以語料庫為本之台灣英語學習者錯誤學術動名詞搭配詞分析

全文

(1)國立臺灣師範大學英語學系碩士. 論文. Master’s Thesis Department of English National Taiwan Normal University. 以語料庫為本之台灣英語學習者錯誤學術動名詞搭配詞分析. Verb-Noun Miscollocation in Academic Writing: A Corpus-based Study of EFL Advanced Learners’ English. 指導教授：陳浩然博士 Advisor: Dr. Hao-Jan Chen 研究生：鄭妤真 Student: Yu-Chen Cheng. 中華民國一零五年七月 July, 2016.

(2) 中文摘要搭配詞在過往的研究已被認為是成功的學術英文寫作的要點之一，這個論點對學習者特別重要的原因有一: 如學習者將搭配詞正確地使用於學術英文寫作時，將意味著學習者對學術英文有深刻了解以及有所認識。研究指出，學習者使用搭配詞的方式與母語者有所不同，但對於研究學習者在學術英文上之搭配詞使用錯誤卻尚無較多文獻。有鑑於此，本文旨在分析台灣碩士生在學術英文中之錯誤動名詞搭配詞。因搭配詞在不同學科下會有所不同，所以本文特別挑選兩個學科來討論學習者錯誤動名詞搭配詞，此二學科為:應用英語以及英語教學。本文以三個研究問題為論文主軸。第一個問題在於辨認出在應用英語以及英語教學碩士論文裡，常見的錯誤動名詞搭配詞。第二問題在於發掘台灣學習者的錯誤動名詞搭配詞之錯誤類別。第三問題而是探討學習者之錯誤動名詞搭配詞背後造成的可能原因。以便於回答上述研究問題，本文使用兩個大型語料庫，該語料庫為學習者碩士論文語料庫，以及參照語料庫。學習者碩士論文語料庫是由 Yang 在 2015 年建構而成。該語料庫包含 494 篇碩士論文，分別來自 10 間不同的應用英語以及英語教學研究所。另一方面，參照語料庫則是包含兩個來自 COCA 與 BNC 之學術子語料庫。學習者碩士論文語料庫以及參照語料庫的大小分別為一千一百萬字與十二億字。至於研究工具，則是使用線上查詢系統 (Sketch Engine)來搜尋錯誤動名詞搭配詞。本文經比對學習者碩士論文語料庫以及參照語料庫後，發現總共有 171 項、 1171 筆的錯誤動名詞搭配詞。將所有的錯誤動名詞搭配詞依照錯誤類型分類後，可以大約概括分成四種錯誤類型，此四種錯誤類型分別為: 動詞錯誤、名詞錯誤、. I.

(3) 介係詞錯誤或遺失，以及搭配詞不存在。142 項(946 筆)錯誤動名詞被歸類於錯誤動詞；16 項 (124 筆)為錯誤名詞；6 項(36 筆)為介係詞錯誤或遺失；7 項 (75 筆) 為搭配詞不存在。除此四大錯誤分類外，因錯誤動詞數眾多，三種類型之錯誤動詞也可在本研究中發現:同義動詞錯誤(62 項、443 筆)、短語動詞錯誤 (2 項、8 筆)、其他動詞錯誤 (78 項、495 筆)。於錯誤成因之分析下，七種錯誤成因也概括解釋所有 171 項錯誤動名詞搭配詞。第一項錯誤成因是母語影響-直接翻譯，共有 95 項(608 筆)錯誤動名詞搭配詞直接由母語翻譯為英語。第二錯誤成因母語影響-分類別錯誤 (split category)包含 10 項 (88 筆) 錯誤動名詞搭配詞。第三錯誤成因為同義詞混淆，有 39 項 (261 筆) 錯誤搭配詞為同義詞選錯誤。第四錯誤成因是錯誤類比，有 9 項 (38 筆) 錯誤搭配詞被歸類於此成因。第五錯誤成因是語意過度延伸，包含 7 項 (40 筆) 錯誤搭配詞。第六錯誤成因是拼字類似，有 9 項 (52 筆) 錯誤搭配詞因拼字類似而造成學習者混淆。最後，有 1 項 (39 筆)為動名詞皆錯誤。本文研究結果顯示，台灣學習者對在學術英文搭配詞中，錯誤動詞的頻率最高，表示台灣學習者在搭配詞中，動詞選擇最為困難，尤其是針對同義詞的選擇，對學生來說更加容易造成混淆。另外，根據錯誤成因研究結果顯示，母語負面影響為造成錯誤搭配詞的主要原因，需要學習者多加注意。研究者最後依據實驗結果，建議學術英文教學者將常見錯誤動名詞納入教學，並提醒學生母語與英語間的差異。. 關鍵字: 錯誤搭配詞、動名詞搭配詞、語料庫分析、學術寫作. II.

(4) Abstract Collocation has been recognized as the key to successful academic writing. This fact is especially true for learners because it signifies that learners are prepared and well-acquainted with the culture of academic writing. Several studies have reported the differences of collocational use by advanced learners and native speakers, however, the literature for learner miscollocations in academic writing is few. As a result, this study aims to investigate verb-noun miscollocation in master theses by Taiwanese EFL learners. Since collocations are highly discipline-specific, two disciplines were chosen in this study, namely, Applied Linguistics and English Teaching. There were three research questions governing this present study. The first question is to identify the common miscollocations made by Taiwanese master students in Applied Linguistics and English Teaching. The second question is to uncover the type of errors made by learners most. The last question is to investigate the possible causes of the miscollocations found in the academic prose. To answer the aforementioned research questions, two corpora, a learner corpus and a reference corpus, were utilized in this study. The learner corpus, which was composed of 494 master theses from 10 different Applied Linguistics and English Teaching programs in Taiwan, were compiled by Yang (2015). On the other hand, the reference corpus contained two academic subcorpora from BNC and COCA. The sizes of the learner and reference corpus were 11 million and 1.2 billion words respectively. As for instruments, an online query system, Sketch Engine, was adopted to identify target nouns and miscollocations. Altogether, 171 miscollocations with a total frequency of 1171 tokens were discovered in the research. All miscollocations were categorized into four main error types: misused verb, misused noun, misused or missing preposition, and combination III.

(5) non-existent. 142 types (946 tokens) of miscollocations were categorized as misused verb; 16 types (124 token) were misused noun; 6 (36 tokens) types were misused or missing preposition and 7 (75 tokens) types of collocations were non-existent. For the first type, misused verb, three patterns of misuse were also detected, which were, synonymous verb, phrasal verb and other verb. The number of miscollocations of these three sub-types were 62 types (443 tokens), 2 types (8 tokens), and 78 types (495 tokens). The analysis of causes yield seven possible explanations for miscollocations, L1 transfer-direct translation, L1 transfer-split category, misused synonym, false analogy, overgeneralization, approximation and others. 95 types (608 tokens) of miscollocations were influenced by a direct translation from Chinese. 10 types (88 tokens) of miscollocations were labeled as L1 transfer-split category. 39 types (261 tokens) of miscollocations were caused by selecting the wrong synonyms. 9 types (83 tokens) of miscollocations were false analogy. 7 types (40 tokens) of miscollocations contained verbs/nouns that were used over-extensively. 9 types (52 tokens) of miscollocations were categorized in approximation, and 1 type (39 tokens) of miscollocation had both erroneous nouns and verbs. Several interesting findings were observed. First of all, according to the result of error types, misused verbs was the most common error type in this study, indicating that proper verb usage poses the most difficulty for learners. Most of the misused verbs were also caused by confusion of synonymous verbs. Secondly, echoing with previous studies, there were several verbs that often baffled learners, such as gain, get, obtain, attain and reach. Lastly, based on the analysis of possible causes, the high frequency of miscollocation under L1 influence suggested that first language played a crucial role in learner collocations. Several pedagogical implications and future research directions IV.

