstract 68.00 % 73.33 % 77.33 % 74.00 % 72.67 % 73.67 % light V A 2 Literal 80.00 % 83.33 % 89.33 % 87.33 % 82.67 % 85.33 % Different Abstract 19.33 % 50.67 % 24.00 % 68.00 % 21.67 % 59.34 % light V A 3 Literal 24.00 % 46.67 % 32.67 % 64.67 % 28.34 % 55.67 % Content V Abstract 26.67 % 45.33 % 35.33 % 58.00 % 31.00 % 51.67 % TABLE XVIII-1: Percentages of accurate collocations of each type in Experiments 2 and 4. (Figures 6-4, 6-5, 6-6)

393

REFERENCES

Aitchison, J. (1987). Reproductive furniture and extinguished professors. In R. Steele & T. Threadgold (Eds.), Language topics: Essays in honour of Michael Halliday Vol. 2 (pp. 24-49). Amsterdam: John Benjamins.

Akhmanova, O. (1974). Word-Combination: Theory and Method. Moscow: Moscow University Press.

Allen, S. E. M. (1998). Categories within the verb category: learning the causative in Inuktitut. Linguistics, 36, 627-32.

Bahns, J. (1993). Lexical collocations: A contrastive view. ELTJ, 47(1), 56-63.

Bahns, J., and Eldaw, M. (1990). Should we teach EFL students collocations? Paper given at 9th World Congress of Applied Linguistics, Thessaloniki.

Bardovi-Harlig, K. (2008). Between pragmatics and grammar: The L2 acquisition of formulas. A paper presented at the Second Language Research Forum. University of Hawai‘i at Mānoa, Honolulu, Hawai‘i.

Barlow, M. (2000). Usage, blends, and grammar. In M. Barlow and S. Kemmer (Eds.), Usage-based models of language (pp. 315-345). Stanford, CA, CSLI Publications.

Benson, M. (1985). Collocations and idioms. In R Ilson (Ed.), Dictionaries , lexicography and language learning. (pp. 61 – 68). Oxford: Pergamon Press.

Benson, M., Benson, E., and Ilson, R. (1986). The BBI Combinatory dictionary of English word combinations. Amsterdam/Philadelphia, John Benjamin’s Publishing Company.

Bentivogli, L., and Pianta, E. (2003). Beyond lexical units: Enriching wordnets with phrasets. In proceedings of the research note sessions of the 10th conference of the European chapter of the association for computational linguistics (EACL’03), 67-70. Budapest, Hungary, April 2003.

Biskup, D. (1992). L1 influence on learners’ renderings of English collocations: A Polish/German empirical study, In Arnaud and Bejoint (Eds), Vocabulary and applied linguistics. London: Macmillan.

Bley-Vroman, R. (1996). Patterns, constructions, and SLA theory. Talk given at the University of Hawai‘i, January 31, 1996.

Bloom, L. (1973). One word at a time: The use of single word utterances before syntax. The Hague, Morton.

Bloom, L., Lightbrown, P., & Hood, L. (1975). Structure and variation in child language. Monographs of the Society for Research in Child Development, 40 (2, Serial No.

394

160).

Bolander, M. (1989). Prefabs, patterns and rules in interaction? Formulaic speech in adult learners’ L2 Swedish. In K. Hyltenstam and L. K. Obler (Eds.), Bilingualism across the lifespam. Cambridge, Cambridge University Press, 73-86.

Bohn, O. –S. (1986). Formulas, frame structures, and stereotypes in early syntactic development: some new evidence from L2 acquisition. Linguistics 24: 185-202.

Bonk, W. J. (1996). Testing ESL learners’ knowledge of collocations. Unpublished master’s thesis, University of Hawai‘i.

Braine, M. D. S. (1976). Children’s first word combinations. Monographs of the Society for Research in Child Development. 41, 1.

Brown, R. (1973). A First Language, the Early Stages. Cambridge, MA, Harvard University Press.

