论文部分内容阅读
【摘 要】完型填空是一種比较传统和广受欢迎的测试方式， C- 测试是对完型填空的修改。尽管C- 测试在当前的英语测试中并为广泛使用，但是许多研究已经证明C- 测试在一定程度上优于完型填空。本篇论文主要总结了一些研究者的成果。
【关键词】完型填空； C- 测试； 测试；信度
Language test paper
Cui Wen-juan
（Ningxia Construction Vocational and Technical College Yinchuan Ningxia 750000）
【Abstract】Close Test is a traditional and popular way of test， while C-Test is a modification of Close Test. Though C-Test do not enjoy great popularity in the current English testing， but many studies have proved that in some degrees C- Test is a better way than Close Test. This paper generates some achievements and influential words of some researchers.
【Key words】Close Test；C-Test；test； reliability
Close Test is a very popular test method in many English tests， such as CET4， CET6， and TEM4. Close Test is a test in which words are deleted mechanically， regardless of the function of the word. It has many merits such as it can test a student's overall ability， it is reliable and easy to construct. It also has many unavoidable disadvantages， for example： 1.Different passages give different results and the deletion of different sets of words result in different results. 2.it is not possible to predict accurately from this is people's ability with respect to the variety separate skills， such as speaking， writing， etc. C-Test is a modified form of Close Test， in which the second half of some words are deleted. Many people assume it is better than the traditional Close Test. For example， 1， C-Test can generate exact scoring， because in one space， there is only one answer.2， A C-Test of 100 items takes little space and not really so much time to complete. In the paper， the first chapter is devoted to illustrate Close Test， and in the second chapter， a detailed study of C-Test is given. In the last part， a conclusion is reached： C-Test is a better method than Close Test in several respects.
1. Chapter I Close test： a general view
1.1 About Close test.
The close procedure was originated in the 1950s as a means of assessing the difficulty of a reading text for native speakers， but within three years its originator suggested that it could be used for assessing the progress of second and foreign language learners. Ulrich Raatz regarded cloze tests as the most important representative of this family of foreign language tests. To construct a cloze test of the classical type a fairly long text is needed. Since it merely functions as a sample of the language the text should be authentic， but its content is not important， although it should not be biased in any way. After the initial run-in sentence every nth - very often every 7th - word is deleted. This is a simplification of a random sampling procedure for word deletion. It is recommended that a test should have at least 50 blanks in order to ensure adequate sampling. When the test is processed two sampling processes are operating simultaneously： the text （and its redundancy） is being sampled by the blanks； the examinee's competence is being sampled by restoring the missing elements. Cloze tests were repeatedly reported to have high reliabilities and to correlate highly with other language tests （validity）. However the majority of these studies were performed in the USA on highly heterogeneous groups of learners of English as a foreign language. （Ulrich Raatz， Introduction to language testing and C-Tests） J.Charles Alderson， etc defines close test as the test in which words are deleted mechanically. Each word is deleted regardless of what the function of the word is.（J. Charles Alderson， etc. 1995， p 55）
In his A language testing handbook， Andrew Hawison talked about the principle of close test， "The principle is that single words are taken 'fillers'， one only for each gap.
In his testing for language teacher， Arthur Hughes explained close test as this： in its original form， the close procedure involves deleting a number of words in a passage， leaving blanks， and requiring the person taking the test to attempt to replace the original words. After a short unmutilated 'lead-in'， it is usually about every seventh word which is deleted.
According to Andrew Hawison， there are two types of gap： function gaps， such as conjunctions， prepositions， articles， etc， and for this kind of gap， there is only one correct answer. For example， "I visit him ____ the morning..." in the gap， only "in" can be fitted here. Another type is semantic gap， such as adjectives， verbs， adverbs， nouns， and for one gap， there are several alternatives， as long as they are semantically related. For instance， "there is a _____ lady in the garden..." in the gap， "beautiful"， "pretty"， "attractive" ，etc can be used in the gap.
Here is an example of close test：
He has been called the "missing link." Half-man， half-beast. He is supposed to live in the highest mountain in the world-Mount Everest.
He is known as the Abominable Snowman. The ___1__ of the Snowman has been around for ___2 __. Climbers in the 1920s reported finding marks like those of human feet high up on the side of Mound Everest. The native people said they___3___this creature and called it the "Yeti，" and they said that they had___5__caught Yetis on two occasions ___4 __none has ever been produced an evidence（證据）。
Over the years， the story of the Yetis has___6___.In 1951， Eric Shipton took photographs of a set of tracks in the snow of Everest. Shipton believed that they were not ____7___the tracks of a monkey or bear and ____8___that the Abominable Snowman might really____9___.
