CONTENT

Acknowledgements…………………………………………………….2

Abstract (Chinese)….………………………………………………….3

Abstract (English)……………………………………………………...5

 

Part I Introduction—a Survey on General Reading Theory

1.1    Interactive approaches to ESL/EFL reading……………………..8

1.2   SL/FL reading, a reading problem or a language problem? …….16

Part II Reader factors that affect reading………………………..20

2.1 Background knowledge and ESL/EFL reading………………….21

2.2 Language proficiency and ESL/EFL reading……………………23

2.3 Reading skills and ESL/EFL reading……………………………25

Part III An Experiment……………………………………………28

3.1 Subjects………………………………………………………….31

3.2 Reading material…………………………………………………31

3.3 Design and procedures…………………………………………..32

3.4 Data collection…………………………………………………...33

Part IV Results and Implications

4.1 Overall case discussion………………………………………….34

4.2 Group discussion………………………………………………...37

 Part IV Conclusion

5.1 Conclusion ………………………………………………………45

5.2 Pedagogical implications………………………………………...45

5.3 Suggestions for further research………………………………..48

Bibliography……………………………………………………….. 50

Appendix…………………………………………………………….52

Acknowledgements

 

I am deeply grateful to the following people for helping me to make this paper possible:

To Prof. Zhu Yue for his instruction, patience, kindness and advice;

To my teachers in the School of Foreign Studies of Anhui University: Prof. Zhou Fangzhu, Prof. Hong Zengliu, Prof. Zhang Zuwu, Prof. He Gongjie, Prof. Li Yongfang, Prof. Xiao Shuhui, and Prof. Cao Yi for their lectures;

To my colleagues in the Department of College English Teaching of Anhui University for their help and encouragement;

To my husband and daughter and friends for their support.

 

内容提要

影响阅读的因素有语言因素和非语言因素。语言因素指阅读材料而言，包括语篇所使用的词汇、句法、所涉及的话题等。非语言因素有读者因素和情景因素。读者因素指阅读者的阅读动机、阅读兴趣、其智力及语言水平、已有知识等。情景因素指阅读活动时的社会和物理环境。根据阅读的互动模式，阅读是这些因素相互作用的过程。

本文对三种读者因素及它们对阅读的影响进行了研究。1、背景知识在理解语篇时必不可少。根据图式理论，语篇本身没有任何意义，只是为阅读者根据已有知识构建意义提供线索。2、尽管研究者对语言水平影响外语阅读的程度持有不同看法，他们的研究成果有达成共识之处：基本的语言技能在阅读活动中必不可少；已有的阅读技能在外语阅读中的运用受到语言水平的制约。3、阅读技能在阅读中起着重要作用，对此，研究者们感兴趣的是外语阅读者已经具备的母语阅读能力是否向外语阅读迁移。达到认可的一点是母语阅读确有向外语阅读迁移的倾向(Carson et al., 1990:256)。Perkins(1989:7)等人发现，母语和外语阅读理解之间有显著意义的相关。王初明等人(1992：173)认为需要细致深入的实验来研究阅读的各个组成部分，甚至理解能力亦可进一步分解以便揭示高层次阅读的奥秘。目前，国外的研究多为第二语言研究，其结果不一定适合中国的外语学习者，而国内这方面的实验大多是以外语专业的学生为对象；另外，在他们的实验中，用于测试阅读理解的题目难度不一致，有就文章事实提问的简单题也有较高难度的推断类高层次理解题，作者认为，阅读理解题的性质不同对读者的认知能力和语言水平的要求也就不同，需要分别对待。