Caroli, M. T. (1998). Relating collocations to foreign language learning. Unpublished MA thesis. University of Reading. Department of Linguistic Science.

Chujo, K., Yoshimori, T., Hasegawa, S., Nishigaki, C., & Yamazaki, A. (2007). The vocabulary of selected high school English textbooks. Nihon daigaku seisan kogyo bu kenkyu hokoku B, 40. http://www.cit.nihon- u.ac.jp/kenkyu/kankoubutsu/houkoku_b/Vol.40/houkoku_b40.7.pdf

Clark, R. (1974). Performing without competence. Journal of Child Language, 1, 1-10.

Clément, R. (1986). Second language proficiency and acculturation: an investigation of the effects of language status and individual characteristics, Journal of Language and Social Psychology, 5: 271-90.

Collins Cobuild. (1999). Collins Cobuild English words in use: A dictionary of collocations. London, Harper Collins Publishers.

Cowie, A. P. (1994). Phraseology, in R. E. Asher & J. Simpson (Eds.), The Encyclopedia of Language and Linguistics. Oxford: Pergamon Press, 3168-71.

Cruttenden, A. (1981). Item-learning and system-learning. Journal of Psycholinguistic Research, 10, 1, 79-88.

Crystal, D. (1997). A dictionary of linguistics and phonology. 4th edition. Cambridge, MA: Blackwell Publishers.

Dechert, H., & Lennon, P. (1989). Collocational blends of advanced language learners: A preliminary analysis. In W. Olesky (Ed.), Contrastive pragmatics. (pp 38-51). Amsterdam: John Benjamins.

De Cock, S., Granger, S., Leech, G., & McEnery, T. (1998). An automated approach to the phrasicon of EFL learners. In S. Granger (Ed.), Learner English on computer

395

(pp. 67-79). London: Longman.

Elbers, L. (1995). Production as a source of input for analysis: evidence from the developmental course of a word-blend. Journal of Child Language, 22, 1, 47-71.

Ellis, N. C. (1996). Sequencing in SLA: Phonological memory, chunking and points of order. Studies in Second Language Acquisition, 18, 91–426.

Ellis, N. C. (2001). Memory for language. In P. Robinson (Ed.), Cognition and second language instruction (pp. 33–68). Cambridge: Cambridge University Press.

Ellis, N. C. (2003). Constructions, chunking, and connectionism: The emergence of second language structure. In C. Doughty & M. H. Long (Eds.), Handbook of second language acquisition. Oxford: Blackwell, 33-68.

Ellis, R. (1984). Formulaic speech in early classroom second language development. In J Handscombe, R. Owen, and B. Taylor (Eds.), 1984: On TESOL ’83: The question of control. Washington DC: Teachers of English to speakers of other languages.

Fayez-Hussein, R. (1990). Collocations: The missing link in vocabulary acquisition. In J. Fisiak (Ed.) Papers and studies in contrastive project, 26 (pp.123-136). Poznan: Adam Micklewiz University.

Fillmore, C. (1979). On fluency. In C. Fillmore, D. Kempler, and W. S-Y. Wang (Eds.), Individual differences in language ability and language behavior, New York, Academic Press, 85-101.

Fillmore, L. W. (1976). The second time around: Cognitive and social strategies in second language acquisition, unpublished PhD dissertation, Stanford University.

Ginzburg, R. S., Khidekel, S. S., Knyazava, G. Y., & Sankin, A. A. (1979). A course in modern English lexicology. 2nd edn., Moscow: Vusshaya Shkola.

Gitsaki, C. (1996). English collocations in second language acquisition. Unpublished Ph.D dissertation, University of Queensland, Australia.

Gläser, R. (1986). Phraseologie der Englischen Sprache. 2nd edn., 1990, Leipzig: VEB Verlag Enzyklopädie.

Granger, S. (1998). Prefabricated patterns in advanced EFL writing: collocations and formulae. In A. P. Cowie (Ed.), Phraseology: theory, analysis and application (pp.145-160). London: Oxford University Press.