Further efforts have been made to find out about Yetis. But the only things people have ever found were ___10_ _footprints. Most believe the footprints are nothing more than ___11___animal tracks， which had been made ___12___as the melted（融化）and refroze in the snow. ____13___，in 1964， a Russian scientist said that the Abominable Snowman was ____14___and was a remaining link with the prehistoric humans. But， _15 __，no evidence has ever ___16___been produced. These days， only a few people continue to take the story of the Abominable Snowman ____17____. But if they ever ____18____catching one， they may face a real _____19____： Would they put it in a___ 20____or give it a room in a hotel？
（ ）1. A. event B. story C. adventure D. description
（ ） 2. A. centuries B. too long C. some time D. many years
（ ） 3. A. heard from B. cared for C. knew of D. read about
（ ） 4. A. even B. hardly C. certainly D. probably
（ ） 5. A. as B. though C. when D. until
（ ） 6. A. developed B. changed C. occurred D. continued
（ ）7. A. entirely B. naturally C. clearly D.simply
（ ）8. A. found B. declared C. felt D. doubled
（ ）9. A. exist B. escape C. disappear D. return
（ ） 10.A. clearer B. more C. possible D. rare
（ ） 11.A.huge B. recent C. ordinary D. frightening
（ ）12.A.strange B.large C.deep D. rough
（ ）13.A.In the end B. Therefore C. After all D. However
（ ）14.A.imagined B. real C. special D. familiar
（ ）15.A.so B. besides C. again D. instead
（ ）16.A.rightiy B. actually C. normally D. particularly
（ ）17.A.lightly B. jokingly C. seriously D. properly
（ ）18.A.succeed in B. insist on C. depend on D. join in
（ ）19.A.decision B. Situation C. subject D.problem
（ ） 20.A.zoo B.mountain C.museum D.laboratory
1.2 Disadvantages of Close test.
Close test bring many convenience and advantages to language testers and students， but nothing is perfect， so as to close test. The following are several linguists' view of the disadvantages of close test.
According to Arthur Hughes， there are two problems facing close test.
1， From close test， some underlying ability could be measured through the measure， but it is not possible to predict accurately from this is people's ability with respect to the variety separate skills， such as speaking， writing， etc.
2， Different passages give different results and the deletion of different sets of words result in different results. （Arthur Hughes， Testing for Language Teachers）
J. Charles Alderson， Caroline Clapham and Dianne Wall pointed out several problems of close test：
1， One problem with nth word deletion tasks is that the choice of the first deletion can have an effect on the validity of the test to experiments comparing cased on the sane text， but with different initial gaps， and therefore different spaces through the passage， have shown that the tests vary in both validity and reliability （Alderson， 1979， and Klein-Braley， 1981）. 2， another disadvantage is that an nth word deletion close test is not really amended. If the tester decides to reinstate the difficult word and delete another one nearby， then the principle of nth word deletion comes less authentic.
3， marking close tests can be difficult since there may be many possible answers for any one gap， and there is often disagreement as to what answers are acceptable. It's time-consuming to produce a comprehensive answer which may require wide protesting of the test and then lengthy discussion on the appropriateness of different answers. To avoid this， some testers only accept the exact word that was used in the original text. This naturally leads to lower final scores but dose not usually change the students' ranks.
4， Unless the aim of the close test is to test overall language proficiency， as advocated by Dller 1979， such tests may be a wasteful way of testing. Few of the items in any one passage may test the aspects of language with which the tester is concerned..
In 1979 and 1983 Alderson and in 1981 Klein-Braley reported studies using more homogeneous samples of examinees and came to the conclusion that cloze tests， while theoretically sound， suffered from a number of major technical defects： In order to ensure a sufficient number of items， cloze tests have to be relatively long. Because of the deletion principle， cloze tests usually consist of only one longer text. This can result in text specificity and thus in test bias. The factors 'text'， 'deletion rate' and 'starting point' affect reliability and validity coefficients. If the exact method of scoring is used， then cloze tests are often too difficult for adult educated native speakers. If the acceptable method of scoring is used， a large subjective component enters the scoring， and the tests are much less reliable. Moreover scoring the tests consumes much more time. The difficulty of cloze tests depends on the proportion of structure and content words deleted.