本实验在Perkins等（1989）和王初明等（1992）的研究基础上进行并有所发展。它以非英语专业本科生为研究对象，以实验对象的英汉语言水平、英汉阅读理解（高层次的理解）及英汉阅读速度为变量，将背景知识、阅读动机、参与意识等作为控制变量。将合格对象的六个项目的数据先用SPSS软件包中的Pearson相关分析程序进行统计分析，得出变量总体相关距阵，再分别将六个项目排序，得出十二个高分组和低分组，进一步观察变量间在相关程度上的变化。作者对数据的意义作了分析。实验结果显示：汉英阅读速度之间存在着迁移倾向（r=.56**），但是阅读速度的迁移在英语水平、汉英阅读速度、阅读理解等不同层面上呈现不同的程度；汉英阅读在较高层次的理解上的迁移倾向不明显（r=.04）；汉语水平对汉语阅读速度、阅读理解不能产生显著差别，汉语阅读主要是阅读技巧问题；英语水平在英语阅读中起着重要作用，英语阅读是语言问题也是阅读技能问题。

基于外语阅读与外语水平关系的讨论，本文指出大量阅读不仅可以提高阅读能力, 还是提高语言水平的有效途径。基于背景知识和阅读关系的探讨，本文向外语学习者和阅读材料编辑者推荐“狭窄阅读”法，即阅读类似话题或同一作者的作品，这样可帮助读者建立新图式或激活已有图式，提高阅读效果。鉴于阅读速度的迁移倾向,本文认为外语阅读速度慢的学习者可有意识地借助母语阅读改变阅读习惯。

关键词：阅读；读者因素；阅读速度；阅读理解；阅读技能；迁移；

Abstract

The factors that affect reading include language factors and non- language factors. Language factors relate to reading material, including the lexical items used, syntax, topic, etc.. Non-language factors include reader factors and situational factors. Reader factors refer to reader’s motivation and interest in reading, his intelligence and language proficiency, prior knowledge, etc. Situational factors refer to the social and physical situations of reading activity. According to the Interactive Model of reading, reading is the process in which all these factors interact with each other.

This paper is a study on three of the reader factors and how they affect reading. 1. Background knowledge is essential to understanding a text. According to schema theory, a text does not carry meaning by itself. Rather, it only provides directions for readers as to how they should construct meaning from their own previously acquired knowledge. 2. While researchers have different opinions on how much language proficiency affects FL or SL reading, their findings indicate that basic language competence is involved in reading comprehension and FL proficiency ceiling may restrict the degree to which good reading skills are applied to FL reading. 3. Reading skills play an important role in reading. In this point, the researchers are interested in whether the NL reading abilities that an SL or FL reader has already processed will transfer to his SL or FL reading. One point that has been agreed on is there is certain tendency to transfer from NL to SL or FL reading (Carson et al., 1990:256). Perkins et al. (1989:7) find that there is a significant correlation between NL and FL reading comprehension. Wang Chuming et al. (1992:173) consider that it is necessary to research into each component of reading, and even comprehension may be further divided to reveal the secret of high-level comprehension. Researches done abroad are mainly SL studies. Therefore, the result may not be suitable for Chinese FL learners. The researchers at home mostly take English majors as their subjects. Moreover, in their experiment, questions to test reading comprehension vary in difficulty—there are simple fact-finding ones as well as demanding high-level ones. This writer thinks that comprehension questions different in difficulty require different cognitive ability and language proficiency from readers. Therefore, it is necessary to deal with them respectively.

This experiment is based on the research work done by Perkins et al. (1989) and Wang Chuming et al. (1992), but it goes further. It uses non-English major college students as subjects, and takes such factors of subjects’ as variables: language proficiency, reading speed, and the ability to make inference standing for reading comprehension, and it controls such variables as background knowledge, motivation and sense of participation. Data of the six items are analyzed by Pearson correlation procedure in SPSS software to gain the overall correlation matrix. Then, they are respectively put in order to get six highest groups and six lowest groups for further investigation into the changes of their correlation. The writer gives her analysis on the indication of these data. The result shows that in general, there exists a tendency to transfer between Chinese and English reading speed(r=.56**); but the degrees of transfer in reading speed vary with the subjects’ English proficiency, Chinese and English reading speeds and reading comprehension; there is no significant tendency to transfer between English and Chinese reading on high-level comprehension(r=.04); reading in Chinese is mainly a problem of reading skills; English language proficiency makes significant difference in English reading speed and high-level comprehension; reading in English is both a problem of language proficiency and reading skills.