(6) were also raised at the end of this thesis.. Keywords: miscollocation, Verb-Noun collocation, corpus analysis, academic writing. V.

(7) Acknowledgement This thesis would not have become a reality without the support of many individuals. First of all, I would like to express my sincere gratitude towards my advisor, Dr. Hao-Jan Chen. His expert advice and valuable comments guided me throughout the whole graduate study. His infectious passion for research and kind encouragements had been the main driving force throughout the hard times of thesis writing. He patiently answered every question and provided me with insightful suggestions that assisted me not only in my thesis but in my life as a whole. I will always cherish every suggestion and the knowledge he shared with all my life. I would also like to take this opportunity to thank Dr. Yi-Chien Lee and Dr. ZhaoMing Gao, for being my committee members despite their busy schedules. Dr. Lee and Dr. Gao offered tremendous help by reading the drafts and providing insightful advice to improve the quality of my thesis and presentation. My special thanks goes to Dave, my soul mate and best friend for life. He spent ample time helping me proofread my thesis and participated in the research. His warm support and encouragement was the shelter for me whenever I was wiped out by the storm. His patience and wisdom lead me through every difficult step I took. He believed in me when I doubted myself the most. Without him, my graduate study would not have been possible. I would also like to thank Mr. and Mrs. Allwein for their constructive feedback on the research. Their kind help made this thesis more accurate and interesting. Richard also deserves some recognition. He was there for me many a day, patiently sitting and watching me while I was hard at work. Thanks also go out to my friends, Jason, Lisa, Ignace, and Christine. Thank you for offering suggestions when I needed help for my thesis and presentation. I would also like to express my appreciation to Puma and Jean VI.

(8) for offering a shoulder to cry on and a laughter to share when I was down and low. Last, but not least, I would like to thank my family: my father, my mother, my sister and my grandfather, for all the support throughout the process of my graduate study. Their heart-warming encouragements always kept me going when I felt like quitting. Thank you for taking care of me and for providing me so that I could concentrate on my studies. Words cannot describe my sincere gratitude. Finally, thank you, grandmother. You will be forever missed and never forgotten.. VII.

(9) TABLE OF CONTENTS. 中文摘要……………………………………………………….. I ABSTRACT…………………………………………………… III ACKNOWLEDGEMENT……………………………………. VI LIST OF TABLE……………………………………………... XI LIST OF FIGURE……………………………………………. XIII CHAPTER I INTRODUCTION……………………………... 1 1.1 Background of the Study………………………………………………. 1 1.2 Motivation of the Study……………………………………………….. 3 1.3 Purpose of the Study…………………………………………………... 6 1.4 Definition of Key Terms………………………………………………. 7. CHAPTER II LITERATURE REVIEW……………………. 9 2.1 Notion of Collocation………………………………………………….. 9 2.1.1 Phraseological View of Collocation……………………………. 9 2.1.2 Frequency-based View of Collocation…………………………. 12 2.1.3 Grammatical Categorization of Collocation…………………… 14 2.2 Advanced ESL/EFL learners’ Difficulties of Collocation……………... 15 2.2.1 Advanced EFL learners’ Difficulties of Collocation in Academic writing……………………………..……………………... 18 2.3 Collocation Error by Advanced EFL learners…………………………. 21 2.4. The Present Study…………………………………………………….. 27. CHAPTER III METHODOLOGY………………….............. 29. VIII.

(10) 3.1 Corpora used in this study……………………………………………... 29 3.1.1 Master Theses corpus…………………………………………... 29 3.1.2 Reference Corpus………………………………………………. 30 3.2 Instrument……………………………………………………………... 31 3.3 Data Extraction and Analysis………………………………………….. 36 3.3.1 Noun Selection…………………………………………………. 37 3.3.2 Miscollocation Extraction……………………………………… 38 3.3.3 Types and Causes of Miscollocations………………………….. 42. CHAPTER IV RESULT………….…………………............... 45 4.1 Overall Result and Error Types………………………………………... 45 4.2 Analysis of Verb-Noun Miscollocations of Types of Errors…………... 47 4.2.1 Misused verb…………………………………………………… 48 4.2.1.1 Synonymous verbs……………………………………… 49 4.2.1.2 Phrasal Verb…………………………………………….. 53 4.2.1.3 Misused or Missing Prepositions……………………….. 53 4.2.1.4 Other verb………………………………………………. 55 4.2.2 Misused Noun………………………………………………….. 59 4.2.3 Combination not existed……………………………………….. 62 4.2.4 Most misused Verbal collocates in Verb-Noun Miscollocation... 63 4.3 Possible Causes of Verb-Noun Miscollocations………………………. 65 4.3.1 L1 Transfer: Direct Translation………………………………… 67 4.3.2 L1 Transfer: Split Category…………………………………….. 71 4.3.3 Misuse Synonym……………………………………………….. 72 4.3.4 False Analogy…………………………………………………... 74 4.3.5 Overgeneralization……………………………………………... 75 IX.

(11) 4.3.6 Others…………………………………………………………... 76 4.4 Discussion on Verb-Noun Miscollocations in Academic Writing……... 77. CHAPTER V CONCLUSION…..…………………................ 80 5.1 Summary of the Study…………………………………………………. 80 5.2 Pedagogical Implications……………………………………………… 81 5.3 Limitation of This Study………………………………………………. 83 5.4 Suggestions for Future Research………………………………………. 83. REFERENCES………………………………………………... 85 APPENDICES………………………………………………… 91. X.

(12) LIST OF TABLE. Table 2.1. Definition of collocation types by Jones and Sinclair (1974)…. 13. Table 2.2. Definition of upward/downward collocation by Sinclair (1991). 14 Table 2.3. Sub-categorizations of lexical collocations (Benson et al., 1987)………………………………………………………………………. 15 Table 2.4. Types of mistakes in collocation (Nesselhauf, 2003)…………... 23 Table 2.5. Types of collocation error (Wang & Shaw, 2008)……………… 24 Table 2.6. Causes of miscollocation (Liu, 1999)…………………………... 25 Table 3.1. Details of Master Thesis Corpus……………………………….. 30 Table 3.2. Comparison of Master Thesis Corpus (MTC) and COCABNCacademic corpus………………………………………………. 31 Table 3.3. Types of miscollocation (Nesselhauf, 2003; Wang & Shaw, 2008)……………………………………………………………………….. 43 Table 3.4. An adopted version of Liu’s (1999) list of miscollocation causes………………………………………………………………………. 43 Table 4.1. Overall statistical data of miscollocation………………………. 45 Table 4.2. Types and tokens of miscollocation of each error type………… 47 Table 4.3. Overall statistical data of miscollocation………………………. 49 Table 4.4. Misused verb error type: synonym verbs……………………..... 49 Table 4.5. Misused verb error type: phrasal……………………………….. 53 Table 4.6. Misused or missing preposition error type……………………... 54 Table 4.7. Misused verb error type: other verbs…………………………… 56 Table 4.8. Misused noun error type………………………………………... 60 Table 4.9. Combination not existed……………………………………….. 62 XI.