Ulrich Raatz also put forward his point of view： Many of the cloze tests reported in the literature are less reliable than originally assumed. （Ulrich Raatz， Introduction to language testing and C-Tests）
2. Chapter II An elaboration of C- Test
2.1 About C-Test.
A C-Test is an integrative written test of general language proficiency based on the concept of reduced redundancy. A C-Test consists of five to six short authentic texts， each complete as a sense unit in itself. In these texts the first sentence is left standing. Then the 'rule of two' is applied： beginning at word two in sentence two the second half of every second word is deleted. Numbers and proper names are usually left undamaged， but otherwise the deletion is entirely mechanical. The process is continued until the required number of blanks has been produced （in the canonical C-Test either 20 or 25）. Then the text is allowed to run on to its natural conclusion. The instructions for the examinees say something like 'In this test parts of some of the words have been damaged. Please replace the missing parts.' Texts are arranged in order of difficulty with the easiest text first. （Ulrich Raatz， Introduction to language testing and C-Tests） east 100 items. The deletion rate and the starting point for deletions should be fixed.
The words affected by the deletions should be a genuinely representative sample of the elements of the text.
Examinees with special knowledge should not be favoured by specific texts， therefore the new test ought to consist of a number of different texts.
Only exact scoring should be possible so as to ensure objectivity.
Native speakers ought to be able to make virtually perfect scores on the test： 90'% or higher. If native speakers cannot make scores higher than 90%， then the text should not be used for non-native speakers.
The new test should be reliable， valid and easy to develop.The first results presented in 1982 showed that all these criteria were fulfilled. Subsequent research （cf. Grotjahn， Klein-Braley， Raatz， this volume） has confirmed these early findings. （Ulrich Raatz， Introduction to language testing and C-Tests）
The C-test is really a variety of close， which its originators claim is superior to the kind of close test. Instead of whole words， it is the second half of every word which is deleted. （Arthur Hughes， Testing for Language Teachers）
In 1981 Raatz and Klein-Braley presented C-Tests as a technical improvement over cloze tests. They had set up a number of conditions which the new test format should fulfill：
" The new test should be much shorter， but at the same time it should have at least 100 items.
" The deletion rate and the starting point for deletions should be fixed.
" The words affected by the deletions should be a genuinely representative sample of the elements of the text.
" Examinees with special knowledge should not be favoured by specific texts， therefore the new test ought to consist of a number of different texts.
" Only exact scoring should be possible so as to ensure objectivity.
" Native speakers ought to be able to make virtually perfect scores on the test： 90'% or higher. If native speakers cannot make scores higher than 90%， then the text should not be used for non-native speakers.
" The new test should be reliable， valid and easy to develop. The first results presented in 1982 showed that all these criteria were fulfilled. Subsequent research （cf. Grotjahn， Klein-Braley， Raatz， this volume） has confirmed these early findings.
Lucy Katona and Zoltan Dornye （The C-Test： A Teacher Friendly Way to Test Language Proficiency） pointed out that The C-test is an integrative testing instrument that measures overall language competence， very much like the cloze test. It consists of four to six short， preferably authentic， texts in the target language， to which "the rule of two" has been applied： the second half of every second word has been deleted， beginning with the second word of the second sentence； the first and last sentences are left intact. If a word has an odd number of letters， the "bigger" part is omitted， e.g.， proud becomes pr-. One-letter words， such as I， are ignored in the counting. The students' task is to restore the missing parts. In a typical C-test there are 100 gaps-that is， missing parts. Only entirely correct restorations are accepted. By way of illustration， here is a short C-test： "One cool autumn evening， Bob L.， a young professional， returned home from a trip to the supermarket to find his computer gone. Gone！ All so- of cr- thoughts ra- through h- mind： H- it be- stolen？ H- it be- kidnapped？ H- searched h- house f- a cl- until h- noticed a sm- piece o- printout pa- stuck un- a maga- on h- refrigerator do-. His he- sank a- he re- this sim- message： can't continue， file closed， bye."
C-Tests belong to the family of language tests using the reduced redundancy approach towards test development. They are intended as a technical improvement on cloze tests. C-Tests consist of 4 to 6 authentic texts in the language required， should be selected from different sources and must be appropriate in content and difficulty level for the target group envisaged. C-Tests have been used in a number of languages， .e.g. English， French， German， Italian， Russian， Spanish and Portuguese. For English and German texts readability type prediction formulas for estimating difficulty before test use are available.
