高中生英文字彙量與閱讀表現之關聯

全文

(1)國立台灣師範大學英語學系碩. 士. 論. 文. Master Thesis Graduate Institute of English National Taiwan Normal University. 高中生英文字彙量與閱讀表現之關聯. The Correlation between Senior High School Students’ Vocabulary Size and Reading Performance. 指導教授：葉錫南教授 Advisor: Dr. Hsi-nan Yeh 研究生：許碧婷. 中華民國九十八年六月 June, 2009.

(2) 摘要本研究旨在探討台灣高中生之字彙量與大學學測英文閱讀測驗表現之間的關聯。同時，希望藉此了解學測英文閱讀測驗字彙量的需求，以幫助高中生訂定字彙學習目標。參與本研究者為三百零六位就讀北台灣兩所高中二年級的學生。本研究以 Nation (2007) 的字彙量測驗(Vocabulary Size Test)前五個等級，對參與者進行字彙量程度測量，並且施行一份學測相關閱讀測驗以評估其閱讀理解能力。根據兩項測驗成績結果，分析高中生字彙量對英文閱讀表現之影響。本研究所採用的字彙量測驗，係 Nation 根據其所編的 BNC 字族表(BNC Word Family List)製作，而 BNC 字族表是依據英國國家語料庫（British National Corpus）彙編而成。與過去研究工具相較，本研究所用之字族表及字彙量測驗，具有反映英語字彙使用現況的優勢。本研究的主要結果如下：一、本研究受試學生之字彙量與閱讀表現之間的關係達顯著相關(r=.624**)。擁有較多字彙量的受試學生在閱讀測驗的表現顯著優於字彙量較少的受試學生。二、本研究以 BNC 字族表分析學測英文閱讀測驗選文，平均而言，BNC 字族表上的前三千個高頻率字族(high frequency word families)提供近百分之九十二的文章覆蓋率(text coverage)，若將專有名詞計入，則三千字族涵蓋近百分之九十五的學測閱讀測驗選文字彙，達到 Laufer (1988) 所建議之基本閱讀理解字彙需求：95%文章覆蓋率。此外比較不同字彙量受試學生的閱讀表現，結果也顯示三千字族在高中生學測英文閱讀測驗表現上極具重要性。三、本研究的受試高二學生之字彙量差異性極大，且大部份的受試者字彙量程度未足三千字族，亦即未達到學測英文閱讀測驗字彙量需求。基於字彙量與閱讀表現之高度相關，以及充足的字彙量對高中生在學測閱讀測驗表現的重要影響力，本研究建議高中英文教師應協助學生持續增進字彙能力。教學上，教師須先了解學生實際字彙能力，而後對於不同字彙量程度的學生，選取適切的教學方法，以切合學習者不同字彙發展階段的需求。對於低字彙量程 i.

(3) 度的學生，以有效、快速地增加字彙量為學習目標，尤其應重視高頻率字彙優先學習；對於已具備高字彙量的學生，教師可鼓勵學生廣泛閱讀，以進一步提升其字彙能力及英文閱讀實力。. ii.

(4) ABSTRACT This study investigates how vocabulary size correlates with reading performance, specifically, how Taiwanese senior high school students’ receptive vocabulary size influences their reading comprehension in Taiwan’s college entrance examination. This study also aims to estimate the number of word families needed for the reading texts of the Scholastic Aptitude English Test (SAET) and to suggest a reasonable vocabulary learning goal for Taiwanese senior high school students. The subjects were 306 2nd-year students from two senior high schools in northern Taiwan. A reading test sampled from the SAET was administrated to assess the subjects’ reading comprehension. This study made use of the first five levels of Nation’s Vocabulary Size Test (Nation, 2007) to measure the subjects’ receptive vocabulary size. The Vocabulary Size Test is based on the BNC Word Family List, which was developed from the British National Corpus. It is believed that the BNC Word Family List can better reflect the nature of language use of typical English native speakers today and the Vocabulary Size Test may yield proper judgments on the vocabulary size of ESL/EFL learners. The major findings of the study are summarized as follows:. (1) The subjects’. vocabulary sizes and their reading performances on the SAET test had a positive and significant correlation (r=.624**). (2) The first 3,000 word families of the BNC Word Family List made up approximately 92 per cent of the running words of the SAET texts on average, and if proper nouns were included, the coverage provided by the 3,000 word families reached close to the 95% text coverage recommended by Laufer (1988) for successful reading comprehension. Moreover, the most frequent 3,000 word families on the BNC Word Family List were found to have a critical effect on students’ reading performance. (3) There was a great variability in the iii.

(5) second-year senior high school students’ vocabulary sizes and a high percentage of the subjects in this study encountered vocabulary deficiency in the 3,000 word families. Considering the importance of vocabulary size to reading performance, this study calls for students’ continuous development in vocabulary size and acquisition of the 3,000 high frequency word families first to facilitate their reading comprehension.. English teachers in Taiwan are suggested to be aware of individual. student’s vocabulary competence first and then provide proper assistance to students’ vocabulary growth.. For students whose vocabulary sizes still fall short of the size. requirement, the vocabulary learning goal is to build larger vocabulary size and an explicit, intentional learning through intensive reading is the most effective way to enhance their vocabulary acquisition. For students whose vocabulary sizes have already achieved high level, implicit and incidental learning through extensive reading may be most beneficial to the advancement of their lexical knowledge and overall reading proficiency.. iv.

(6) ACKNOWLEDGEMENTS I am fortunate to have had support and contributions from many people surrounding me throughout this study. Had not been for their assistance, this thesis would never come into existence. My deepest gratitude goes to my advisor, Professor His-nan Yeh, who inspired me both in doing this study and in being a good teacher. Over the course of this study, he squeezed time out from his tight schedule to revise my draft and discuss with me patiently, giving me insightful advice. With his constant encouragement, he guided me through all difficulties to the final self-fulfillment.. I greatly admire his. professional knowledge and his intense enthusiasm about education.. Professor Yeh. is really a wonderful mentor and role model to me. My heartfelt thanks also go to two committee members, Professor Hao-jan Chen, and Professor Hsueh-ying Yu, who provided me with constructive suggestions to improve the quality of my thesis. Without their professional comments, this thesis would not be complete. I am earnestly grateful to these two great teachers for their invaluable guidance in making my thesis more refined. I am deeply thankful to my dear colleague, Ms. Chiung-wen Chang, who offered me helpful ideas and opinions to solve problems I encountered in the process. Special thanks to my other colleagues in National Keelung Senior High School for their precious friendship. Whenever I felt frustrated by the invisible burden, they comforted me with supportive words. I would offer my sincere thanks to my fellow English teachers in National Keelung Senior High School and National Keelung Girls’ Senior High School, for assisting me in administering tests to their students.. I am grateful to all of the. teachers and the students for their participation and cooperation in this study. v.

(7) Finally, I would like to dedicate this thesis to my family, especially to my parents, my mother-in-law, my beloved husband and my darling daughter and son. They have been with me through thick and thin during the entire master program to offer support, and constant belief in my abilities. It was their love that gave me strength to carry on and to complete this thesis.. vi.