(13) Table 4.10. Most misused verbs in miscollocations………………………. 64 Table 4.11. Possible causes of miscollocation……………………………. 66 Table 4.12. Possible cause of miscollocation: L1 transfer-direct translation…………………………………………………………………. 68 Table 4.13. Possible cause of miscollocation: L1 transfer-split category...... 72 Table 4.14. Possible cause of miscollocation: misuse Synonym………….. 73 Table 4.15. Possible cause of miscollocation: false analogy………………. 75 Table 4.16. Possible cause of miscollocation: overgeneralization………… 76 Table 4.17. Possible cause of miscollocation: others……………………… 77. XII.

(14) LIST OF FIGURE. Figure 3.1. Word list function……………………………………………... 32 Figure 3.2. Word list of Master Thesis Corpus…………………………… 33 Figure 3.3. Interface of Sketch Diff function……………………………… 34 Figure 3.4. Sketch Diff result of “feedback” in two corpora……………... 36 Figure 3.5. Sketch Diff result of target noun “progress” in two corpora…………………………………………………………………....... 39 Figure 3.6. Concordance lines of miscollocation “reach progress”……….. 39 Figure 3.7. Suspicious verbal collocates of the targeted noun “word” in MTC corpus………………………………………………………………... 40. Figure 3.8. Concordance lines and frequency counts of “listening comprehension”……………………………………………………………. 41 Figure 3.9. Flow chart of data extraction…………………………………... XIII. 44.

(15) Chapter One: Introduction. 1.1 Background of the Study Collocation, prefabricated lexical patterns of a reoccurring nature, has been recognized as one of the fundamental components of the English language. Previous research found out a large portion of English speakers’ language, both written and spoken, is made up of recurring lexical patterns (Biber & Conrad, 2005). Therefore, ways to mastering reoccurring, prefabricated clusters have attracted the interest of scholars in language learning research (Arnaud & Savignon, 1997; Bahns & Eldaw, 1993; Biskup, 1992; Durrant & Schmitt, 2009; Fan, 2009; Farghal & Obiedat, 1995; Granger, 1998; Gyllstad, 2005; Howarth, 1998; Laufer & Waldman, 2011; Nesselhauf, 2003, 2005; Panahifar, 2013; Peacock, 2012; Wang & Shaw, 2008). For foreign language learners, mastering the use of collocation is especially crucial since it might positively influence the process and outcome of language learning. Nesselhaul (2005) stated several advantages of collocation for language learners, namely reducing mental effort and reaching native-like fluency. First of all, utilizing collocation reduces mental effort when producing language (Nesselhaul, 2005:2). According to Schmitt (2000), the human brain can only store a limited amount of information in real time due to the small capacity of short-term memory. In order to accelerate processing information, some “frequently-used lexical chunks” would be stored in long-term memory and be retrieved when needed. Retrieving lexical chunks from long-term memory does not require on-line cognitive effort to decode individual word meaning, indicating that more cognitive effort can be used to make sense of new information. Such psycholinguistic phenomenon is beneficial for non-native English 1.

(16) speakers because they need extra mental effort to process English compared to native speakers (Nesselhaul, 2005). The other major advantage of collocation is that it aids learners in speaking and writing more native-like (Fan, 2009; Howarth, 1998; Laufer & Waldman, 2011; Nesselhauf, 2003, 2005). In a set of naturally produced utterances by native speakers, there are some words that tend to appear together in a phrase, while others, which are equally acceptable grammatically, sound odd (Wray, 2002). In a study done by Biber, Conrad and Reppen (1998), for example, the phrase “large number” was found to be more commonly used than “great number”. If learners are able to produce such prefabricated and idiomatic phrases adequately, the language they produce will be less awkward and more native-like. Based on the advantage and features of collocation above, it has been acknowledged that collocations are crucial linguistic patterns for foreign learners to acquire. In the setting of English for Academic Purpose, proper usage of collocation might determine whether the piece of work is successful or not (Wu, Chang, Mitamura, & Chang, 2010). One prominent characteristic of academic writing is that precise word choice and clear statements should be emphasized throughout the entire work. Unlike other registers such as fiction, news or political speeches, where ambiguity is allowed, any sign of obscurity might come across as immature writing (Howarth, 1998). Hence, collocation usage in academic prose should be as precise as possible to be able to convey the author’s message without confounding the readers. Howarth (1998) also advocated the indispensability of collocation in academic writing, stating “conventional collocations are essential for effective communication, and their use by non-native writer is a clear sign that these learners have made an essential adjustment to the 2.

(17) academic culture they are entering in” (p. 186). In other words, it is mandatory to gain a clear understanding of collocation to be able to compose a successful academic prose. Even though the importance of collocation has been long established, foreign language learners still have difficulty with producing conventionally acceptable collocations (Ferraro, Nazar, Ramos, & Wanner, 2014; Granger, 1998; Howarth, 1998; Nesselhauf, 2003). Since collocations do not entirely follow syntactic and semantic rules, and word substitution is allowed, learners were often found to produce collocational mistakes. (Fan, 2009; KUO, 2009; Laufer & Waldman, 2011; Nesselhauf, 2003, 2005). For instance, Nesselhaulf (2003) reported that V-N collocations with less restrictions often troubled learners more than ones that did not allow substituting elements in the collocation. Furthermore, the phenomenon of making collocational mistakes seems to be prevalent among learners of all English proficiency levels. Laufer and Waldman (2011) pointed out that compared to intermediate and basic learners, advanced learners produced more varieties of collocations, but made more mistakes as well. Given the previous findings, learners might have a deficiency with producing collocations and the causes might be far more complicated than general vocabulary problems.. 1.2 Motivation of the Study Several studies were dedicated to learner difficulty of collocation in academic settings (Chang, Chang, Chen, & Liou, 2008; Durrant, 2009; Howarth, 1998; Wu et al., 2010). The results all suggested that learners possess limited knowledge of native-like collocation, and they tend to overuse some high-frequency collocations. These studies made insightful discoveries in learner collocation, however, the conclusions mostly 3.