Hakuta, K. (1974). Prefabricated patterns and the emergence of structure in second language acquisition. Language Learning 24, 287-298.

Hatch, E. (1974). Second language learning-universals? Working Papers in Bilingualism, 3, 1-18

396

Hatch, E., Peck, S. & Wagner-Gough, J. (1979). A look at process in child second language acquisition. In Ochs, E & Schieffelin, B (Eds.) Developmental pragmatics. Academic Press, New York.

House, J. (1996). Developing pragmatic fluency in English as a foreign language: Routines and Metapragmatic awareness. Studies in Second Language Acquisition, 18, 225-252.

Howarth, P. (1994). A computer-assisted study of collocations in academic prose, with special reference to grammatical structure and stylistic value. Unpublished Ph.D. dissertation, University of Lees, England.

Howarth, P. (1996). Phraseology in English academic writing: Some implications for language learning and dictionary making. Lexicographica Series Maior, 75. Tübingen, Max Niemeyer.

Howarth, P. (1998). Phraseology and second language proficiency. Applied Linguistics, 19 (1): 24-44.

Hussein, R. (1991). Collocations: The missing link in vocabulary acquisition amongst EFL learners. Papers and Studies in Contrastive Linguistics, Vol. 26, (pp. 123-136). Adam Mickiewicz University: Poznan.

Irujo, S. (1986). A piece of cake: learning and teaching idioms. ELT Journal 40 (3): 236- 242.

Karniol, R. (1990). Second language acquisition via immersion in daycare. Journal of Child Language, 17, 147-170.

Kellerman E. (1977). Towards a characterization of the strategy of transfer in second language learning. Interlanguage Studies Bulletin 2(1), 58-145.

Kellerman E. (1978). Giving learners a break: native language intuitions as a source of predictions about transferability. Working Papers on Bilingualism, 15, 59-92.

Kellerman E. (1979). Transfer and non-transfer: where are we now? Studies in Second Language Acquisition, 2, 37-57.

Kellerman E. (2000). Lo que la fruta puede decirnos acerca de la transferencia léxico- semántica : una dimensión no estructural de las percepciones que tiene el aprendiz sobre las relaciones lingüísticas. In Muñoz C. (Ed.) Segundas lenguas: adquisición en el aula (pp.21-37). Barcelona: Ariel.

Kellerman, E., & Jordens, P. (1977). Towards a characterization of the strategy of transfer in second- language learning. Interlanguage studies bulletin, Vol. 2, No. 1, (pp. 58-145).

Kizuka, H. (1995). A compact dictionary of verb-noun collocations. Tokyo: Japan Times.

Kizuka, H. (2002). Nihonjin ga machigaeyasui eigo eigoyourei jiten. Tokyo: Japan Times.

397

Kjellemer, G. (1991). A mint of phrases. In K. Aijimer &B. Altenberg (Eds.), English corpus linguistics, Studies in honour of Jan Svartvik (pp. 111-127). London: Longman.

Kjellmer, G. (1994). A Dictionary of English Collocations: Based on the Brown Corpus, 3 volume set. Oxford University Press.

Kjellmer, G. (1995). A dictionary of English collocations. Oxford, Oxford University Press.

Konishi, Y (1989) Kidoairaku eigo hyogen jiten. Tokyo: Hokuseido shoten.

Koya, T. (2002). Nihonjin eigogakushushano eigo korokeishon nouryokuni ataeru nihongo eikyo [first language influence on the acquisition of English collocation by Japanese learners]. The Journal of Japan Association o for Practical English, 10, 63-77.

Koya, T. (2005). The acquisition of basic collocations by Japanese learners of English. A doctorate dissertation, Waseda University, Tokyo. http://dspace.wul.waseda.ac.jp/dspace/handle/2065/5285

Krashen, S., & Scarcella, R. (1978). On routines and patterns in language acquisition and performance. Language Learning, 28(2), 283-300.