(8) TABLE OF CONTENTS Abstract (Chinese) ......................................................................................... i Abstract (English) .......................................................................................... iii Acknowledgements........................................................................................ v Table of Contents ........................................................................................... vii List of Tables.................................................................................................. ix. Chapter One Introduction .......................................................... 1 Background and Motivation .................................................................... 1 Purpose of the Study ................................................................................ 6 Research Questions.................................................................................. 7 Significance of the Study ......................................................................... 7. Chapter Two Literature Review................................................. 9 Vocabulary and Reading Comprehension ................................................ 9 Vocabulary Size ....................................................................................... 11 Vocabulary Size and Text Coverage............................................... 12 Threshold Vocabulary Size for Reading Comprehension ............. 13 Text Coverage of the SAET Reading Texts by the English Reference Word List.............................................................. 15 Vocabulary Size of EFL/ESL Readers ........................................... 19 Vocabulary Size of Taiwanese EFL Learners................................. 20 Word Frequency....................................................................................... 22 BNC Word Family List ............................................................................ 24 Comparison between the BNC Word Family List and the English Reference Word List ................................................ 26 Vocabulary Test........................................................................................ 36 The Vocabulary Levels Test............................................................. 37 The Vocabulary Size Test................................................................. 39. Chapter Three Methodology...................................................... 42 Participants............................................................................................... 42 Instruments............................................................................................... 43 Vocabulary Test.................................................................................. 43 Reading Comprehension Test ............................................................ 46 Procedures................................................................................................ 49 Data Analysis ........................................................................................... 49. Chapter Four Results and Discussion ........................................ 52 Vocabulary Size of the Participants ......................................................... 52 Correlation between Vocabulary Size and Reading................................. 56 vii.

(9) Reading Performance at Distinct Levels of Vocabulary Size ................. 59 The Vocabulary Size Needed for Comprehending the SAET Reading Texts.................................................................................. 66. Chapter Five Conclusions.......................................................... 72 Summary of the Major Findings .............................................................. 72 Implications of the Study ......................................................................... 74 Limitations of the Study and Directions for Future Research ................. 78. References..................................................................................... 81 Appendixes ................................................................................... 86 Appendix A Vocabulary Test................................................................ 86 Appendix B Reading Comprehension Test.......................................... 91 Appendix C Readability Test of 2005-2009 SAET Reading Text ....... 97 Appendix D Subjects’ vocabulary size at each of the 1,000 words frequency levels.............................................................. 98 Appendix E Subjects’ accumulated vocabulary size at the five levels............................................................................... 101 Appendix F Statistic analyses on the three groups at 3,000 words level with over 1,100 words at L4&L5........................... 106 Appendix G Asian Examinees’ Scores on the TOEFL Internet-Based Test......................................................... 107. viii.

(10) LIST OF TABLES Table 1 Table 2 Table 3. Vocabulary size estimates and exam equivalents ................................. 11 Vocabulary size and coverage in novels for teenagers ......................... 13 2002-2006 SAET reading text coverage distribution by CEEC’s. Table 4 Table 5. English Reference Word List ................................................................ 17 English vocabulary size of foreign learners.......................................... 20 Text coverage by the GSL and the AWL in four different kinds. Table 7. of text .................................................................................................... 23 Distributions of the 6,480 words against the BNC Word Family List ........................................................................................................ 29 Distributions of the 4,320 words of Level 1 to Level 4 of. Table 8. English Reference Word List against the BNC Word Family List ........................................................................................................ 30 Examples of differences between the two lists, Words vs.. Table 6. Families................................................................................................. 32 Table 9 Examples of differences between the two lists, in level classification ......................................................................................... 34 Table 10 Characteristics of the chosen reading samples from SAET.................. 48 Table 11 Table 12 Table 13. Vocabulary size at the five frequency levels ........................................ 53 Numbers and percentage of subjects at the five frequency levels ........ 54 Correlation between vocabulary size at the 5 frequency levels and reading performance ...................................................................... 56. Table 14 Table 15 Table 16. Reading scores across three groups at 3,000 words level..................... 60 One-Way ANOVA for three groups at 3,000 words level ................... 61 Post Hoc Multiple Comparisons among three groups at 3,000 words level............................................................................................ 61 Table 17 Comparisons among the three groups at 3,000 words level (with different vocabulary size at the 4th and 5th frequency levels).................................................................................................... 64 Table 18 2002-2006 SAET reading text coverage distribution by the Table 19. Table 20. BNC Word Family List ........................................................................ 67 2002-2006 SAET reading text coverage distribution comparison of BNC Word Family List and CEEC’s English Reference Word List ............................................................................. 68 Text Coverage of the 6 passages of the Reading Test........................... 69. ix.

(11) CHAPTER ONE INTRODUCTION. Background and Motivation In Taiwan, with the growth of economy and the trend of English as an international language, the government has enacted an English-language policy to commence English teaching and learning from the third grade in primary schools. Hence, English has become a required subject in the early-stage of the nine-year compulsory education.. However, even with the enhancement of English language. policy in education, students’ English ability remains an unresolved thorny issue. According to data from TOFEL (Test of English as a Foreign Language), the mean scores of Taiwanese examinees who took the TOEFL Internet-based version of the TOEFL test (TOEFL iBT) in 2007 and 2008 are 72 and 73, respectively, lagging behind most of our neighboring Asian countries, as shown on Appendix G. Such an unsatisfactory outcome is echoed by another statistic report released by IELTS (International English Language Testing System).. The mean band score of. Taiwanese academic-purpose test takers in 2007 is 5.59 out of the nine-band scale, ranked number 16 among the 20 most frequently IELTS-taking regions (from http: // www.ielts.org/teachers_and_researchers/analysis_of_test_data/test-taker_performance _2006.aspx). The performance of Taiwanese examinees on these two major English language tests has raised concern over Taiwanese students’ English proficiency in this global village, where English competence is regarded as international competitiveness (Chang, 2007). The underperformance of Taiwanese examinees may reflect the pedagogical 1.

(12) practice and weakness in English education in Taiwan. The practice includes that English is treated as a foreign language rather than a second language.. As a result,. Taiwanese students are motivated to learn English mainly because English is a required subject in the curriculums and in the entrance examinations as well. Another practice is related to the way in which English is taught or is tested. pedagogy tends to be rather reading orientated.. The. English curriculums starting from. the senior high stage predominantly emphasize on students’ reading skills rather than the mastery of all four language abilities. Students are introduced to reading-centered teaching materials and are often measured by reading-oriented tests.. Furthermore,. under the examination-oriented culture, the two English examinations in the college entrance examination: the Scholastic Aptitude English Test (SAET) and the Department Required English Test (DRET) will inevitably continue to have great impact on English education quality in Taiwan and even on students’ level of English competence.. In consideration of the crucial role of college entrance examination to. high school students’ English education, this study focuses on investigating the reading in the context of examination, specifically the Scholastic Aptitude English Test. In order to help students cope with the challenge of taking college entrance examination, this study tries to investigate the correlation between vocabulary size and reading performance in the SAET. The SAET is a large-scale standardized achievement test aiming to measure high school students’ basic English competence in Taiwan and the test result is used by almost all colleges to screen the qualified candidates applying for admittance.. The. SAET contains questions that ask for grammatical, vocabulary knowledge, reading comprehension skills, and writing skills. As a SAET test composes of 70% of reading and 30% of writing, it is more to evaluate examinees’ competence in reading and writing rather than the macro-skills of listening and speaking. 2. Among all the test.

(13) items, reading comprehension section occupies a primary allocation of 30 percent, so it seems reasonable to assume that English reading competence is important for students’ success in this examination. Many of the past studies have shown that adequate vocabulary is important for fluent reading and a well-developed vocabulary is a prerequisite for fluent reading (e.g., Allington, 2000; Juel, 1995; Chall, 1983, as cited in Joshi, 2005). Students with poor vocabulary would struggle to decode the basic elements of a text and find it hard to develop any higher level understanding of the content (Read, 2000). Studies in reading performance bring together the assumption that vocabulary competence may strongly influence reading comprehension (Alderson, 2000).. It has been long noted. that learners’ acquisition of vocabulary is essential for reading comprehension and reading development.. Struggling readers often do not make gains in their reading. comprehension because of a limited vocabulary (Rupley, 2005).. In the researcher’s. teaching experience, senior high school students find that insufficient vocabulary size is a great obstacle to their success in comprehending the reading texts, while taking the two levels of English tests in Taiwan’s college entrance examination. Their small amount of vocabulary frequently prevents them from choosing the right answer in a reading test. Despite the fact that many high school students have learned English for a lengthy period of time starting from their fifth grades in the elementary schools, some of them still experience lexical gaps in comprehending English reading texts.. Their. inadequacy of vocabulary or difficulty in vocabulary learning may have held them back from making progress on most formal language tests. Based on the researcher’s observation, students with larger vocabulary size tend to perform better in the reading tests, and thus obtain higher scores in English examinations. Hsu’s (2004) research on Taiwan’s college entrance examination also showed that vocabulary recognition 3.