(18) focused on the underuse/overuse behavior of learner collocation. There is an urgent need to look into learner miscollcoation in academic settings as well. The work of Woolard (2000) revealed that one useful way to raise learner awareness of collocation is to direct his or her attention towards the collocation error he or she produced. Lewis (2000) also claimed that it is equally important for EFL learners to know what is wrong as to what is correct. Hence, there is a necessity to investigate miscollocation to help raise collocational awareness in learners, and as a result, improve the learner’s academic writings. Aside from the need to investigate miscollocation in the academic setting, another research niche that is worth delving into is the discipline-specific collocation error. As pointed out by Yang (2015), collocations in academic prose can be topic or genrespecific, and therefore studies should further identify collocations that appear in individual fields of study exclusively. Ward (2007) compared lexical characteristics of undergraduate textbooks in chemical, civil, electrical, industrial, and mechanical engineering, concluding that complex noun phrases found in the engineering text is highly discipline-specific. For instance, “gas +” collocation is a typical phrase in chemical engineering. Some examples were also provided by Hyland and Tse (2007), such as communicative competence in applied linguistics, situated learning in education, and social constructivism in social science, echoing the fact that collocations in academic prose are subject-specific (Hyland & Tse, 2007:247). For foreign language learners in applied linguistics, in order to enhance academic collocational knowledge, it is essential to understand and be capable of using discipline-specific collocations correctly. As Bhatia (2002) proposed, "Students interacting with different disciplines need to develop communication skills that may not be an extension of general literacy 4.

(19) to handle academic discourse, but a range of literacies to handle disciplinary variation in academic discourse" (p. 27). To sum things up, based on the observations above, investigation on miscollocation in applied linguistics academic writing is warranted. Although the above studies yield interesting results, cautious approaches might need to be taken with analyzing the data. One limitation of past miscollocation studies is that the amount of data and the size of the corpus used is relatively limited. The learner miscollocation study by Nessehaulf (2005) is one of the most comprehensive analysis to date. She investigated behaviors of learner collocations in the German component of ICLE with corpus extraction techniques. The corpus, comprising 154,191 words, were carefully tagged and modified to fit the purpose of the research. Howarth (1996), on the other hand, also manually examined verb-noun miscollocation by nonnative speakers with different first languages. The corpus utilized in his study is roughly about 22,000 words. Despite the thorough investigations of both studies setting the ground for all subsequent collocational research, the corpus size can still be expanded. The miscollocations extracted provided evidence to improve our understanding of learner colloaction error; however the collocation errors found were rather limited (Lin, 2010). As Liu (2013) stated, the larger the corpus is, the more significant the examples that could be discovered to increase the validity of the final results. To sum it up, three ways to improve learner collocational research in academic settings are investigating learner miscollocation, targeting the research in a specific discipline, and increasing the size of the corpus. In order to further understand the phenomenon and causes of EFL learners’ collocation use, there is a need for employing a corpus with a larger size to look into learner collocation errors in discipline-specific academic prose. 5.

(20) 1.3 Purpose of the Study This present study aims to investigate common miscollocation in academic prose by Taiwanese EFL learners. One specific academic field, applied linguistics, was chosen since this paper hopes to shed light on discipline-specific learner miscollocation. The reason for choosing academic prose in applied linguistics specifically was because according to Durrant (2009), the vocabulary needs of students from Arts and Humanities are characteristically different from students in other disciplines. Hence there is a special need to investigate academic texts in such disciplines alone. In terms of the type of collocation examined, verb-noun (V-N) collocation was the main focus of this study. Liu (1999) indicated that verb-noun (V-N) collocation is the most difficult type of collocation that troubled learners. Research was conducted on Taiwanese learner miscollocation using English Taiwan Learner Corpus (English TLC), 233 out of 265 lexical miscollocations were found out to be V-N collocation. Nessehaulf (2003) also pointed out that the wrong choice of verbs was the most common mistake made by learners. The subtle differences between semantically similar verbs usually caused confusion. Therefore, a thorough investigation on V-N collocation might facilitate learners with understanding their major issues with collocations in academic writing in order to produce a piece of academic writing of a higher quality. Another purpose of this study is to adopt an efficient tool to extract suspicious miscollocations automatically. Previous methods of extracting V-N miscollocations mostly consisted of manually extracting all V-N collocations, followed by verifying all V-N collocations with BNC and dictionaries (Howarth, 1998; Laufer & Waldman, 2011; Nesselhauf, 2003). This method ensured that every V-N collocation was examined and it also eliminated the chance of selecting non V-N collocation accidentally. However, 6.

(21) the process of extracting V-N miscollocations might be time-consuming, and the involvement of subjectivity in the process might make some of the decisions to be open to discussion. To overcome the problems mentioned above, this study utilized a more efficient method of miscollocation extraction by comparing a large non-native corpus directly to BNC and COCA academic components via Sketch Engine, an online corpus query engine. First adopted in collocation study by Liu (2013), Sketch Engine was originally designed to create a profile of grammatical relations a word is in (Kilgarriff, Rychly, Smrz, & Tugwell, 2004). However, through the function of Sketch Diff, it was possible to investigate how a word behaves in two different corpora. Aside from adopting a more efficient method of miscollocation extraction, this study also utilized one of the largest Chinese learner academic corpus to date, which was compiled by Yang (2015). With a large corpus and powerful corpus query engine, it was hoped to provide some interesting yet insightful conclusions for the research field of L2 learner collocation.. 1.4 Definition of Key Terms 1.4.1 Verb-Noun collocation In the lexical collocation definition developed by Benson, Benson and Ilson (1986), lexical collocation consist of nouns, adjectives, verbs, and adverbs. One type of lexical collocation is formed by a lexical verb and a noun as object, which was referred in this present paper as verb-noun collocation. 1.4.2 Miscollocation Based on previous research (Laufer & Waldman, 2011; Nesselhauf, 2003, 2005; Wang & Shaw, 2008), a collocation were judged as unacceptable if the collocation: 7.

(22) cannot be found in dictionaries (Oxford Advanced Learner’s Dictionary, the Collins COBUILD English Dictionary, The BBI Dictionary of English Word Combinations, and Oxford Dictionary of English Idioms), found less than 10 hits in BNC, not being accepted by native speakers. This present paper followed a similar line of judgement, and referring to a collocation as miscollocation if the collocation did not appear in BNC/COCA academic subcopora, not being found in BNC, and not being accepted by one native English speaker.. 8.

(23) Chapter Two: Literature Review This chapter reviewed previous studies on ESL/EFL learners’ collocations. The review started out discussing three different notions of collocations which were adopted in previous research. The second part of this chapter summarized some difficulties EFL learners might have when producing collocations, especially in academic writings. The third part emphasized on the types and causes of miscollocations found in learners’ writings. Lastly, this section ended with the aim and research questions answered in this present study.. 2.1. Notion of Collocation Collocations, also known as formulaic sequences, prefabs, or chunks, are one of the major lexical competences of second-language learners. There has not been a clearcut definition for collocation due to its complexity (Fan, 2009). Three broad definitions, namely phraseological tradition, frequency-based view and grammatical categorization, can be discovered in previous research (Barfield & Gyllstad, 2009). Three notions were discussed in details below.. 2.1.1 Phraseological View of Collocation The core of phraseological view is the paradigmatic and associative relationship between the lexical items in one collocation. Unlike frequency-based view, which emphasizes on frequency, phraseological view is more concerned with the fixedness and substitutability of the lexical items. Cowie (1981) first approached the phraseological notion of collocation and defined collocation “a composite unit which permits the substitutability of items for at least one of its constituent elements” (P. 224). He further raised two parallel 9.