Based on the study of FL proficiency and FL reading, the writer believes that large amount of exposure to written material is an approach to improve both the language proficiency and reading ability. Based on the study of background knowledge and reading, the writer recommends “narrow reading” to FL learners and book compilers: reading materials with similar topics or by the same authors will help to build up new schema or activate existing schema which will improve reading. Finally, since there is tendency to transfer between NL and FL reading speed, it is suggested that FL learners with slow reading speed change their reading habit purposefully through NL reading.

 

Key Words: reading; reader factors; reading speed; reading comprehension; reading skills; transfer
PART I

 

Introduction---a Survey on General Reading Theory

 

A reading knowledge of a foreign language is often important to academic studies, professional success, and personal development. This is particularly true of English as so much professional, technical and scientific literature is published in English today. Reading ability is often all that is needed by learners of English as a Foreign Language (EFL), as well as of other foreign languages.  Despite this specific need for the foreign language, it is the common experience, at least of EFL teachers, that most students fail to learn to read adequately in the foreign language. Very frequently, students reading in a foreign language seem to read with less understanding than they are expected, and to read considerably more slowly than they reportedly read in their first language.

  Then what has caused the problem? Researchers on language and other relative subjects have done much research into the essence of reading process. Some of them are still influential to studies in both foreign language (FL) or second language (SL) reading and native language (NL) reading.

1.1  Interactive Approaches to ESL/EFL reading

  That reading is not a passive, but rather an active, and in fact an interactive process has been recognized for some time in first or native language reading (Goodman 1967, 1971; Kolers 1969; Wardhaugh 1969; Smith 1971; Rumelhart 1977; Adams and Collins 1979). However, only in recent decades has second language or foreign language reading been viewed as an active, rather than a passive process.

  Early research work on second language reading, specifically on reading in English as a second language, assums a rather passive, bottom-up view. The authors express their meaning by using letters and words and building up a meaning for a text from the smallest textual units at the "bottom" (letters and words) to larger and larger units at the "top" (phrases, clauses, inter-sentential linkages). Problems of second language reading and reading comprehension are viewed as essentially decoding[①] problems, deriving meaning from print. This bottom-up reading approach, which is also called phonics approach, dominates both first and second language research and theory until the end of the 60s in the last century. The central notion behind this bottom-up approach is that reading is basically a matter of decoding a series of written symbols into their aural equivalents.

 Cambourne (1979) provides the following illustration of how the process is supposed to work (Nunan, 1995:64):

Print   Every letter discriminated   Phonemes and graphemes matched   Blending  Pronunciation   Meaning

The derivation of meaning is the end process in which the language is translated from one form of symbolic representation to another.

To the end of the 70s, the so-called psycholinguistic model of reading, which had earlier exerted a strong influence on views of first or native language reading (Goodman 1967, 1971; Smith 1971) began to have an impact on views of second language reading. Goodman describes reading as a "psycholinguistic guessing game," in which the "reader reconstructs...a message which has been encoded by a writer as a graphic display" (Goodman, 1971:135). In this model, the reader need not (and the efficient reader does not) use all of the textual cues. The better the reader is able to make correct prediction, the less confirming via the text is necessary (Goodman 1973:164). According to this point of view, the reader reconstructs meaning from written language by using the graph-phonic, syntactic and semantic systems of the language, but he or she merely uses cues from these three levels of language to predict meaning, and, most important, confirms those predictions by relating them to his or her past experiences and knowledge of the language.

Although Goodman does not characterize his theory as a top-down model, and continues to resist this characterization himself (Goodman, 1981), several other reading experts (Anderson 1978; Cziko 1978) have characterized it as basically a concept-driven, top-down pattern in which “higher-level processes interact with, and direct the flow of information through lower-level processes” (Stanovich 1980:34). In any event, the impact that Goodman’s psycholinguistic theory has on both first or native language reading, and later on second or foreign language reading, is to make the reader an active participant in the reading process, making and confirming predictions, primarily from his or her background knowledge of the various linguistic levels (graph-phonic, syntactic, and semantic, in the broadest sense of that term).