Here are two samples of C-Test：
1 Haywood football team needs new players
You li_ _ football a_ _ you do_ _ have a te_ _ to pl_ _ with？ Th_ _ come t_ our ne_ _ training o_ Wednesday after_ _ _ _ in t_ _ new spo___ hall a_ Haywood. Every ____ is wel_ _ _ _ _ here， ju__ bring yo_ _ football sh_ _ _ along a_ _ have f_ _. Our n_ _ trainer i_ perfect， la_ _ year w_ even be_ _ the team of Nottingham High School. We're waiting for you！
2. Fill in the missing words
Before the Dixons _____ to Seahouses， Mrs Dixon needed s _ _ _ things f _ __ _ the shop. She t _ _ _ Sarah to bring three t _ _ _ of cat f _ _ _， four s _ _ _ _ bottles _ _ water and a ___ of muesli. Sarah didn't！_____ to her mother. She b _____ thew____ things and f _ _ _ _ _ to bring the muesli. Mrs Dixon didn't want _ _ _ chocolate， but Sarah bought five _ _ _ _ of chocolate.
_ _ _ _ she came home， Sarah's mother was a _ _ _ _. She shouted： "You n _ _ _ _ listen to what I tell you." Sarah said： "Sorry， mum， but I'm soe _______ about our t __ _ _ to Seahouses." Then Dad _ _ _ _ in and said： "Don't w _ _ _ _ ！ We can get all th _ _ _ things in Seahouses， t _ erits of
2.2 Merits of C-Test.
C-Tests have proved to be highly reliable and to have high empirical validity in terms of correlations with other language proficiency tests. Their construct validity has been the subject of considerable research which is documented in the C-Test bibliography. In The C-Test： A Teacher Friendly Way to Test Language Proficiency， Lucy Katona and Zoltan Dornye came to such a conclusion：
1. The C-test appears to correlate well with our other language-proficiency tests， which proves that it is a good and reliable testing instrument. It is also an integrative language-testing method， correlating highly with both vocabulary and grammar scores， as well as speaking skills （圖1）.
图1
2. The C-test measures general language proficiency more accurately than the cloze， and it is a lot easier to construct and to score （图2）.
图2
3. We recommend the exact scoring method （that is， accepting only entirely correct restorations）， as we found that accepting words with spelling errors made no significant difference to the student rankings.
2.3 Demerits of C-Test.
Disadvantages of C-Test are as follows： C-Test is harder to read than a close passage， and correct responses can often be found in the surrounding text. （Arthur Hughes， Testing for Language Teachers）
Alderson（1995） pointed two problems facing C-Test： 1， The first few letters of a missing word are given in a C-Test text will reduce the number of possible answers for any one gap.2， it does not provide enough clues for educated native speakers to be able to draw._."
Chapter III C-Test： a better method over Close Test.
According to Lucy Katona and Zoltan Dornye， the C-test was developed as a modification of the cloze test， which is a frequently used， major language-testing instrument， extremely popular because of the ease of constructing it and its high reliability and validity. （The cloze test consists of a longer text of which every fifth to tenth complete word is left out.）
In the paper， The Comparative Study of the Validation between the Classical Cloze Test and C-Test， through research and questionnaire， Zhang Ling & Han Hui came to such a conclusion： 1， close format is superior in terms of its constructing and scoring method and reliability. 2， as pragmatic test format， which close format has more content coverage to measure a candidate's abilities， including linguistic and non-linguistic aspects. 3， which close test could provide more information to predict a candidate's actual performance in language learning. The study shows that C-Test could be adopted in vocabulary practice， while classical test could be tested in pragmatic language test.
The C-test appears to have some advantages over the cloze test： 1. As students are confronted with a variety of short passages， a better sampling of content areas is possible. Also， a person with special knowledge in a certain field does not have an unfair advantage.
2. By "damaging" every second word， we can obtain a more representative sample of all the different language elements in the text than in the cloze， where normally every fifth or sixth word is left out.
3. Many more items can be included in much shorter texts， making the test less time-consuming for the students than the cloze.
4. Unlike the cloze test， scoring is easy and objective， as there is only one acceptable solution in most cases.
5. As a rule， students actually like doing C-tests， whereas the cloze test is one of the most frustrating test types for learners.