(24) explanations that governed the notion of collocation. Firstly, in one collocation, one item was used in a figurative sense. For instance, in the word combination, explode a bomb was defined as free-combination since all of the lexical items were used in their literal meaning. However, explode a myth, was categorized as collocation because the word explode was used in the sense of “to prove something is wrong”, which is different from its original meaning, “to burst”. The figurative sense of explode restricted the object-nouns following behind it. Only a small handful of words could be attached behind it, such as theory, idea, notion (Cowie, 1981, P.226). Aside from determination of collocates by the figurative sense of one lexical item, the second explanation can be viewed from the other way around. The context, on the other hand, might trigger the figurative meaning of one element. Cowie claimed that, “a single item in a figurative sense is determined by a limited set of items used in a literal sense” (P. 226). Taking the word combination explode a myth as an example again, explode was only interpreted figuratively in the literal context of myth, notion, idea, theory. There might not be other context which triggered the figurative sense of explode. Howarth (1996) followed Cowie’s work and purposed that the meaning of collocation was not completely figurative as a whole. There was still evidence that each component contributed to the meaning of collocation individually. Only one component was in its “specialized” sense and thus made the word combination had figurative sense. For instance, foot the bill (collocation) and fill the bill (idiom), even though foot was not in its literal sense (as in the meaning of “pay”), the semantic meaning of foot can still be extracted from the collocation. However, the semantic sense of fill the bill was one inseparable unit. To make the definition of collocation more precise, there were three conditions in which the lexical item was considered specialized (Aisenstadt, 1979): 1. With a narrow, specific meaning. E.g., Shrug ones shoulder (Howarth, 1996, 10.

(25) P. 39) 2. Having a figurative, secondary meaning. E.g., File an application 3. Delexicalized meaning. E.g., Catch fire Imagining a scale of semantic transparency and specialized sense of lexical items in a one word combination, the more narrow, specific or delexicalized the sense of the word is, the more restricted such a combination might be. Nesselhaulf (2005) expanded Cowie’s viewpoints and defined collocation as “words that habitually occur together with restriction of substitutability and relative transparency of meaning” (Nesselhaulf, 2005, P. 2). She established a term named “restricted sense” to explain the semantic restriction in a verb-object-noun collocation. Based on the degree of restrictedness of lexical items in one collocation, a continuum of idiomaticity was presented free of restriction on one side (free combination) and fully restricted on the other (idiom). Two criteria were also developed to determine the degree of restrictedness: 1. The sense of verb is so specific that it only takes a small range of nouns following behind it. 2. All semantically and syntactically acceptable nouns cannot be used with a specific sense of verb. On the basis of restrictedness, three categories of word combinations were therefore further developed, free combination, collocation, and idiom. In free combination, all items were used in an unrestricted sense (e.g., want a car), whereas in collocation, the verbal item was restricted and only took a small set of nouns as collocates (e.g., dial a number). As for idioms, both verb and noun in the word combination were used restrictedly, not allowing any kind of substitution (e.g., sweeten the pill). In short, as long as either one of the criteria or both were observed in the verbal component, such a 11.

(26) combination was then classified as collocation. Similar to Cowie’s (1981) definition of collocation, Nesselhaulf (2005) also had an underlying concept that nouns behaved the same in both free combination and collocation, but verbs behaved differently. Nesselhaulf’s analysis of word combination clarified the hazy definition of collocation by using verb as the core of classification for collocation.. 2.1.2 Frequency-based View of Collocation A Frequency-based definition considers collocation as a set of words within a certain distance appearing frequently (Barfield & Gyllstad, 2009; Fan, 2009). Pioneering work under this tradition can be traced back to Firth’s study (1957). He proposed “collocability” within collocation, using the example “dark night” to demonstrate that the meaning of “night” is complete when combined with the word “dark”. In other words, the meaning of a word is more intact with the presence of other supplementary words which it has collocability with. Furthermore, the meaning of collocation cannot be simply predicted from each component in the collocation lexically. The meaning of collocation, as Firth argued, should be “an abstraction at the syntagmatic level” (P. 196). Halliday (1966) followed Firth’s concept of collocation, and expanded his concept by introducing the following terms: node, collocate and span of collocation. Node is the primary word in the collocation. Collocate is the frequently co-occurring word and span is the environment where collocate would appear. In Halliday’s view of collocation, predictions of lexical items can be made since some co-occurring items appear “greater than chance” (P.156). Discussing collocation from a statistical approach, Sinclair (1991) combined ideas from both Firth and Halliday, and defined collocation as “the occurrence of two or more words with a short space of each other in the text” (P.170). Two types of collocation 12.

(27) were distinguished by Sinclair based on the definition above, which were, casual collocation, and significant collocation. Significant collocations could be distinguished by the higher frequency of co-occurring word combinations when compared to the respective frequencies of each word in a given length of text (Jones & Sinclair, 1974). For instance, the word combination “the club” might be a casual collocation instead of a significant collocation because “the” occurs more often than the word combination the club. However, join club, on the other hand, could possibly be a significant collocation. In this present study, Sinclair’s definition was adopted and the frequency of word combinations was the key to decide whether the targeted collocation is acceptable or not.. Table 2.1. Definition of Collocation Types by Jones and Sinclair (1974) Collocation Type. Definition. Example. Casual. Frequency of word combination < Frequencies each word.. The club. Significant. Frequency of word combination > Frequencies each word.. Join club. Later in his study, Sinclair (1991) made an observation that a lexical item which appeared most frequently in general had a less clear and more opaque meaning. To put it in another sense, the more frequent lexical item was “delexicalized” and more difficult to analyze or explain. Based on this observation, Sinclair presented two types of collocation: upward collocation and downward collocation. Upward collocations included words that tend to collocate with lexical items of higher frequency than itself, whereas downward collocations consisted of words that collocated with words with lower frequency. To distinguish upward and downward collocate statistically, Sinclair proposed that the frequency of collocate should be plus or minus 15% of the frequency. 13.

(28) of the node1. Using the word back as an example, upward collocates were down, from, into, at, and downward collocates were arrive, bring, behind, again. The former type of collocations were often used as grammatical frames; on the other hand, the latter type of collocations denoted more semantic sense (Sinclair, 1991, P.116). Sinclair’s statistical definition of collocation provided insights of the “lexical realization of the situational context” (Moon, 1987, p22). Table 2.2. Definition of Upward/Downward Collocation by Sinclair (1991) Collocation Definition Upward Downward. words collocating with lexical items with higher frequency than itself words collocating with lexical items with lower frequency than itself. Example Back from Bring back. 2.1.3 Grammatical Categorization of Collocation Different from the frequency tradition and phraseological view, grammatical categorization of collocation did not tend to define collocation from a semantic aspect. The BBI dictionary of English Word Combinations (Benson, Benson, & Ilson, 1987) presented two categories for collocation, grammatical collocation and lexical collocation. Grammatical collocation was described as a word combination with a “dominant word” (usually contained the core meaning of collocation) which was followed by a preposition or a clause (Benson et al, 1987, P. XIX). A typical grammatical collocation might be decide on. Lexical collocation, however, only contained nouns, verbs, adjectives, and adverbs. In contrast to grammatical collocations, grammatical words could not be seen in lexical collocations. Benson et al. (1987) further divided lexical collocation into seven sub-categories (L1 to L7): verb +. For instance, if the frequency of node is 100 times in English, upward collocate should appear at least 115 times, and downward collocates should appear less than 85 times. 14 1.