(14) significantly determines the examinee’s test performance. There is no doubt that improving the ability of reading is an urgent matter for most EFL learners in Taiwan.. In addition, as becoming a competent reader is often. associated with developing a large number of words in the target language, apparently, the first priority for Taiwanese students is to develop large vocabulary. Growth in vocabulary may enable learners who are experiencing reading difficulties to better comprehend what they read, make inferences within and between texts they read, and increase their ability to comprehend (Heilman, Blair, & Rupley, 2002; Ruddell, 1994). Studies (Laufer 1992, 1997; Nation, 1993) have found that there is a threshold for vocabulary associated with reading comprehension. Laufer (1992) referred to threshold for vocabulary as “the number of words the reader must possess in his lexicon to be able to read in L2”.. A reader with vocabulary size below the threshold. will have difficulties in L2 reading. The important role of vocabulary knowledge in second language reading comprehension has been explored in past research (e.g., Laufer, 1992, 1996; Liu & Nation, 1985).. However, most of the past research studied reading for general. purpose and the vocabulary sizes were measured by the Vocabulary Levels Test (VLT).. A weakness of these studies was that the vocabulary test available at the. time was limited to the VLT.. The VLT is based on the General Service Word List to. test the first 2,000 words of English (West, 1953); to estimate beyond the first 2,000 word families, it uses other lists such as the old Thorndike and Lorge List (1944) and the University Word List (Xue & Nation, 1984). Moreover, the VLT covers five non-continual frequency levels of vocabulary: the 2nd 1000, the 3rd 1000, the 5th 1000, the Academic Word List, and the 10th 1000. To overcome this difficulty, Professor Nation developed a substantial number of word family lists from the British National Corpus, hereafter referred to as the BNC Word Family List. 4. Furthermore,.

(15) in attempt to fill the gaps in the old Vocabulary Levels Test, Nation also made a Vocabulary Size Test (VST) (Nation, 2007) based on the BNC Word Family List. Since the BNC Word Family List is developed in the most recent time and is made based on the British National Corpus, the list reflects the nature of language use of typical English native speakers in modern time. The BNC Word Family List, argued by Nation (2006), may provide more accurate estimates of the number of word-families needed to read and to listen to English.. It is expected that the VST,. sampled from the most frequent 14,000 BNC Word Family List may yield proper judgments on the vocabulary size of English learners.. This study makes use of the. latest-developed VST to measure students’ receptive vocabulary size; it aims to examine the correlation between vocabulary size and reading and to explore the number of words needed for comprehending SAET reading texts. It is reported that when a reader knows more than 95% of the running words of a text, comprehension is possible (Hu & Nation, 2000; Hirsh and Nation, 1992; Laufer, 1989, 1992; Liu and Nation, 1985).. That is to say, a vocabulary size that covers 95%. of SAET reading texts may serve as a reasonable vocabulary learning goal for senior high school students in Taiwan.. With a view to investigating the vocabulary load of. the SAET texts, this study examined the text coverage of the SAET reading texts and made a comparison of the text coverage figures between the BNC Word Family List and the English Reference Word List, a list released by CEEC in Taiwan, specifically for the reference for SAET and DRET.. In setting an effective vocabulary learning. goal aiming at facilitating reading comprehension, senior high students should not only know how much vocabulary to learn, but also be aware of what frequency of words to learn. Frequency refers to the degree of occurrence of the words and frequency ranking closely affects the order in which the words are acquired and should be learned.. High frequency words have proven to provide high coverage in a 5.

(16) wide range and variety of texts (Nation, 2004).. Nation (2006) claimed that the BNC. Word Family List may be a good representation of the commonly used vocabulary and ranked in a reasonable frequency sequence.. The high frequency words in the. BNC Word Family List are very likely to be more immediately useful for EFL young learners and therefore represent a practical vocabulary learning goal.. This study. looked at the high frequency words in the BNC Word Family List to provide Taiwanese senior high school students with a clearer picture of what high frequency word families they need to concentrate on first. Performing well in the college entrance examination is an important learning goal for most Taiwanese senior high school students. That is the reason why this study is situated in the context of the SAET.. This study used the most recent BNC Word. Family List and its corresponding vocabulary test: the Vocabulary Size Test to estimate how many word families Taiwanese high school students have and to investigate how their vocabulary sizes correlate with their reading performance on the SAET.. Purpose of the Study This study sets out to explore what role vocabulary size plays in reading comprehension. In addition to the correlation between vocabulary size and reading comprehension, this study reports on the vocabulary sizes of high school students. Furthermore, students with different vocabulary sizes and reading performance are compared and analyzed to see how large a vocabulary is needed to deal with the reading texts of the SAET.. Therefore, the purpose of this study is to suggest a. reasonable vocabulary learning goal for Taiwanese senior high school students and to estimate how close students are to this crucial vocabulary goal. The findings from this study provide some empirical evidence to help struggling readers in senior high 6.

(17) schools with more informative guidance in vocabulary learning.. Research Questions The study aims to examine how students’ receptive vocabulary sizes correlate with their reading comprehension in the college entrance examination in Taiwan. The researcher explores the following questions: 1.What are the vocabulary size levels of the participants? 2.What is the correlation between vocabulary size and reading performance? How do participants with distinct levels of vocabulary size differ in their reading performance?. 3. How large a vocabulary size is needed for comprehending the SAET reading texts?. Significance of the Study In the examination-oriented culture, obtaining high scores on the entrance examinations is considered the most significant motivation for English learning in Taiwan. One of the biggest challenges for Taiwanese senior high school students is the deficits in English vocabulary and the limited time for either direct vocabulary instruction in English classes or incidental acquisition through extensive reading in out-of-school settings. Students who are exposed much less to English words thus experience slow vocabulary development.. Their limited English vocabulary hinders. them from comprehending texts and may further result in their poor performance in the English tests. With these considerations lying behind, the present study chooses to focus on investigating the correlation between vocabulary size and reading performance in the SAET.. The significance of this study could be viewed from both 7.

(18) instructional and learning perspectives.. From the instructional perspective, this. study provides teachers with precise descriptions of the status quo of students’ lexical competence and reading comprehension ability.. As bi-polarity phenomenon is a. serious problem in Taiwan’s English teaching and learning environment (Chang, 2006), teachers are faced with difficulties in teaching to mixed-proficiency class. By identifying individual student’s real competence level and learning problems, teachers will be able to invest more resources in areas of greatest need—the slow English learners, to help retrieve their learning confidence in English, and to provide individual students with more suitable and appropriate teaching material and methods. From the learning perspective, students have to possess adequate vocabulary in order to carry out the academic reading activities and pass the reading-oriented college entrance examinations. However, because of the limited time available for English learning and vocabulary development, most high school students in Taiwan have to work with great effort to meet the requirements set in the English curriculum and to achieve high scores in the college entrance examinations. The findings of this study can provide a better basis for decisions about what vocabulary size is good for high school students’ reading comprehension and whether it is worthwhile for them to spend time on and thus it helps reduce students’ learning burden.. 8.