6. A C-Test of 100 items takes little space and not really so much time to complete.
References：
Alderson， J. Charles， etc.1995 .Language Test Construction and Evaluation. Cambridge University Press
Hughes Arthur， 1941. Testing for Language Teachers. Cambridge University Press
Harrison Andrew. 1983， A language Testing Handbook. Macmillan Press， London
Raatz Ulrich，Klein-Braley Christine，etc. Introduction to language testing and C-Tests
參考文献
[1] 张玲，韩慧，2005，传统完形填空与C-试题的效度对比研究， 新疆教育学院学报， 2005年9月第21卷第3期
收稿日期：2012-12-12
【关键词】完型填空； C- 测试； 测试；信度
Language test paper
Cui Wen-juan
（Ningxia Construction Vocational and Technical College Yinchuan Ningxia 750000）
【Abstract】Close Test is a traditional and popular way of test， while C-Test is a modification of Close Test. Though C-Test do not enjoy great popularity in the current English testing， but many studies have proved that in some degrees C- Test is a better way than Close Test. This paper generates some achievements and influential words of some researchers.
【Key words】Close Test；C-Test；test； reliability
Close Test is a very popular test method in many English tests， such as CET4， CET6， and TEM4. Close Test is a test in which words are deleted mechanically， regardless of the function of the word. It has many merits such as it can test a student's overall ability， it is reliable and easy to construct. It also has many unavoidable disadvantages， for example： 1.Different passages give different results and the deletion of different sets of words result in different results. 2.it is not possible to predict accurately from this is people's ability with respect to the variety separate skills， such as speaking， writing， etc. C-Test is a modified form of Close Test， in which the second half of some words are deleted. Many people assume it is better than the traditional Close Test. For example， 1， C-Test can generate exact scoring， because in one space， there is only one answer.2， A C-Test of 100 items takes little space and not really so much time to complete. In the paper， the first chapter is devoted to illustrate Close Test， and in the second chapter， a detailed study of C-Test is given. In the last part， a conclusion is reached： C-Test is a better method than Close Test in several respects.
1. Chapter I Close test： a general view
1.1 About Close test.
The close procedure was originated in the 1950s as a means of assessing the difficulty of a reading text for native speakers， but within three years its originator suggested that it could be used for assessing the progress of second and foreign language learners. Ulrich Raatz regarded cloze tests as the most important representative of this family of foreign language tests. To construct a cloze test of the classical type a fairly long text is needed. Since it merely functions as a sample of the language the text should be authentic， but its content is not important， although it should not be biased in any way. After the initial run-in sentence every nth - very often every 7th - word is deleted. This is a simplification of a random sampling procedure for word deletion. It is recommended that a test should have at least 50 blanks in order to ensure adequate sampling. When the test is processed two sampling processes are operating simultaneously： the text （and its redundancy） is being sampled by the blanks； the examinee's competence is being sampled by restoring the missing elements. Cloze tests were repeatedly reported to have high reliabilities and to correlate highly with other language tests （validity）. However the majority of these studies were performed in the USA on highly heterogeneous groups of learners of English as a foreign language. （Ulrich Raatz， Introduction to language testing and C-Tests） J.Charles Alderson， etc defines close test as the test in which words are deleted mechanically. Each word is deleted regardless of what the function of the word is.（J. Charles Alderson， etc. 1995， p 55）
In his A language testing handbook， Andrew Hawison talked about the principle of close test， "The principle is that single words are taken 'fillers'， one only for each gap.
In his testing for language teacher， Arthur Hughes explained close test as this： in its original form， the close procedure involves deleting a number of words in a passage， leaving blanks， and requiring the person taking the test to attempt to replace the original words. After a short unmutilated 'lead-in'， it is usually about every seventh word which is deleted.
According to Andrew Hawison， there are two types of gap： function gaps， such as conjunctions， prepositions， articles， etc， and for this kind of gap， there is only one correct answer. For example， "I visit him ____ the morning..." in the gap， only "in" can be fitted here. Another type is semantic gap， such as adjectives， verbs， adverbs， nouns， and for one gap， there are several alternatives， as long as they are semantically related. For instance， "there is a _____ lady in the garden..." in the gap， "beautiful"， "pretty"， "attractive" ，etc can be used in the gap.
Here is an example of close test：
He has been called the "missing link." Half-man， half-beast. He is supposed to live in the highest mountain in the world-Mount Everest.
He is known as the Abominable Snowman. The ___1__ of the Snowman has been around for ___2 __. Climbers in the 1920s reported finding marks like those of human feet high up on the side of Mound Everest. The native people said they___3___this creature and called it the "Yeti，" and they said that they had___5__caught Yetis on two occasions ___4 __none has ever been produced an evidence（證据）。
Over the years， the story of the Yetis has___6___.In 1951， Eric Shipton took photographs of a set of tracks in the snow of Everest. Shipton believed that they were not ____7___the tracks of a monkey or bear and ____8___that the Abominable Snowman might really____9___.