(29) noun/pronoun (prepositional phrase), verb + noun, adjective + noun, noun + verb, noun1 + of + noun2, adverb + adjective, verb + adverb (Examples see Table 2.3) Table 2.3. Sub-categorizations of lexical collocations (Benson et al., 1987) Collocation Type. Grammatical Structure. Example. L1. verb + noun/pronoun (prepositional phrase). set a record. L2. verb + noun. break a code. L3. adjective + noun. strong tea. L4. noun + verb. blood circulates. L5. noun1 + of + noun2. a school of whales. L6. adverb + adjective. closely acquainted. L7. verb + adverb. apologize humbly. Some differences between L1 and L2 was that verbs in L1 denoted a sense of creation and/or activation, such as make an impression, whereas L2 collocation consisted of verbs which had a meaning of eradication and/or nullification, like demolish a house. Based on this classification of collocation, this present study intended to look into L1 and L2 types of collocation. Grammatical collocations were not included in the research. 2.2 Advanced ESL/EFL learners’ Difficulties of Collocation Due to the importance of collocation in language learning, a large amount of studies were carried out to investigate the features and patterns of learners’ collocation (Bahns & Eldaw, 1993; Barfield & Gyllstad, 2009; Biskup, 1992; Durrant, 2009; Farghal & Obiedat, 1995; Granger, 1998; Gyllstad, 2005; Laufer & Waldman, 2011; Nesselhauf, 2003, 2005; Panahifar, 2013; Peacock, 2012; Wang & Shaw, 2008). It was discovered that ESL/EFL learners did not possess enough collocational knowledge in general (Bahns & Eldaw, 1993; Biskup, 1992; Farghal & Obiedat, 1995; Granger, 1998; Gyllstad, 2005; Howarth, 1998; Nesselhauf, 2003). Bahns & Eldaw (1993) utilized translation and fill-in-blank methods to examine the relationship between general 15.

(30) knowledge of vocabulary and collocation use. Using noun as node and verb as collocate, Bahns & Eldaw selected 15 V-N collocations to investigate collocational knowledge of 58 German advanced learners. The result suggested that learners were more likely to make mistakes in collocation production than they are with general vocabulary production. Bahns & Eldaw later concluded that learners’ problems of collocation were far more complicated than simple general vocabulary issues. Possible causes could be attributed to not being aware of potential difficulty of collocation. By delving into productive knowledge of collocation, Farghal and Obiedat (1995) used fill-in-the-blank tests to investigate learners’ coping strategies when encountering unfamiliar collocations. Overall analysis showed that both junior and senior college students were deficient in collocation, therefore four strategies were used for lexical simplification, namely voidance, transfer, synonymy, and paraphrasing. Among the four skills, synonym was used most frequently, and thus Farghal and Obiedat (1995) later claimed that learners were unaware of “collocational restrictions of lexical items” (P. 321). Learners were also unable to acknowledge collocation as prefabricated patterns in which random substitution of words were not allowed. Discussing collocational development and difficulties at different proficiency levels, Gitsaki (1997) investigated how 275 Greek junior high school students made use of collocation in composition, blank-filling and translation tasks. Students were divided into three groups: post-beginning, intermediate, post-intermediate. Results suggested that the accuracy of both lexical and grammatical collocations were low with the beginning level learners. Gitsak (1997) explained that lexical collocations were relatively shorter than grammatical collocations and therefore beginning level students tend to memorize lexical collocation as unanalyzed chunks. Learners were also found to have more problems with grammatical collocations containing prepositions because 16.

(31) of L1 interference. Aside from studies using traditional elicitation techniques, some studies resort to corpus technology and analyzed learner collocation with large amount of texts. With help of computer software, comparison among native and non-native corpora provides more insights into collocation problems learners might encounter. In a corpus-based study done by Granger (1998), International Corpus of Learner English (ICLE) was utilized and investigated how -ly adverb + adjective (e.g., perfectly natural) collocation were used by French learners. ICLE is a computerized corpus containing EFL writings from all over the world. Comparing French components of ICLE (251,318 words) and the native corpus (234,514 words), Granger used text retrieval software to extract all words ended with –ly and manually deleted the words that did not serve a semantic purpose of intensifier. The result demonstrated that learners underused booster amplifiers (e.g., highly) compared to native English speakers, and furthermore, overused two particular amplifiers, completely, totally, which Granger referred as “safe bets”. With an underdeveloped sense of collocation restriction, learners tended to underuse native-like collocation and overuse collocations which were not typical in English. Laufer & Waldman (2011), on the other hand, discussed collocations used in 759 argumentative essays by basic, intermediate and advanced level of Israeli students. 220 nouns appearing at least 20 times were extracted from the native corpus. V-N collocations were then selected in both corpora, and checked with BNC, dictionaries, and natives for their validity. Laufer & Waldman found out that all learners produced significantly less collocation than native speakers regardless of proficiency level. In terms of collocation error, advanced learners made more mistakes than other levels of learners, even though advanced learners, at the same time, had the most amount of 17.

(32) collocation produced. They later claimed that the reason for such a high ratio of collocation error (one third of all collocation produced) was that learners processed individual words in the word combination, instead of prefabricated patterns.. 2.2.1 Advanced EFL learners’ Difficulties of Collocation in Academic setting Turning to learners’ collocational use in academic writings, several studies (Durrant & Schmitt, 2009; Howarth, 1998; Li & Schmitt, 2010; Yang, 2015) investigated the difference of collocation use between native and non-native speakers. Shared findings suggested that the use of collocations in academic setting deviated from native norms of collocational use. Howarth (1998) conducted research on V-N collocations of academic writings by non-native speakers. The native corpus contained texts extracted from LOB corpus and published texts in sociology, education and law (240,000 running words), whereas 10 sets of graduate school assignments from non-native speakers in applied linguistics were collected to compile a non-native corpus (25,000 running words). The total amount of 5300 V-N collocations from native corpus and 1200 collocations from nonnative corpus were extracted by using a list of high frequency verbs. All collocations were checked with BB1 Combinatory Dictionary of English and Oxford Dictionary of Current Idiomatic English and three levels of restrictedness of word combination (free combination, collocation, and idiom) were identified. Results found out that conventional collocations and idioms found in non-native corpus were 25 % of the entire corpus, relatively lower than the native norm (38%), indicating that learners had a lower level of knowledge of collocation. Other findings also showed that less restricted collocations were found in the non-native corpus, and more unacceptable collocation produced by non-native postgraduate learners. Howarth suggested that learners lacked of awareness of appropriately deploying lexical items in collocation. 18.