(19) CHAPTER TWO LITERATURE REVIEW This chapter gives a review of the literature concerning vocabulary size, reading comprehension, and the connection between them. To begin with, research into the relationship between vocabulary size and reading comprehension is reviewed. perspectives on the nature of vocabulary size and word lists are discussed.. Next,. Also, we. take a preliminary look at the text coverage figures of the SAET reading texts.. At. last, two major tests involved in measuring EFL/ESL learners’ vocabulary size are discussed.. Vocabulary and Reading Comprehension Knowledge of words and their meanings is a crucial component of language proficiency, both for first language acquisition and for second and foreign language learning. No text comprehension is possible, either in one’s native language or in a foreign language, without understanding the text’s vocabulary (Laufer, 1992). Vocabulary is an essential building block of language and has a strong relationship with reading comprehension (Schmitt, 2000). A lot of second language studies (e.g., Qian, 2002; Nation, 2001; Laufer, 1996) considered vocabulary knowledge an important source of variation in reading comprehension and demonstrated that reading comprehension is related to vocabulary knowledge, more strongly than to the other components of reading and that vocabulary serves as a good predictor of reading success (Qian, 2002). Vocabulary knowledge has crucial effects on higher-level language processes such as grammatical processing, construction of schemata and text models (Adams & Collins, 1977; Chall, 9.

(20) 1987).. Readers may compensate for their lack of knowledge of particular words by. using contextual clues, schema and background knowledge of the subject matter to guess and comprehend the text.. However, problems with vocabulary may pose. greater obstacles for second or foreign language readers, at limited proficiency levels. It is probably because that for readers with limited vocabulary, the processing of lower-level elements may take up too large a part of cognitive resources which otherwise should have been allotted to higher-level processing. The. sizable. correlation. between. vocabulary. knowledge. and. reading. comprehension has been a great interest for reading educators for decades (Davis, 1944, 1968).. In terms of correlation studies, Laufer’s (1992) study, in which 92. first-year college students participated, discovered a correlation of .50 between vocabulary size—measured by Vocabulary Levels Test by Nation (1983)—and reading comprehension, and a correlation of .75 between vocabulary size—measured by Eurocenters Vocabulary Test by Meara and Jones (1989)—and reading comprehension. Still in another study (Laufer, 1996) a correlation of .71 was found. Research with Japanese learners of English (Koda, 1989, cited in Qian, 1999) also yielded correlations of .69 to .74 between vocabulary ability and reading comprehensions. In Taiwan, Huang (1999, cited in Huang 2003) found that Taiwan students’ reading comprehension scores were 69% explained by their vocabulary knowledge score. Laufer (2001) pointed out that the correlation of vocabulary knowledge and reading is particularly strong. As vocabulary is one of the best predictors of overall language ability, perhaps it is not surprising that vocabulary knowledge and its size should link so well with general language exams and hierarchies of ability such as the Common European Framework for Foreign Languages (Council of Europe, 2001).. Vocabulary can be. shown to predict performance in a number of specific language skills on which these 10.

(21) exams and hierarchies rely. For example, Meara and Buxton (1987) successfully correlated vocabulary size with scores on reading comprehension tests, as shown on Table 1. Table 1 Vocabulary size estimates and exam equivalents CEF level. Cambridge ESOL Test. Vocabulary size/5,000. Al. Starters, etc.. < 1,500. A2. KET. 1,500-2,500. Bl. PET. 2,750-3,250. B2. FCE. 3,250-3,750. CI. CAE. 3,750-4,500. C2. CPE. 4,500-5,000. Source: Meara & Milton (2003: 12) The findings in past research on first language reading and second language reading demonstrate the role of vocabulary in reading process, and the rather high correlations between reading and vocabulary.. Vocabulary Size Many reading researchers claim that adequate knowledge of words is a prerequisite for effective language use. Words are the basic building blocks of language, the units of meaning from which larger structures such as sentence, paragraphs and whole texts are formed (Read, 2000). Researchers (Read, 1989; Qian, 1999) asserted that vocabulary knowledge should at least comprise two aspects: vocabulary size (breadth) and vocabulary quality (depth).. Vocabulary size refers to. the number of words the meaning of which a learner has at least some superficial knowledge while vocabulary quality means how well a learner knows a word (Qian, 11.

(22) 2004).. The criteria for knowing a word are usually quite liberal in past studies.. Nation (1990) classified vocabulary knowledge into two primary dimensions, namely, the receptive and productive knowledge.. Because reading text is presented in the. written form of the language, it is recognized that receptive vocabulary knowledge has more association with reading and has important contribution to reading comprehension. Several estimates of receptive and productive vocabularies conclude that receptive vocabulary is double that of productive vocabulary. This study focuses on measuring the receptive vocabulary size rather than the productive one, as it is to explore how vocabulary size correlates with the reading performance in a college entrance examination for Taiwanese EFL learners in the senior high school stage.. Vocabulary Size and Text Coverage Text coverage refers to “the percentage of running words in a text or corpus that is also in, or covered by, a particular word list” (Nation & Waring, 1997). In deciding how much vocabulary is needed to read a text, we have to know first how much unknown vocabulary can be tolerated in a text before it interferes with comprehension. As vocabulary size for adequate comprehension of a reading text is related to the ratio of unknown to known words, it alone does not represent any concrete standard to be used for setting vocabulary learning goals but has to be considered in conjunction with text coverage to be meaningful. Text coverage is a prerequisite with which we make estimation of the vocabulary size a leaner needs to have to be able to cope with numerous reading texts.. If we want to explore the two questions: “To what extend. can a reader comprehend the reading text?”, and “What is the relationship between vocabulary size and reading?”, we have to know both what coverage of the words in a text is and what vocabulary size of a reader is.. In taking reading tasks, skilled. readers can tolerate a small proportion of unknown words in a text without disruption 12.

(23) of comprehension and can even infer the meanings of the unknown words from sufficiently rich contexts.. However, if the proportion of unknown words is too high,. comprehension is disrupted (Carver, 1994).. Liu and Nation (1985) and Nation (1990). stated that 95% coverage of general texts was a percentage necessary for successfully using context to infer the meanings of unknown words.. Research by Laufer (1988a). also suggested that 95 per cent coverage was sufficient to allow reasonably successful guessing of the meaning of the unknown words and thus reasonable comprehension of a text.. The 95 per cent coverage means that one word in every twenty is unknown.. Hirsh and Nation (1992) conducted a study looking at the text coverage of novels written for teenage readers.. As shown on Table 2, the study concluded that a. vocabulary size of 3,000 to 5,000 words, covering 95 to 98 per cent of the novel text, provided a very good basis for second language learners’ reading comprehension. Table 2 Vocabulary size and coverage in novels for teenagers Vocabulary Size. % Coverage. Density of Unknown Words. 2,000 words. 90. one in every 10. 2,000+ proper nouns. 93.7. one in every 16. 2,600 words. 96. one in every 25. 5,000 words. 98.5. one in every 67. Research by Hirsh and Nation (1992), cited on Nation & Waring (1997). Threshold Vocabulary Size for Reading Comprehension In the studies of vocabulary and reading, the threshold level refers to the size of vocabulary needed to deal with reading materials with acceptable comprehension. It was found that learners whose vocabulary was below a certain threshold level struggled to decode the basic elements of a text, to the extent that they found it hard to. 13.

(24) develop any higher level understanding of the content (Read, 2000).. Laufer and Sim. (1985, cited in Hu & Nation, 2000) conducted a study to determine the threshold of comprehending an EAP (English for Academic Purpose) text in the First Certificate in English Exam.. It was concluded from their study that a threshold vocabulary was. the minimum vocabulary level that readers needed to achieve first to activate their background knowledge, reading skills and reading strategies for reading comprehension. The estimates of vocabulary needed for effective understanding of authentic texts may vary, depending on purpose of reading task and reading text, and specific topic.. For general reading purposes, Liu and Nation (1985) and Nation (1990). suggested that knowledge of 3,000 words was needed for understanding general texts. In addition, Fox (1987) noted that a vocabulary of some 10,000 words would be adequate receptive vocabulary, equaling to approximately 7,000 separate word families defined by Nation, 2001.. An even larger estimate rises from such reading. materials as journalistic magazines. Hu and Nation (2000) examined the relationship between text coverage and reading comprehension for non-native speakers of English with a fiction text. They found from the study that at 80% of text coverage (that is, 1 unknown word out of every 5 words) provided by the first 2,000 word families, reading comprehension did not occur because reading skills and background knowledge could not be activated to make up for lack of vocabulary knowledge. With a text coverage of 90% (i.e., 1 unknown word in 10), then a small minority gained adequate comprehension. With a text coverage of 95% (i.e., 1 unknown word in 20), a few more gained adequate comprehension. At 100% coverage, most gained adequate comprehension. It was proposed that 98% text coverage (i.e., 1 unknown word in 50) would be needed for most learners to gain adequate comprehension. After reaching the threshold of the suggested 98% coverage figure, further increase brought no 14.