Further efforts have been made to find out about Yetis. But the only things people have ever found were ___10_ _footprints. Most believe the footprints are nothing more than ___11___animal tracks， which had been made ___12___as the melted（融化）and refroze in the snow. ____13___，in 1964， a Russian scientist said that the Abominable Snowman was ____14___and was a remaining link with the prehistoric humans. But， _15 __，no evidence has ever ___16___been produced. These days， only a few people continue to take the story of the Abominable Snowman ____17____. But if they ever ____18____catching one， they may face a real _____19____： Would they put it in a___ 20____or give it a room in a hotel？
（ ）1. A. event B. story C. adventure D. description
（ ） 2. A. centuries B. too long C. some time D. many years
（ ） 3. A. heard from B. cared for C. knew of D. read about
（ ） 4. A. even B. hardly C. certainly D. probably
（ ） 5. A. as B. though C. when D. until
（ ） 6. A. developed B. changed C. occurred D. continued
（ ）7. A. entirely B. naturally C. clearly D.simply
（ ）8. A. found B. declared C. felt D. doubled
（ ）9. A. exist B. escape C. disappear D. return
（ ） 10.A. clearer B. more C. possible D. rare
（ ） 11.A.huge B. recent C. ordinary D. frightening
（ ）12.A.strange B.large C.deep D. rough
（ ）13.A.In the end B. Therefore C. After all D. However
（ ）14.A.imagined B. real C. special D. familiar
（ ）15.A.so B. besides C. again D. instead
（ ）16.A.rightiy B. actually C. normally D. particularly
（ ）17.A.lightly B. jokingly C. seriously D. properly
（ ）18.A.succeed in B. insist on C. depend on D. join in
（ ）19.A.decision B. Situation C. subject D.problem
（ ） 20.A.zoo B.mountain C.museum D.laboratory
1.2 Disadvantages of Close test.
Close test bring many convenience and advantages to language testers and students， but nothing is perfect， so as to close test. The following are several linguists' view of the disadvantages of close test.
According to Arthur Hughes， there are two problems facing close test.
1， From close test， some underlying ability could be measured through the measure， but it is not possible to predict accurately from this is people's ability with respect to the variety separate skills， such as speaking， writing， etc.
2， Different passages give different results and the deletion of different sets of words result in different results. （Arthur Hughes， Testing for Language Teachers）
J. Charles Alderson， Caroline Clapham and Dianne Wall pointed out several problems of close test：
1， One problem with nth word deletion tasks is that the choice of the first deletion can have an effect on the validity of the test to experiments comparing cased on the sane text， but with different initial gaps， and therefore different spaces through the passage， have shown that the tests vary in both validity and reliability （Alderson， 1979， and Klein-Braley， 1981）. 2， another disadvantage is that an nth word deletion close test is not really amended. If the tester decides to reinstate the difficult word and delete another one nearby， then the principle of nth word deletion comes less authentic.
3， marking close tests can be difficult since there may be many possible answers for any one gap， and there is often disagreement as to what answers are acceptable. It's time-consuming to produce a comprehensive answer which may require wide protesting of the test and then lengthy discussion on the appropriateness of different answers. To avoid this， some testers only accept the exact word that was used in the original text. This naturally leads to lower final scores but dose not usually change the students' ranks.
4， Unless the aim of the close test is to test overall language proficiency， as advocated by Dller 1979， such tests may be a wasteful way of testing. Few of the items in any one passage may test the aspects of language with which the tester is concerned..
In 1979 and 1983 Alderson and in 1981 Klein-Braley reported studies using more homogeneous samples of examinees and came to the conclusion that cloze tests， while theoretically sound， suffered from a number of major technical defects： In order to ensure a sufficient number of items， cloze tests have to be relatively long. Because of the deletion principle， cloze tests usually consist of only one longer text. This can result in text specificity and thus in test bias. The factors 'text'， 'deletion rate' and 'starting point' affect reliability and validity coefficients. If the exact method of scoring is used， then cloze tests are often too difficult for adult educated native speakers. If the acceptable method of scoring is used， a large subjective component enters the scoring， and the tests are much less reliable. Moreover scoring the tests consumes much more time. The difficulty of cloze tests depends on the proportion of structure and content words deleted.