(33) Using methodologies which take advantage of frequency, Durrant and Schmitt (2009) intended to find out how non-native learners make use of word combinations in academic writings. They studied adjective-noun and noun-noun collocations by ESL/EFL learners of different L1 and collected two lengths of text (short, long) to form four corpora: NS long, NNS long, NS short, and NNS short. A native speaker corpora consisted of long texts from Prospect magazine and academic essays by students in Applied Linguistics and short texts extracted from LOCESS corpus. NNS corpora, on the other hand, contained long texts from British, Turkish academic essays, and short texts from Bulgarian subcorpus of ICLE. A total of 10,839 collocationss from the 96 texts and only 1,355 was produced by non-native speakers (12%). Two main findings were raised. First, non-native writers tend to use less low-frequency (found less in five times in BNC) collocations than native writers. The pattern was consistent in both short and long texts, however, the trend seemed to be more prominent in longer texts. Secondly, taking t-score and MI value into account, non-native learners showed a greater reliance on strong collocations (high t-score), but underuse collocations with low MI value in comparison with native speakers. The result of this study indicated that productions by non-native speakers showed a degree of conservatism, and hence they rely on collocations that were common in English Li and Schmitt (2010) conducted a longitudinal study with learner’s development of collocation learning. With a similar thought from Durrant and Schimtt (2009), Li and Schmitt also utilized a frequency tradition to investigate collocational behaviors by 4 Chinese postgraduate students studying at the University of Nottingham. To understand the improvement of collocation learning overseas, a total of 32 essays and 4 dissertations were collected from the four subjects during one academic year. All adjective-noun collocations were then extracted from learner corpus (150,000 words). 19.

(34) The result showed that 299 types of adjective-noun collocation were identified, and among all, only 41% were classified as frequent and strongly-associated. Li and Schmitt suggested that learners are more inclined to use collocations with high t scores and low MI scores, they also used certain collocations respectively. As for the development of collocational use, little improvement was discovered in numbers of robust collocations in their academic writings. Employing a semi-automatic method, Yang (2015) investigated adjective-noun and verb-noun collocations by Taiwanese postgraduate students in Applied linguistic. Using large academic corpora, she compared the collocational behavior in theses by Taiwanese students with those in published research papers. The sizes of the postgraduate corpus and the published author corpus were 11.7 million and 11.8 million respectively. Yang used 29 most frequent core nouns as a baseline and extracted verbal, adjective collocates from both corpora via Sketch Diff function built in Sketch Engine. All extracted adjective-noun and verb-noun collocations were then examined manually to exclude collocations which contained verbs that can be widely collocated with many words (e.g., consider and regard). A log-likelihood test was later conducted to clarify the underuse and overuse behaviors of learner collocations. The findings showed that published authors’ adjective-noun and verb-noun collocation repertoire was richer than Taiwanese. Low type-token ratio of V-N/A-N collocations in Taiwanese postgraduate theses indicated an ill-balance knowledge of academic collocation. Due to a smaller collocation repertoire, Taiwanese learners were also found to be overly dependent on a few collocations they feel comfortable using.. 20.

(35) 2.3 Collocation Error by Advanced EFL learners Collocational mistakes by EFL learners have been highly researched in order to understand the difficulties that learners might have when producing collocations. A common finding from previous collocational studies pointed out that learners were not aware of collocation restrictedness (Bahns & Eldaw, 1993; Fan, 2009; Granger, 1998; Howarth, 1998; Laufer & Waldman, 2011; Nesselhauf, 2003, 2005; Wang & Shaw, 2008), and thus they tended to make mistakes. Studies on miscollocation took a step further and analyzed the types (Nessehaulf, 2003; Wang & Shaw, 2008) and the causes (Liu, 2013; Wu, Chang, Mitamura, & Chang, 2010) of miscollocations. Liu (2002) investigated verb-noun miscollocations produced by Taiwanese high school and college students via analyzing English Taiwan Learner Corpora (EnglishTLC). EnglishTLC was a one-million-word database built for a web-based writing environment named IWill, which allowed learners to upload their essays and for teachers to correct the essays. To correct the essays on IWill, teachers were provided some comment tags such as, wrong phrase, word verbs, etc, to insert in parts of essays which were written incorrectly. By searching the comment tags in IWill, Liu was able to extract possible miscollocations. After extracting the miscollocations, she further searched the words which were marked as incorrect repeatedly through the KIWIC function. The rationale behind this method was that if a word was judged as incorrect, it might also pose difficulty for other learners. 265 miscollocations were extracted with this method. Further analysis revealed that a high portion of 82% of the miscollocations were verb-noun collocations and over 95% of all the verb-noun miscollocations were from an incorrect use of verbs. The result suggested that learners had difficulty in selecting the correct verbs in verb-noun collocations. Liu also hypothesized that there might be relationships between wrong verbs and their correct counterparts which could 21.

(36) be detected through semantic investigation of the verb pairs (wrong verb paired with the correct one). An analysis of synonym, troponin, and hypernym was then utilized to investigate all the semantic-related collocational errors. The findings showed that synonym relation of the verb pairs troubled learners the most. Adopting German sub-corpus of ICLE, Nessehaulf (2003) examined restricted verb-noun collocations from 32 argumentative essays from German-speaking university students of English. Word combinations were divided into free combination, restricted collocation, and idioms in terms of the restrictedness in the combination. Nessehaulf first extracted all V-N collocations in the corpus manually, and classified word pairs based on their restrictedness. In order to determine the acceptability of collocation in English, dictionaries and native speakers’ judgement were adopted. In total, 1072 verb-noun combinations were extracted from the corpus and near a quarter of the combinations (255 out of 1072) were judged to have at least one mistake. Regarding distribution of mistakes between free combination, collocation and idiom, collocation had the highest rate of error. This finding indicated that collocations posed the greatest difficulty for German learners, especially the ones that verbal component allowed a wider range of nouns as object (i.e., less restricted). The underlying assumption was that learners were too “creative” and had not taken collocational restrictedness into consideration. L1 influence was also, on the other hand, accountable for 45% of the mistakes, making it the major factor of learner collocational error. Nessehaulf also listed out 9 types mistakes in collocations:. 22.

(37) Table 2.4. Types of mistakes in collocation (Nesselhauf, 2003). Type. Example. Wrong choice of verb. *carry out races (hold race). Wrong choice of noun. *close lacks (close gaps). Combination exists but not used correctly. *take notice (to notice) *hold children within bounds (show. Combination does not exist. children where the boundaries lie). Preposition of a prepositional verb missing,. *fail in one’s exam (fail one’s exam). present though unacceptable Preposition of a noun missing, present though unacceptable. Article or pronoun missing, present though. *raise the question about (raise question) *get the permission (get permission). unacceptable Noun used in singular instead of plural or vice versa Syntactic structure wrong. *pass one’s judgement (pass judgement) *make sb. friends (make friends with sb.). Emphasizing L1 background as one major factor of learner collocation error, Wang and Shaw (2008) compare collocations produced of high frequent verbs written by Swedish-speaking learner of English (SSLE) and Chinese-speaking learners of English (CSLE). Each corpus was of the same size, 100 essays of 200 words. They first lemmatized all selected verbs and searched for noun collocates in a ± 5 word span. Inspired by Nessehaulf (2003), Wang and Shaw categorized all word combinations into free combination, collocation and idiom. Combinations were counted as correct if they appeared in the BBI dictionary and at least 5 texts in BNC. If any ambiguity between the dictionary and BNC emerged, a native speaker was inquired for the accuracy of such a combination. Regarding the type of error of SSLE and CSLE, the result showed that mistakes were similar despite the differences in the L1 background. In lexical choice, both SSLE and CSLE made frequent mistakes with the verb do and make. The error, on the other hand, found in restricted collocation was mainly unacceptable 23.