Lewis, M. (1993). The lexical approach. Hove: Teacher Training Publications.

Lewis, M. (2000). Teaching collocation: Further developments in the lexical approach. London, Language Teaching Publications.

Lieven, E. V. M., Pine, J. M. & Baldwin, G.. (1997). Lexically-based learning and early grammatical development. Journal of Child Language 24: 187-219.

Lipka, L.(1991). An Outline of English Lexicology. Forschung & Studium Anglistik 3, 2nd edn., 1992, Tübingen: Max Niemeyer.

Locke, J. K. (1997). A theory of neurolinguistic development. Brain and language, 58, 265-326.

Long, M. (1985). A role for instruction in second language acquisition: task-based language teaching. In K. Hyltenstam and M. Pienemann (Eds.), Modelling and assessing second language acquisition. Clevedon, Avon: Multilingual Matters.

Long, M. and Crookes, G. (1992). Three approaches to task-based syllabus design. TESOL Quarterly 26, 27-56.

Marton, W. (1977). Foreign vocabulary learning as problem no.1 of language teaching at the advanced level. Interlanguage studies bulletin 2, 33-57.

Marton, W. (1978). Dydaktyka jezyka obcego w szkole sredniej: Podejscie kognitywne. Warsaw: PWN.

398

McCarthy, M. (1990). Vocabulary. Oxford, Oxford University Press.

Mitchell, R. and Myles, F. (2004). Second language learning theories. London: Edward Arnold.

Miyakoshi, T. (2004). Collocations and second language acquisition: The acquisition of English adjectival constructions. University of Hawaii, Working papers in linguistics, 35(1), 1-24.

Mohmoud, A. (2005). Collocation errors made by Arab learners of English. Asian EFL journal, volume teachers’ articles, August. http://www.asian-efl- journal.com/pta_August_05_ma.php

Moon, R. E. (1998). Fixed Expressions and Idioms in English: A Corpus-Based Approach. Oxford Studies in Lexicology and Lexicography. Oxford: Clarendon Press.

Murano, R. (2004). L1 influence on learners’ use of high-frequency verb + noun collocations. ARELE: Annual Review of English Language Education in Japan, 14, 1-10.

Myles, F., Mitchell, R. & Hooper, J. (1999). Interrogative chunks in French L2. Studies in Second Language Acquisition, 21, 49-80.

Nattinger, J., and DeCarrico, J. (1992). Lexical phrases and language teaching. Oxford, Oxford University Press.

Nesselhauf, N. (2003). Transfer at the locutional level: An investigation of German- speaking and French-speaking learners of English. In Tschichold, C. (Ed.), English core linguistics. Essays in honour of D. J. Allerton (pp.269-286). Bern: Peter Lang.

Okitsu, F. (2000). Machigaiyasui eigohyogen. Tokyo: Sanshusha.

Oxford. (2002). Oxford collocations: Dictionary for students of English. Oxford, Oxford University Press.

Paul, K., and Nishimura, T. (1994). Nihonjin eigono misu 200. Tokyo: Kenkyusha.

Paul, K., and Nishimura, T. (1996). Zoku nihonjin eigono misu 200. Tokyo: Kenkyusha.

Pawley, A., & Syder, F. H. (1983). Two puzzles for linguistic theory: nativelike selection and nativelike fluency, in J. C. Richards and R. W. Schmidt (Eds.), Language and Communication (pp. 191-225). London: Longman.

Peters, A. M. (1983). The units of language acquisition. Cambridge: Cambridge University Press.

Peters, A. M. (1985). Language segmentation: Operating principles for the analysis and perception of language, in D.I. Slobin (Ed.), The Crosslinguistic Study of Language

399

Acquisition, vol.2. (pp.1029-1067), Hillsdale, NJ: Lawrence Erlbaum Inc.

Plunkett, K. (1993). Lexical segmentation and vocabulary growth in early language acquisition. Journal of Child Language, 20, 43-60.