(25) noticeable effects on comprehension, reading pleasure, or reading speed. In the case of academic texts, knowing the meanings of 3,000 word families covered approximately 88% of a text. Another 4% of the text was covered by proper nouns and another 3% by technical vocabulary. If readers knew these words, it was easier for them to infer the meaning from academic texts. Laufer (1992) found that in order to reach roughly 95 per cent coverage of the running words in the reading text, readers needed a minimum of 3,000 word families to understand unedited text.. She. concluded that a vocabulary below the 3,000 word-family level was a more significant factor in limiting English reading comprehension for academic purposes than learners’ general academic ability, such as L1 reading skills.. Some other studies suggested. that to reach 95 per cent coverage, a vocabulary size of about 4,000 word families would be needed—2,000 high-frequency general service words, 570 Academic List words (Coxhead, 1998) and 1,000 or more technical words, proper nouns, and ‘low-frequency words’ (Nation 2001: 147). Chen (1999) quoted King’s (1989) estimate as saying that 16,000 words and 20,000 words were needed for EFL learners to read Newsweek and Time, respectively. The most recent study on estimate of vocabulary was made by Nation (2006) based on the BNC Word Family List. Nation argued that if 98% coverage of a text was needed for unassisted comprehension, then an 8,000 to 9,000 word-family vocabulary was needed for comprehension of authentic written texts (e.g., novels or newspapers) and a vocabulary of 6,000 to 7,000 word families for spoken texts.. Text Coverage of the SAET Reading Texts by the English Reference Word List The review of literature showed that when deciding what vocabulary second language learners need, we should take into consideration the three important factors: 15.

(26) vocabulary size, text coverage and word frequency.. To help students solve the. problem of inefficiency and inadequacy in vocabulary learning, it is necessary to decide a reasonable vocabulary learning goal for them to achieve.. As mentioned. earlier, before setting vocabulary learning goals, it is essential to find out how much vocabulary students need to know for specific language uses, such as reading newspapers, or novels, watching movies, or taking part in a conversation (Nation, 2006).. Students in Taiwan have to possess adequate vocabulary to carry out the. reading tasks in the reading-oriented college entrance examinations. However, with limited time available for their vocabulary development, they have to consider what the cost/benefit is behind learning these words.. That is to think about the association. of the English tests in the college entrance examination with the text coverage and frequency of words.. To facilitate senior high school students’ vocabulary learning,. the College Entrance Examination Center (CEEC) released the English Reference Word List (ERWL) (Jeng et al., 2002) as an English reference word list for the two kinds of English tests in the college entrance examination of Taiwan, namely the Scholastic Aptitude English Test (SAET) and the Department Required English Test (DRET). The compilation of the English Reference Word Lists was sponsored by the CEEC and was carried out between September 1, 2000 and June 30, 2002. The List contains 6,480 words in American English, selected on the basis of their word frequencies and in consideration of the Chinese cultural background. These 6,480 words are classified into six levels, each level with 1,080 words. The words of levels 1 to 4 are used as reference for SAET, and the words of levels 1 to 6, for DRET (extracted from CEEC website).. The word lists are merely for reference but do not. set limits to the English vocabulary to appear in the two examinations. As the English Reference Word List functions as a leading guidance for vocabulary learning in Taiwan senior high schools, it is significant to know its text 16.

(27) coverage percentage of the SAET reading comprehension section. Table 3 shows the coverage of the 2002 to 2006 SAET reading texts by the English Reference Word List. The distribution of text coverage is shown in the order of ascending level. Table 3. 2002-2006 SAET reading text coverage distribution by CEEC’s English Reference Word List (cumulative %) Year. 2002. 2003. 2004. 2005. 2006. Mean. 1st 1,080. 59.51. 62.78. 60.23. 59.17. 59.42. 60.22. 2nd 1,080. 70.00. 75.20. 72.27. 72.21. 73.03. 75.54. 3rd 1,080. 76.69. 82.01. 79.85. 79.03. 79.10. 79.34. 4th 1,080. 82.15. 86.67. 82.90. 84.65. 83.95. 84.04. 5th 1,080. 83.65. 88.53. 84.42. 85.51. 84.91. 85.40. 6th 1,080. 84.37. 89.42. 86.00. 86.41. 86.45. 86.53. Proper Nouns. 94.21. 94.08. 93.94. 92.26. 92.16. 93.33. (Off-list). (5.79). (5.92). (6.06). (7.74). (7.84). (6.67). 100.00. 100.00. 100.00. 100.00. 100.00. 100.00. Level Ranking. Source: Yu (2006), Analysis on the 2006 SAET Test Items As shown, only about 80% and 85% of the words in the SAET reading texts were within the first three and four levels of the CEEC’s English Reference Word List respectively.. Words in Level one of the English Reference Word List made up an. average of 60% of the text, Level 2, 15%; Level 3, 4% and Level 4, 5%. Aside from that, most proper nouns, such as people’s names, the names of cities, and the names of mountains, not included in the list accounted for an average of 6% text coverage. The accumulated text coverage by all the six levels of the 6,480 words in the English Reference Word List was 86%.. 17.

(28) The compilation of the English Reference Word List is not based on word family but on the basic lexical item including its lemma, the inflected forms of the base word, and sometimes its regular derived forms such as “-ment”.. If we count the coverage. in terms of word family, applying the conversion formula by Nation (1983, cited in Laufer, 1992): 1 word family converted into 1.6 lexical items, the rate of coverage in each level of the English Reference Word List will be slightly higher.. Because a. word family list covers a broader range and a larger number of words than a lemma list, it is likely a word family list may provide higher text coverage than a lemma list. As discussed earlier, the relation between text coverage (percentage of running words in a text) and vocabulary size (ratio of unknown to known words) would affect the degree to which a reader could comprehend a reading text.. If a learner had a. vocabulary size of the first three levels of the English Reference Word List, he would comprehend near to 80 percent of the reading texts in SAET, which meant that he would meet one unknown word in every five known words.. A 95 percent coverage. of running words in a text is necessary for second language learners to successfully understand a reading text (Liu and Nation, 1985).. If a reader does not have a. vocabulary size of the 95% coverage of the text, it is neither possible for him to compensate adequately for his lack of such vocabulary knowledge to guess or infer vocabulary from context, nor will he be able to achieve an ideal comprehension of the reading text. Nevertheless, as shown by Table 3, even all the six levels of the 6,480 words together in the English Reference Word List hardly reached 90% coverage of the SAET reading text.. If the necessity of a vocabulary size of 95 percent text. coverage was applied, all the 6,480 words in the English Reference Word List seemed inadequate to prepare students well enough to comprehend the texts in the SAET reading comprehension section.. 18.