Ulrich Raatz also put forward his point of view： Many of the cloze tests reported in the literature are less reliable than originally assumed. （Ulrich Raatz， Introduction to language testing and C-Tests）
2. Chapter II An elaboration of C- Test
2.1 About C-Test.
A C-Test is an integrative written test of general language proficiency based on the concept of reduced redundancy. A C-Test consists of five to six short authentic texts， each complete as a sense unit in itself. In these texts the first sentence is left standing. Then the 'rule of two' is applied： beginning at word two in sentence two the second half of every second word is deleted. Numbers and proper names are usually left undamaged， but otherwise the deletion is entirely mechanical. The process is continued until the required number of blanks has been produced （in the canonical C-Test either 20 or 25）. Then the text is allowed to run on to its natural conclusion. The instructions for the examinees say something like 'In this test parts of some of the words have been damaged. Please replace the missing parts.' Texts are arranged in order of difficulty with the easiest text first. （Ulrich Raatz， Introduction to language testing and C-Tests） east 100 items. The deletion rate and the starting point for deletions should be fixed.
The words affected by the deletions should be a genuinely representative sample of the elements of the text.
Examinees with special knowledge should not be favoured by specific texts， therefore the new test ought to consist of a number of different texts.
Only exact scoring should be possible so as to ensure objectivity.
Native speakers ought to be able to make virtually perfect scores on the test： 90'% or higher. If native speakers cannot make scores higher than 90%， then the text should not be used for non-native speakers.
The new test should be reliable， valid and easy to develop.The first results presented in 1982 showed that all these criteria were fulfilled. Subsequent research （cf. Grotjahn， Klein-Braley， Raatz， this volume） has confirmed these early findings. （Ulrich Raatz， Introduction to language testing and C-Tests）
The C-test is really a variety of close， which its originators claim is superior to the kind of close test. Instead of whole words， it is the second half of every word which is deleted. （Arthur Hughes， Testing for Language Teachers）
In 1981 Raatz and Klein-Braley presented C-Tests as a technical improvement over cloze tests. They had set up a number of conditions which the new test format should fulfill：
" The new test should be much shorter， but at the same time it should have at least 100 items.
" The deletion rate and the starting point for deletions should be fixed.
" The words affected by the deletions should be a genuinely representative sample of the elements of the text.
" Examinees with special knowledge should not be favoured by specific texts， therefore the new test ought to consist of a number of different texts.
" Only exact scoring should be possible so as to ensure objectivity.
" Native speakers ought to be able to make virtually perfect scores on the test： 90'% or higher. If native speakers cannot make scores higher than 90%， then the text should not be used for non-native speakers.
" The new test should be reliable， valid and easy to develop. The first results presented in 1982 showed that all these criteria were fulfilled. Subsequent research （cf. Grotjahn， Klein-Braley， Raatz， this volume） has confirmed these early findings.
Lucy Katona and Zoltan Dornye （The C-Test： A Teacher Friendly Way to Test Language Proficiency） pointed out that The C-test is an integrative testing instrument that measures overall language competence， very much like the cloze test. It consists of four to six short， preferably authentic， texts in the target language， to which "the rule of two" has been applied： the second half of every second word has been deleted， beginning with the second word of the second sentence； the first and last sentences are left intact. If a word has an odd number of letters， the "bigger" part is omitted， e.g.， proud becomes pr-. One-letter words， such as I， are ignored in the counting. The students' task is to restore the missing parts. In a typical C-test there are 100 gaps-that is， missing parts. Only entirely correct restorations are accepted. By way of illustration， here is a short C-test： "One cool autumn evening， Bob L.， a young professional， returned home from a trip to the supermarket to find his computer gone. Gone！ All so- of cr- thoughts ra- through h- mind： H- it be- stolen？ H- it be- kidnapped？ H- searched h- house f- a cl- until h- noticed a sm- piece o- printout pa- stuck un- a maga- on h- refrigerator do-. His he- sank a- he re- this sim- message： can't continue， file closed， bye."
C-Tests belong to the family of language tests using the reduced redundancy approach towards test development. They are intended as a technical improvement on cloze tests. C-Tests consist of 4 to 6 authentic texts in the language required， should be selected from different sources and must be appropriate in content and difficulty level for the target group envisaged. C-Tests have been used in a number of languages， .e.g. English， French， German， Italian， Russian， Spanish and Portuguese. For English and German texts readability type prediction formulas for estimating difficulty before test use are available.