Prabhu, N. S. (1987). Second language pedagogy. Oxford: Oxford University Press.

Raupach, M. (1984). Formulae in second language speech production. In H. Dechert, D. Mohle, and M. Raupach, (Eds.), Second language productions (pp.114-137). Tubingen, Gunter Narr.

Read, J. (2000). Assessing vocabulary. Cambridge: Cambridge University Press.

Riopel, W. (1984). The effect of two methods of ESL instruction on the learning of verb plus particle collocations. Unpublished master’s thesis, University of Hawaii.

Shapira, R. G. (1978). The non-learning of English: Case study of an adult. In E M. Hatch (Ed.), Second language acquisition: A book of readings (pp. 246–255). Rowley, MA: Newbury House.

Schmidt, R. W. (1983). Interaction, acculturation, and the acquisition of communicative competence: A case study of an adult. In N. Wolfson & E. Judd (Eds.), Sociolinguistics and language acquisition (pp.137-174). Rowley, MA: Newbury House.

Schmidt, R., and Frota, S. N. (1986). Developing basic conversational ability in a second language: A case study of an adult learner of Portuguese. In R. Day, (Ed.), Talking to learn (pp.237-336). Rowley, MA, Newburry House.

Schmitt, N., and Dunham, B. (1999). Exploring native and non-native intuitions of word frequency. Second Language Research, 15 (4), 389-411.

Schumann, J. H. (1978). Second language acquisition: The pidginization hypothesis. In E M. Hatch (Ed.), Second language acquisition: A book of readings (pp. 256–271). Rowley, MA: Newbury House.

Seki, I. (2001). Nihonjinga yokumachagaeru eigo. Tokyo: Senseido.

Sinclair, J. M. (1987). Collins COBUILD English Language Dictionary. London: Collins.

Snow, C. E. (1981). Social interaction and language acquisition. In P. Dale & D. Ingram (Eds.), Child Language: An international perspective (pp.213-228). Baltimore: University Park Press

Stubbs, M. (2001). Words and phrases: Corpus studies of lexical semantics. Oxford, Blackwell Publishers.

Swain, M. (1983). Communicative competence: some roles of comprehensible input and comprehensive output in its development. A paper presented at the 10th University

400

of Michigan conference on applied linguistics, Ann Arbor.

Tannen, D. and Öztek, P. (1979). Health to our mouths: Formulaic expressions in Turkish and Greek. In Coulmas, F, (Ed.), Conversational routine: Explorations in standardized communication situations and prepatterned speech (pp.37-54). The Hague, Morton.

Terasawa, Y, Debuchi, H. and Weekly, E. (1993). Kotobano romansu – eigono gogen. Tokyo: Iwanami bunko.

Theakston, A., Lieven, E. V. M., Pine, J., and Rowland, C. (2001). The role of performance limitations in the acquisition of verb-argument structure: an alternative account. Journal of Child Language, 28, 127-152.

Tomasello, M. (1992). First verbs: a case study of early grammatical development. Cambridge: Cambridge University Press.

Tomasello, M. (2003). Constructing a language. Cambridge, MA, Harvard University Press.

Vihman, M. (1982). The acquisition of morphology of a bilingual child. Journal of Child Language, 12, 297-324.

Weinert, R. (1995). The role of formulaic language in second language acquisition: a review. Applied Linguistics 16 (2): 180-205.

Wood, M. (1981). A definition of idiom. Manchester, England: Centre for Computational Linguistics, University of Manchester.

Wray, A. (2000). Formulaic sequences in second language teaching: Principle and practice. Applied linguistics, 21, 4, 463-489.

Wray, A. (2002). Formulaic Language and the Lexicon. Cambridge: Cambridge University Press.

Zhang, X. (1993). Teachers’ perceptions of and strategies with lexical anomalies. Paper presented at TESOL conference, April 1993, Atlanta, Georgia.

401