(29) Vocabulary Size of EFL/ESL Readers In literature, estimates of vocabulary size varied enormously because of the wide variation in the methods of testing. Nagy and Anderson (1984) suggested that undergraduates had around 20,000 word families in their vocabulary. Nation (1990) cited studies showing that by 8.5 years of age, English-speaking children had roughly 4,500 word families and it was suggested that first-language learners added between 1,000 and 2,000 word families per year to their vocabulary. The vocabulary size of 15,000-20,000 word families (Schmitt, 2000) of a native speaker is a possible and ideal goal for second language learners, but not all learners have to do that, even if a great number of second language learners do achieve near-native vocabulary size (Nation & Warning, 1997).. It was claimed that learners of English as a foreign. language had between 1,000–2,000 words after 5 years of learning (Barnard, 1961; Quinn, 1968). Following is a table summarizing the vocabulary sizes of EFL/ESL learners reported in research studies. It is clear that most foreign learners’ vocabulary sizes fall well short of the size requirement of achieving 95% text coverage (Schmitt, 2008).. 19.

(30) Table 4. English vocabulary size of foreign learners. Country. Vocabulary Hours of Size instruction. Reference. Japan EFL University. 2000. Shillaw 1995. 800-1200. 2300. Barrow et al. 1999. China English majors. 4000. 1800-2400. Laufer 2001. Indonesia EFL University. 1220. 900. Nurweni & Read 1999. Oman EFL University. 2000. 1350+. Horst et al. 1998. Israel High school graduates. 3500. 1500. Laufer 1998. France High School. 1000. 400. Arnaud et al. 1985. Greece Age 15, high school. 1680. 660. Milton & Meara 1998. Germany Age 15, high school. 1200. 400. Milton & Meara 1998. From Schmitt, 2008, p.332. Vocabulary Size of Taiwanese EFL Learners Unlike the native speaker whose vocabulary develops naturally and grows rapidly, most Taiwanese students take English vocabulary learning as a matter of demand to comply with the curriculum requirement. Because of less motivation and exposure, Taiwanese students’ vocabulary growth is relatively limited and the learning results are unsatisfied. In spite of having studied English for more than eight years, commencing from their 3rd grade in the primary schools, some junior high graduates have just little vocabulary, and even some have no vocabulary competence at all. For those learners with inadequacy of English competence in junior high school, the extended and high requirements of English curricula in senior high make their hardships and pressure get even more severe when they are in senior high school English classroom. English curricula for junior high school stipulate a productive 20.

(31) vocabulary size of 1,000 words and receptive vocabulary size of 2,000 words whereas senior high school English curricula call for much more vocabulary (Yang, 2006). According to the National Curriculum of English for High Schools set by the Ministry of Education, the vocabulary in English textbook readings should be of the 7,000 most frequently used words.. It is a great challenge for not only the students but also for. the English teachers.. English teachers in senior high schools are responsible to. upgrade students’ vocabulary size from 2,000 words to the 7,000 words or more to cater for the needs of students who prepare themselves for the college entrance examinations. In the past, there had been a number of studies on vocabulary size conducted in Taiwan, from junior high level to college level, but with college level most frequently researched. In Chen’s 1998 and 1999 research, he examined the vocabulary size of 174 non-English major college freshmen and found 60% to 70% of his subjects reached the 2,000—3,000 word level while still 30% to 40% of students fell below the 2,000 word level. The vocabulary knowledge of Huang’s (2000, 2004) college-junior subjects were slightly better, with most of them reaching the 2,000 word level and some the 3,000 word level.. It was nevertheless suggested by Yang (2006) that a. vocabulary size of 8,000—10,000 words is desirable for college students in Taiwan if they want to become competent and successful English readers. There were fewer studies focusing on investigating the vocabulary size of senior high school students. Huang’s (2000) study, with 254 third-year senior high school students, showed an average vocabulary size ranged between 1,000 to 2,000 word families; in his study in 2003, with 347 subjects also in the third year of senior high school, he found a similar result of vocabulary size between 1,000 to 2,000 word families.. Chao (2003) found. a slightly higher estimate, with the third-year senior high students having an average vocabulary size between 2,000 words level to 3,000 words level. The review of 21.

(32) selected literature and studies suggest that EFL students in Taiwan commonly lack sufficient vocabulary size to support their success in comprehending reading texts.. Word Frequency A word’s frequency means the degree of probability of encountering that word in speech or print. There is a strong relationship between the frequency with which a word occurs and the likelihood it will be learned. The higher frequency of encountering the word, the more familiar it becomes.. There is no doubt that. increased exposure to the words may result both in increased proficiency and in increased familiarity with the words.. Therefore, word frequency is considered a. crucial variable in text comprehension and is also known to have effects on word learning and word familiarity (Aitchison, 1987). Word frequency in general was a poorly defined variable until the advent of corpora such as the Brown corpus.. In the. past, there were only limited vocabulary lists available such as the General Service List (West, 1953) and the University Word List (UWL) (Xue & Nation, 1984, later remodeled to be the Academic Word List, AWL, Coxhead, 2000). The classic list of high frequency words is Michael West’s General Service List (1953).. It has been proven by many studies that GSL provided coverage of 78% to. 92% of various kinds of written text, averaging around 82% coverage.. Beyond the. 2,000 high frequency words of the GSL, Nation proposed that second language learners with special academic purpose, such as study in upper high school or at university, need to know the 836 words of the University Word List (UWL) (Nation, 1990).. The UWL, later remodeled to the 570-word Academic Word List (AWL). (Coxhead, 2000), provides high coverage of academic text, as shown on Table 5 below. 22.

(33) Table 5 Text coverage by the GSL and the AWL in four different kinds of text Researchers Conversation. Fiction. Newspapers. Academic text. 1st 1,000. 84.3%. 82.3%. 75.6%. 73.5%. 2nd 1,000. 6%. 5.1%. 4.7%. 4.6%. Academic. 1.9%. 1.7%. 3.9%. 8.5%. Total. 92.2%. 88.8%. 84.2%. 86.6%. Nation, 2001, p.17 As shown, it was estimated that the first 1,000 high-frequency word families of English roughly made up 73.5 per cent of academic text, the second 1,000 word families made up 4.6 per cent, the 570 word families of the Academic Word List (AWL) made up roughly 8.5 per cent and the remaining 13.4 per cent of text was made up of low-frequency words. ‘Low frequency’ used to be defined as all those words not on the 2,000-word list of GSL, nor on the Academic Word List. There are several frequency levels missing between high-frequency and low-frequency vocabulary.. The General Service List. plus the Academic Word List contains only three frequency levels and therefore they fail to discriminate among language learners whose vocabulary sizes differ beyond these three levels.. Although low-frequency words are by definition with less text. coverage, they are still important to know; readers able to deal with some of the low-frequency words can enjoy the pleasure of reading better. With the high text coverage of high-frequency words and low coverage of low-frequency words, it is clear that high-frequency words need to be known first for minimal comprehension of text. Therefore, learners have to choose the vocabulary they have to learn, starting from the most frequently used words. The 2,000 word families in General Service List (GSL) are the most commonly recognized words to. 23.

(34) be learned with top priority (Schonell, Meddleton, & Shaw, 1956, cited in Nation and Waring, 1997) and are considered by Fox (1987) to be perfectly adequate productive vocabulary for second language learners. It is assumed that both native- and nonnative-speakers acquire vocabulary largely in the order of its frequency. High-frequency words are generally learned before lower-frequency words.. Such an. order of vocabulary acquisition is associated with the link between vocabulary size and word frequency: the bigger the vocabulary, the more low-frequency words one knows.. BNC Word Family List Even though the General Service Word List (West, 1953) has been widely used in past studies on vocabulary and it remains one of the best available word lists, it is in need of replacement because of its age, errors it contains and its written focus (Nation and Waring, 1997). Since modern electronic corpora have become much larger and more sophisticated, reliable data on word frequency can be acquired to develop some specialized vocabulary frequency lists. These vocabulary lists then can serve as research instruments to make certain kinds of texts analysis more accessible. Leech (1992) claimed that the corpus is a powerful methodology, as it is evidence of actual use of the language and open to objective verification of results. To provide more accurate estimates of the number of word families needed to read and listen to English, Nation recently developed a substantial number of word-family lists from the British National Corpus (BNC). The BNC is a 100 million-token corpus consisting of 90% written text and 10% spoken text. Nation first used the BNC to determine the frequency levels and then applied them to other corpora to make these frequency levels truly represent the vocabulary size of a typical English user.. Therefore, these. fourteen 1,000 word-family lists made from the British National Corpus provide a 24.