The basic notion of this approach is that the reader rather than the text is the heart of the reading process. Cambourne (1979) provides the following schematisation of the approach (Nunan, 1995:65):

Past experience, language intuitions and expectations     Selective aspects of prints    Meaning    Sound pronunciation if necessary

?

  From the diagram, it can be seen that this approach emphasizes the reconstruction of meaning rather than the decoding of form. The interaction of the reader and the text is central to the process, and readers bring to this interaction their knowledge of the subject at hand, knowledge of and expectations about how language works, motivation, interest and attitudes towards the content of the text. Rather than decoding each symbol, or even every word, the reader forms hypotheses about text elements and then ‘samples’ the text to determine whether or not the hypotheses are correct.

  Other second language reading specialists such as Clarke and Silberstein (1977), Clarke (1979), Mackay and Mountford (1979), and Widdowson (1978,1983) view second language reading as an active process in which the second language teacher is an active information processor who predicts while sampling only parts of the actual text.

  In 1979, Coady elaborated on this basic psycholinguistic model for ESL reading and suggested a model in which the ESL reader's background knowledge interacts with conceptual abilities and process strategies to produce comprehension (Coady 1979:5-12) (see Figure 1).

Figure 1

Model of the ESL Reader, Coady (1979:7)

Conceptual abilities            Background knowledge

                 Process strategies

By conceptual ability, Coady means general intellectual capacity. By processing strategies, Coady means various subcomponents of reading ability, including many more general language processing skills that apply to oral language e.g., grapheme-morpho-phoneme correspondences, syllable-morpheme information, syntactic information (deep and surface), lexical meaning, and contextual meaning. Coady (1979:7) observes the role of background knowledge as follows:  

Background knowledge becomes an important variable when we notice, as many have, that students with a Western background of some kind learn English faster, on the average, than those without such a background.

Coady(1979:12) also suggests that background knowledge may be able to compensate for certain syntactic deficiencies:

The subject of reading materials should be of high interest and relate well to the background of the reader, since strong semantic input can help compensate when syntactic control is weak. The interest and background knowledge will enable the student to comprehend at a reasonable rate and keep him involved in the material in spite of its syntactic difficulty.

  Oller (1979) also stresses the importance of taking into consideration psychological as well as linguistic factors in accounting for how people read. He points out that the link between our knowledge of linguistic forms and our knowledge of the world is very close.

  In the top-down view of second language reading, the reader is an active participant in the reading process, making predictions and processing information, with the help of his prior experience and background knowledge. In this view, the reader's prior linguistic knowledge ("linguistic" schemata) and level of proficiency in the second language are important, so is the reader's prior knowledge of the content area of the text ("content" schemata) as well as of the rhetorical structure of the text ("formal" schemata). Researches done and reviewed by Carrell (1983a, 1983b 1983c, 1984a, 1984b, 1984c, 1985; Carrell and Eisterhold 1983; Carrell and Wallace 1983) within the general framework of schema theory have shown the significant roles played by both content and formal schemata in ESL reading.

The introduction of a top-down processing perspective into second language reading has had a profound impact on the field. In fact, it has had such a profound impact that there has been a tendency to view the introduction of a strong top-down processing perspective as a substitute for the bottom-up, decoding view of reading rather than its complement.

However, as schema theory research has attempted to make clear, efficient and effective reading -- be it in a first or second language -- requires both top-down and bottom-up strategies operating interactively. Rumelhart suggests that reading process is in fact a complicated “interactive” process of various types of language knowledge, including that of words, phrases, syntax and semantics (Carrell, et al., 1990:27-31). Rumelhart’s model of reading is illustrated in Figure 2[②]:

Syntactical knowledge

Semantic knowledge

Feature extraction device

Most probable interpretation

Pattern synthesizer

VIS

Grapheme

Input

?