Here are two samples of C-Test：
1 Haywood football team needs new players
You li_ _ football a_ _ you do_ _ have a te_ _ to pl_ _ with？ Th_ _ come t_ our ne_ _ training o_ Wednesday after_ _ _ _ in t_ _ new spo___ hall a_ Haywood. Every ____ is wel_ _ _ _ _ here， ju__ bring yo_ _ football sh_ _ _ along a_ _ have f_ _. Our n_ _ trainer i_ perfect， la_ _ year w_ even be_ _ the team of Nottingham High School. We're waiting for you！
2. Fill in the missing words
Before the Dixons _____ to Seahouses， Mrs Dixon needed s _ _ _ things f _ __ _ the shop. She t _ _ _ Sarah to bring three t _ _ _ of cat f _ _ _， four s _ _ _ _ bottles _ _ water and a ___ of muesli. Sarah didn't！_____ to her mother. She b _____ thew____ things and f _ _ _ _ _ to bring the muesli. Mrs Dixon didn't want _ _ _ chocolate， but Sarah bought five _ _ _ _ of chocolate.
_ _ _ _ she came home， Sarah's mother was a _ _ _ _. She shouted： "You n _ _ _ _ listen to what I tell you." Sarah said： "Sorry， mum， but I'm soe _______ about our t __ _ _ to Seahouses." Then Dad _ _ _ _ in and said： "Don't w _ _ _ _ ！ We can get all th _ _ _ things in Seahouses， t _ erits of
2.2 Merits of C-Test.
C-Tests have proved to be highly reliable and to have high empirical validity in terms of correlations with other language proficiency tests. Their construct validity has been the subject of considerable research which is documented in the C-Test bibliography. In The C-Test： A Teacher Friendly Way to Test Language Proficiency， Lucy Katona and Zoltan Dornye came to such a conclusion：
1. The C-test appears to correlate well with our other language-proficiency tests， which proves that it is a good and reliable testing instrument. It is also an integrative language-testing method， correlating highly with both vocabulary and grammar scores， as well as speaking skills （圖1）.
图1
2. The C-test measures general language proficiency more accurately than the cloze， and it is a lot easier to construct and to score （图2）.
图2
3. We recommend the exact scoring method （that is， accepting only entirely correct restorations）， as we found that accepting words with spelling errors made no significant difference to the student rankings.
2.3 Demerits of C-Test.
Disadvantages of C-Test are as follows： C-Test is harder to read than a close passage， and correct responses can often be found in the surrounding text. （Arthur Hughes， Testing for Language Teachers）
Alderson（1995） pointed two problems facing C-Test： 1， The first few letters of a missing word are given in a C-Test text will reduce the number of possible answers for any one gap.2， it does not provide enough clues for educated native speakers to be able to draw._."
Chapter III C-Test： a better method over Close Test.
According to Lucy Katona and Zoltan Dornye， the C-test was developed as a modification of the cloze test， which is a frequently used， major language-testing instrument， extremely popular because of the ease of constructing it and its high reliability and validity. （The cloze test consists of a longer text of which every fifth to tenth complete word is left out.）
In the paper， The Comparative Study of the Validation between the Classical Cloze Test and C-Test， through research and questionnaire， Zhang Ling & Han Hui came to such a conclusion： 1， close format is superior in terms of its constructing and scoring method and reliability. 2， as pragmatic test format， which close format has more content coverage to measure a candidate's abilities， including linguistic and non-linguistic aspects. 3， which close test could provide more information to predict a candidate's actual performance in language learning. The study shows that C-Test could be adopted in vocabulary practice， while classical test could be tested in pragmatic language test.
The C-test appears to have some advantages over the cloze test： 1. As students are confronted with a variety of short passages， a better sampling of content areas is possible. Also， a person with special knowledge in a certain field does not have an unfair advantage.
2. By "damaging" every second word， we can obtain a more representative sample of all the different language elements in the text than in the cloze， where normally every fifth or sixth word is left out.
3. Many more items can be included in much shorter texts， making the test less time-consuming for the students than the cloze.
4. Unlike the cloze test， scoring is easy and objective， as there is only one acceptable solution in most cases.
5. As a rule， students actually like doing C-tests， whereas the cloze test is one of the most frustrating test types for learners.
6. A C-Test of 100 items takes little space and not really so much time to complete.
References：
Alderson， J. Charles， etc.1995 .Language Test Construction and Evaluation. Cambridge University Press
Hughes Arthur， 1941. Testing for Language Teachers. Cambridge University Press
Harrison Andrew. 1983， A language Testing Handbook. Macmillan Press， London
Raatz Ulrich，Klein-Braley Christine，etc. Introduction to language testing and C-Tests
參考文献
[1] 张玲，韩慧，2005，传统完形填空与C-试题的效度对比研究， 新疆教育学院学报， 2005年9月第21卷第3期
收稿日期：2012-12-12