(38) determiners, which stated by Wang and Shaw, dictate the variability of collocation. The following is the type of collocation error found in this study. Table 2.5. Types of collocation error (Wang & Shaw, 2008). Type. Example. Verb choice error. *take the problem (solve the problem). Noun choice error. *have their lines (have their strategies). Lexical error: adjective. *do some protecting work *have more equipments (have more. Grammatical error: noun plurality. equipment) have limited supply of (have a limited. Grammatical error: determiner. supply of). Grammatical error: preposition. *take heed to (take heed of). Grammatical error: syntactic structure. *do favor to ( do sb. a favor) *have a full functional sanitation (have a. Grammatical error: adverb form. fully functional sanitation). Semantic error: correct collocation doesn’t make sense. *have the ability to worry about it. Aside from categorizing the types of miscollocations, some studies investigated the possible causes behind learners’ collocational errors. Liu (1999) studied the miscollocations found in compositions produced by Taiwanese undergraduate students. The researcher collected 94 compositions, 127 examination papers and extracted collocations manually. Utilizing the definition of lexical collocation presented by Benson, Benson and Ilson (1986), Liu (1999) discovered that errors were found mostly in L1 type (verb + noun/pronoun). He further analyzed the errors and proposed seven possible causes:. 24.

(39) Table 2.6. Causes of miscollocation (Liu, 1999) Cause. Overgeneralization False analogy False concepts Misuse of synonym Negative transfer Word coinage Approximation. Explanation. Example. It is a process which learners tend to use a structure in places it does not apply, making it follow a regular pattern. Learners having incorrect understanding of the rules, and therefore make erroneous structures that stand outside of the rules It describes learners’ lack of knowledge of a pattern, so they tried to apply the rules of a known structure Learners not being aware of the restrictions in a fixed expression and therefore choose the wrong synonym. L1 influences the target language use. *I am worry instead of I am worried *Ask you a favor is not a good application of verb+object+object. Learners have a small range of vocabulary, and therefore, in order to communicate their thoughts, they concoct a new word. Misapplication of words occurs, because of the orthographic resemblance between words,. *See sun-up instead of see the sunrise *Release my pressure instead of relieve my pressure. 25. *do plans instead of make plans *Broaden your eyesight instead of broaden your vision *Drink soup instead of eat soup.

(40) Wu, Chang, Mitamura, & Chang (2010) focused on L1 interference of collocational errors and investigated the miscollocations produced by Taiwanese EFL learners. They argued that learners’ misused and correct verbs (presented as verb pairs) might share similar or identical meaning in Chinese. In order to test such a hypothesis, they investigated the verb pairs extracted by Liu (2002) to see if there was any trace of L1 interference at work. Two concepts regarding L1 interference were adopted, split category and direct translation. Split category, as defined by Wu et al., was that two words in one language can only be translated into one term in another language. For instance, verbs “create” in “*create a song” and “compose” in “compose a song” have the same translation in Chinese, “創作”. Learners who might not know the collocational restrictions and the differences between the two verbs might be influenced by their first language and produce a miscollocation, “*create a song”. As for direct translation, it occurred when learners translated the Chinese word directly into English even if the meaning of two words were completely different. Such a concept could be illustrated in one miscollocation commonly found in writings by Taiwanese learners, “*eat medicine”. Since the action of having medical substance in Chinese is “吃”, for learners who were not familiar with its English counterpart “take (medicine)”, they might incorrectly translate “吃藥” directly to “*eat medicine”. The result supported their hypothesis and indicated that at least 84% of the miscollocations from Liu’s study (2002) could be explained by L1 interference. For verb pairs which were in the same semantic relation (misused and correct verbs were synonym), a great portion of 94% of miscollocation were under the influence of L1. The result indicated that L1 was one of the major cause of miscollocations produced by Taiwanese learners.. 26.

(41) 2.4. The Present Study The studies reviewed in 2.2 and 2.3 all pointed out that appropriately using collocation poses a great difficulty for advanced ESL/EFL learners. Due to their limited knowledge in both lexical usage and collocational restrictions, learners, in general, produce a smaller amount and variety of collocations than native speakers do. In academic writings, learners tend to underuse native-like collocations, or overuse less restricted collocations, indicating that collocational problems are far more complex than possessing a small range of vocabulary and noticing the collocational restrictions. In terms of miscollocation, learners’ greatest difficulty lies in correctly choosing verbal component of collocation. Other problems such as L1 interference hinders learners from producing proper collocation. Despite the studies above provides abundant insights of Advanced ESL/EFL learners collocational deficiency, there are still some room for improvement. For starters, to the researcher’s best knowledge, studies dedicated to EFL learners' miscollocation specifically in academic English has yet been conducted. Since in academic writing, being capable of writing precisely is the key to successful writing, it would be beneficial for both learners and teachers to list out collocations which are difficult to learners, along with causes behind them. Secondly, except for Yang (2015), manual extraction of collocation was utilized by most of the collocational studies reviewed above. However, manual extraction was only feasible if the corpora sizes were small. If a larger corpora were used, maual extraction might not be the ideal method to extract miscollocation.. To set this present. study apart from previous collocational research, an online query system, Sketch Engine, was used to extract miscollocations automatically. As claimed by Liu (2014), extracting miscollocations via Sketch Engine minimized laborious work of filtering out 27.

(42) targeted collocations, and the system also “displayed a precise comparison between the language use of native speakers and that of EFL learners in an efficient manner” (Liu, 2014, P. 89). To better understand learners’ problems of V-N collocation in academic English as well as causes behind them, this present study aims to investigate V-N miscollocations in Taiwanese master theses in Applied Linguistics. A list of collocation errors was also provided in hope to facilitate instructors locating problems most learner would encounter in academic writings. The following research questions were answered in this research: 1.. What are the types miscollocation found in Taiwanese master theses in Applied Linguistics?. 2.. Among the types of miscollocations found in Taiwanese master theses in Applied Linguistics, which type of miscollocation is made by learners the most?. 3.. What might be the causes that miscollocations attribute to?. 28.

(43) Chapter Three:. Methodology. This chapter contained three sections. The first section of this chapter explained the composition of the two corpora in this study, MTC and COCABNCacademic. The second part introduced the tools used to extract V-N collocation, Sketch Engine. The final part of this chapter disclosed the procedure of extracting core nouns and identifying verb-noun miscollocations.. 3.1 Corpora used in this study 3.1.1 Master Theses corpus In order to investigate the miscollocations produced by Taiwanese graduate students, a large learner corpus compiled by Yang (2015) was used in this present study. Yang (2015) compiled a learner corpus of 11 million words, selecting a total of 494 master theses from 10 Applied Linguistics/ English Teaching programs in Taiwan published between years 2003-2012. Since advanced academic writing courses were required for students in the Applied English and English Teaching program, the quality of the writings should be high enough for investigation. With a large corpus size and fairly homogeneous quality of master theses, it is safe to conclude that collocational errors found in the results can be generalized as common errors that most Taiwanese postgraduate students in AL/English Teaching programs might produce. To ensure other influential factors that might affect the results, such as the writers’ first language and the format of the theses, only theses written by Chinese native speakers were selected. Details of Master Thesis corpus were listed in Table 3.1 (adopted from Yang, 2015, P. 26).. 29.