(35) more objective frequency than other traditional word lists. Following are the rules how Nation made the corpus-based list: Most proper nouns, such as people’s names, the names of cities, and the names of mountain ranges, are not included in the lists. There is however a separate list of proper nouns (the 15th). Compound words are included as headwords, even when they are transparent, for example, airbase, alehouse, breastfeed. A few very common abbreviations are included in the lists as family members of the headword. The criteria used to make word families are based on Bauer and Nation’s (1993) level 6, which includes all the affixes from levels 2 to 6, as listed out below. Level 1: A different form is a different word. Capitalization is ignored. Level 2: Regularly inflected words are part of the same family.. The inflectional. categories are — plural; third person singular present tense; past tense; past participle; -ing; comparative; superlative; possessive. Level 3: -able, -er, -ish, -less, -ly, -ness, -th, -y, non-, un-, all with restricted uses. Level 4: -al, -ation, -ess, -ful, -ism, -ist, -ity, -ize, -ment, -ous, in-, all with restricted uses. Level 5: -age (leakage), -al (arrival), -ally (idiotically), -an (American), -ance (clearance), -ant (consultant), -ary (revolutionary), -atory (confirmatory), -dom (kingdom; officialdom), -eer (black marketeer), -en (wooden), -en (widen), -ence (emergence), -ent (absorbent), -ery (bakery; trickery), -ese (Japanese; officialese), -esque (picturesque), -ette (usherette; roomette), -hood (childhood), -i (Israeli), -ian (phonetician; Johnsonian), -ite (Paisleyite; also chemical meaning), -let (coverlet), -ling (duckling), -ly (leisurely), -most (topmost), -ory (contradictory), -ship (studentship), -ward (homeward), -ways (crossways), -wise (endwise; discussion-wise), anti- (anti-inflation), ante- (anteroom), arch- (archbishop), bi- (biplane), 25.

(36) circum- (circumnavigate), counter- (counter-attack), en- (encage; enslave), ex- (ex-president), fore- (forename), hyper- (hyperactive), inter- (inter-African, interweave), mid- (mid-week), mis- (misfit), neo(neo-colonialism),. post-. (semi-automatic), sub-. (post-date),. pro-. (pro-British),. (subclassify; subterranean), un-. semi(untie;. unburden). Level 6: -able, -ee, -ic, -ify, -ion, -ist, -ition, -ive, -th, -y, pre-, re-. As shown, the scheme for defining word families at Level 6 includes inflections and over 80 derivational affixes. Such a large number of affixes result in some large word families, especially among the high-frequency words. However, level 6 only includes bases which are free forms. For example, romance and romantic could not be members of the same family because their base is a bound form, not a free form (Nation, 2007).. Since all the family members at level six meet the criteria of. regularity, frequency, productivity, and predictability, Nation saw Level 6 as an acceptable level for defining a word family. Besides, Nation decided to go with large word families, because he assumed that when reading and listening, a learner who knew at least one of the members of a family well could understand other family members by using knowledge of the most common and regular of the English word building devices (Nation, 2007).. Comparison between the BNC Word Family List and the English Reference Word List As reviewed earlier, the English Reference Word List developed by a team led by Professor Jeng, Hengsyung, is regarded as English reference word list for the two English tests in the college entrance examination: SAET and DRET.. The superiority. of the 6,480 English Reference Word List over other word lists upon SAET reading 26.

(37) test is expected to occur.. However, as discussed earlier, the coverage of the SAET. reading texts by all the six levels of the English Reference Word List was only 86 per cent.. As the 6,480 English Reference Word List is greatly different from the BNC. Word Family List in the way that the former is a lemma list while the latter is a word family list, we speculate that the low text coverage percentage of the English Reference Word List may be caused by the way of word count.. Because lemmas. include a narrower range and a smaller number of words than word families, the lower text coverage by the English Reference Word List may result from conservative lexical items and underestimate of word count in each level of the 6,480-word List. In order to explore above speculation, it is of value to take a further substantial comparison between the English Reference Word List and the BNC Word Family List. The 6,480 English Reference Word List was run level by level first against the BNC Word Family List and then, all the six levels together as well as all the first four levels. The reason to run several levels together is to avoid overlapping-counts of the same word-family members, since some words belonging to the same word family are classified into various levels in the English Reference Word List.. For example, the. word “collect” is ranked in the second level of the English Reference Word List, “collection” in the third level and “collective” and “collector” both in the sixth level. The results of the comparison between the two word lists suggest that the variance in text coverage by the two lists cannot be exclusively explained by the practice of counting words. As shown on Table 6, if we apply different ways of word count to the 6,480 words in the English Reference Word List, the number of types, which means the total number of the different word forms in the list, is 6,458.. In. addition, if the 6,480 words are counted in word family, which is known as a base word accompanied with its inflectional forms and derived forms, the number of word families reaches 5,164. Likewise, for the 4,320 words of the first four levels in the 27.

(38) English Reference Word List, the number of types converts into 4,307 and the number of word families is 3,465, as shown on Table 7.. Both the figures for the number of. types: 6,458 and 4,307 are almost the same as the marked “6,480” and “4,320” words for the total six levels and the first four levels in the List.. However, the numbers of. the word families: 5,164 and 3,465 for the two ranges imply that the English Reference Word List contains in fact a very diverse range of different words. Moreover, words in the English Reference Word List disperse widely against the BNC Word Family List. With respect to the first four levels of the English Reference Word List, which are supposed to serve as sampling reference for SAET, the results show that 32% of the 4,320 words in the English Reference Word List are ranged within the first 1,000 level of the BNC Word Family List, followed by 25% in the second level, 19% in the third, 11% in the forth, and 7% in the fifth.. Accumulatively,. of all the 4,320 words in the first four levels of the CEEC’s English Reference Word List, 57 percent allocate within the 2,000 words level of the BNC Word Family List, around 75%, 85%, and 90% distributing within the 3,000 words level, 4,000 and 5,000 words levels.. If the first four levels of the English Reference Word List work. as sampling reference for the SAET, a large number of the 4,320 words (more than 40%) are out of the most frequently used 2,000 word families, and 25% and 15% are out of the range of the most frequent 3,000 and 4,000 word families of the BNC Word Family List. As for the analysis on all the six levels together, 23% of the 6,480 words in the CEEC’s English Reference Word List allocate in the first 1,000 word families of the BNC Word Family List, followed by 20% in the second frequent level, 16% in the third, 13% in the forth, and 9% in the fifth.. It is accumulated that only about 70% of. the 6,480 words allocate within the 4,000 words level, 80% within the 5,000 words level, and 90% within the 7,000 words level of the BNC Word Family List. Therefore, 28.

(39) our previous speculation regarding the conservative practice of counting words by the English Reference Word List seems inefficient. Instead, the reason for the lower SAET text coverage by the English Reference Word List may lie in the nature of the List itself.. The English Reference Word List, transforming into 5,164 word families. and distributing widely against the BNC Word Family List, seems to cover too large a number of words for Taiwanese senior high school students to refer to. Table 6 Distributions of the 6,480 words against the BNC Word Family List Word Count Number of Count Coverage % Cumulative % Levels. type. family. type. family. type. family. K1 Words :. 1,508. 959. 23.35. 18.57. 23.35. 18.57. K2 Words :. 1,270. 947. 19.67. 18.34. 43.02. 36.91. K3 Words :. 1,028. 880. 15.92. 17.04. 58.93. 53.95. K4 Words :. 819. 723. 12.68. 14.00. 71.62. 67.95. K5 Words :. 596. 547. 9.23. 10.59. 80.85. 78.54. K6 Words :. 380. 365. 5.88. 7.07. 86.73. 85.61. K7 Words :. 253. 243. 3.92. 4.71. 90.65. 90.32. K8 Words :. 156. 151. 2.42. 2.92. 93.06. 93.24. K9 Words :. 120. 118. 1.86. 2.29. 94.92. 95.53. K10 Words :. 90. 87. 1.39. 1.68. 96.31. 97.21. K11 Words :. 46. 46. 0.71. 0.89. 97.03. 98.10. K12 Words :. 25. 24. 0.39. 0.46. 97.41. 98.57. K13 Words :. 25. 25. 0.39. 0.48. 97.80. 99.05. K14 Words :. 8. 8. 0.12. 0.15. 97.93. 99.21. K15 Words :. 13. 13. 0.20. 0.25. 98.13. 99.46. K16 Words :. 10. 10. 0.15. 0.19. 98.28. 99.65. K17 Words :. 6. 6. 0.09. 0.12. 98.37. 99.77. K18 Words :. 5. 5. 0.08. 0.10. 98.45. 99.86. K19 Words :. 7. 6. 0.11. 0.12. 98.56. 99.98. K20 Words :. 1. 1. 0.02. 0.02. 98.58. 100.00. Off-list. 92. 1.42. 0.00. 100.00. 100.00. Total. 6,458. 100.00. 100.00. 100.00. 100.00. 5,164. 29.

(40) Table 7. Distributions of the 4,320 words of Level 1 to Level 4 of English Reference Word List against the BNC Word Family List. Word Count. Number of Count. Coverage %. Cumulative %. type. family. type. family. type. family. K1 Words :. 1,396. 948. 32.41. 27.36. 32.41. 27.36. K2 Words :. 1,084. 861. 25.17. 24.85. 57.58. 52.21. K3 Words :. 713. 649. 16.55. 18.73. 74.14. 70.94. K4 Words :. 435. 396. 10.10. 11.43. 84.23. 82.37. K5 Words :. 263. 251. 6.11. 7.24. 90.34. 89.61. K6 Words :. 141. 137. 3.27. 3.95. 93.62. 93.56. K7 Words :. 81. 79. 1.88. 2.28. 95.50. 95.84. K8 Words :. 39. 35. 0.91. 1.01. 96.40. 96.85. K9 Words :. 31. 31. 0.72. 0.89. 97.12. 97.75. K10 Words :. 22. 21. 0.51. 0.61. 97.63. 98.35. K11 Words :. 7. 7. 0.16. 0.20. 97.79. 98.56. K12 Words :. 9. 9. 0.21. 0.26. 98.00. 98.82. K13 Words :. 12. 12. 0.28. 0.35. 98.28. 99.16. K14 Words :. 3. 3. 0.07. 0.09. 98.35. 99.25. K15 Words :. 8. 8. 0.19. 0.23. 98.54. 99.48. K16 Words :. 7. 7. 0.16. 0.20. 98.70. 99.68. K17 Words :. 6. 6. 0.14. 0.17. 98.84. 99.86. K18 Words :. 4. 4. 0.09. 0.12. 98.93. 99.97. K19 Words :. 1. 1. 0.02. 0.03. 98.96. 100.00. K20 Words :. 0. 0. 0.00. 0.00. 98.96. 100.00. Off-list. 45. 1.04. 0.00. 100.00. 100.00. Total. 4,307. 100.00. 100.00. 100.00. 100.00. Levels. 3,465. 30.

(41) In addition to the difference in lemma- and word-family basis, the ranking of levels between the English Reference Word List and the BNC Word Family List also varies.. Table 8 shows the way how the English Reference Word List differs from the. BNC Word Family List in arrangement of members of the same word families. As the English Reference Word List is produced by reference to a variety of words lists and are developed into a lemma-based list, words belonging to the same word families are ranked dispersedly into different levels. word-family of “employ”.. The first example in Table 8 is the. In the first level of the BNC Word Family List, this. family consists of six lemmas and seven affix members: employ, employed, employee, employees, employer, employers, employing, employment, employs, employable, unemployed, unemployable, unemployment.. Nevertheless, in the English Reference. Word List only four lemmas are listed out: employ, employee, employer and unemployment, among which the first three words are ranked in the third level while the last word in the sixth level. The lemma of employment is attached to another lemma and the lemma of unemployed is eliminated from the List. example is the word family of “collect”.. The second. In the BNC Word Family List, this word. family contains the following members: collect, collects, collecting, collected, uncollected, collective, collectively, collector, collectors, collection, collections, collectable collectables, collectivist, collectivists, collectivism, collectivity. This family consists of five lemmas: the collect lemma with five members, the collective lemma with one member, the collectively lemma also with one single member, the collector lemma with two members, the collection lemma with eight members.. By. contrast, the English Reference Word List lists out four lemmas: collect, collection, collective and collector. These four lemmas disperse across three different levels. The third example: word family of “compare” includes five lemmas and twelve members in the BNC Word Family List whereas there are only three lemmas of it 31.

(42) ranked in two levels of the English Reference Word List. As for the word family of “consult”, there are five lemmas and five more inflectional-affix members listed in the first level of the BNC Word Family List but there are only three lemmas: consult, consultant and consultation which are ranked in two different levels of the English Reference Word List.. The last example in Table 8 is the word family of “simple”.. Both the lists consist of the same four lemmas: simple, simply, simplicity and simplify. Among the four lemmas, one is ranked in the first level, one in the second level and two in the sixth level in the English Reference Word List.. Other than the four. lemmas, there are seven more inflectional members of the family included in the BNC Word Family List and they are all ranked in the same first level. According to Nation (2006), making words lists either based on word types or lemmas does not mean simply adding or deleting word-family members. Rather, each deleted member and its appropriately related forms would have to appear as a separate word-family ordered by range and frequency that make up the lists. Table 8 Examples of differences between the two lists, Words vs. Families 6,480 English Reference Word List rd. vs.. BNC Word Family List. employ (3 ) (ment 0). employ 6 (1st). employee (3rd) employer (3rd) unemployment (6th). employed 0 employee 1 employees 0 employer 1 employers 0 employing 0 employment 1 employs 0 employable 0 unemployed 1 unemployable 0 unemployment 1. 6,480 words. vs.. BNC. 6,480 words 32. vs.. BNC.

(43) collect (2nd) collection (3rd) collective ( 6th) collector (6th). collect 5 ( 1st ) collects 0 collecting 0 collected 0. compare (2nd) comparable (6th) comparative (6th). compare 5 ( 1st) comparable 1 comparably 0 comparability 0. uncollected 0 collective 1 collectively 1 collector 1. incomparable 0 compared 0 compares 0 comparative 1. collectors 0 collection 1 collections 0 collectable 0. comparison 1 comparisons 0 comparing 0 comparatively 1. collectables 0 collectivist 0 collectivists 0 collectivism 0 collectivity 0 consult (4th). consult 5 (1st). simple ( 1st). simple 4 (1st). consultant (4th) consultation (6th). consultancy 1 consultant 1 consultants 0 consultation 1. simply (2nd) simplicity ( 6th) simplify ( 6th). simpleness 0 simpler 0 simplest 0 simplicity 1. consultations 0 consultative 1 consulted 0 consults 0. simplification 0 simplified 0 simplify 1 simplifies 0. consulting 0. simplifying 0 simply 1. In terms of level divisions, there are considerable variations between the BNC Word Family List and the CEEC’s English Reference Word List. Table 9 presents the differences in the classification of words into various levels between the two lists. For example, uncover, unemployment, unlock, unpack are four individual lemmas ranked in the sixth level in the English Reference Word List.. However, in. the BNC Word Family List “un-” is treated as a prefix attached to the base-word 33.