Orthographic knowledge

Lexical knowledge

As seen in Rumelhart’s model, information from syntactic, semantic, lexical, and orthographic sources converge upon the pattern synthesizer. These knowledge sources are providing input simultaneously and a mechanism must be provided which can accept these sources of information, hold the information, and redirect the information as needed. The mechanism which can accomplish these tasks is the massage center. Rumelhart states the functions of the massage center as follows:

The massage center keeps a running list of hypotheses about the nature of the input string. Each knowledge source constantly scans the message center for the appearance of hypothesis relevant to its own sphere of knowledge. Wherever such a hypothesis enters the massage center, the knowledge source in question evaluates the hypothesis in light of its own specialized knowledge. As a result of its analysis, the hypothesis may be confirmed, disconfirmed and removed from the massage center, or a new hypothesis can be added to the message center. This procedure continues until some decision can be reached. At that point the most probable hypothesis is determined to be the right one[③].

The Stanovich model (1980) integrates concepts from a variety of sources.

In his exhaustive review of reading models, Stanovich criticizes the top-down notion that reading proceeds through the generation of hypotheses about up-coming text elements. He points out, for example, that the generation of hypotheses in the manner suggested in top-down approach would actually be more time consuming than decoding would. In the light of the perceived deficiencies of both bottom-up and top-down models, he proposes a third which he calls “an interactive-compensatory model”. As the name indicates, this model suggests that readers process texts by utilizing information provided simultaneously from several different sources, and that they can compensate for deficiencies at one level by drawing on knowledge at other (either higher or lower) levels. These sources include all those looked at separately in bottom-up and top-down processes, that is, phonological, lexical, syntactic, semantic and discoursal knowledge. He states[④]:

Interactive models of reading appear to provide a more accurate conceptualization of reading performance than do strictly top-down or bottom-up models. When combined with an assumption of compensatory processing (that a deficit in any particular process will result in a greater reliance on other knowledge sources, regardless of their level in the processing hierarchy), interactive models provide a better account of the existing data on the use of orthographic structure and sentence context by good and poor readers.

 Stanovich claims that his alternative model is superior because it deals with the shortcomings inherent in other models. The major deficiency of the bottom-up model is that it assumes that the initiation of higher-level processes, such as use of background knowledge, must await lower level processes. The top-down model, on the other hand, does not allow lower level processes to direct higher-level ones. The interactive-compensatory model allows for deficiencies at one level to be compensated for at another. In particular, higher-level processes can compensate for deficiencies at a lower level, and this allows for the possibility that readers with poor reading skills at the grapheme and word levels can compensate for those by using other sources of knowledge such as the syntactic class of a given word or semantic knowledge. Given deficiencies of lower level skills, poor readers may actually be more dependent on higher-level processes than good readers. The last point will be further discussed later.

1.2 SL/FL Reading, a Reading Problem or a Language Problem?

From the aspect of reading process, the factors that affect a reader’s FL reading are various. Some of them are text-related, including the length of words and sentences, word frequency, numbers of compound sentences, propositions, etc. Some are reader-related, such as reading skills, motivation and interest, his knowledge about the language, the text and its background. Other factors are considered situational, which refer to the social and physical situation of a reading activity.

  This thesis is a study of how reader factors affect reading comprehension. Reader factors involve language factors and non-language factors. Language factors refer to a reader’s language level including both foreign language and native language, which is fundamental in reading process. Non-language factors involve a reader’s background knowledge, intelligence, especially his ability to make inference, judgment and conclusion, and psychological quality, such as his sense of participation.

    Research work done both abroad and at home on this topic has focused on whether reading is a language problem or a reading problem. On the one hand, subjects have difficulty understanding the text despite knowing its words and structures. On the other hand, the interpretation of words and syntactic structures, that is, grammar and vocabulary, seems to be the main factor in poorer reading performance in the second language than in the first language. The problem seems to be whether reading in a foreign language is “simply” a problem of knowing words and grammar of the language, or whether there are other causes of the difficulties learners experience. 

  For more than thirty years, this issue of whether reading in an FL is a reading problem or a language problem has been a principal concern in the SL/FL reading research community (such as Yorio, 1971; Clarke, 1979, 1980; Coady, 1979; Cziko, 1978, 1980; Alderson, 1984; Perkins, 1989; Wang Cuoming, 1992).  

  Alderson's (1984) article provides the impetus for the research reported in this paper. In that paper Alderson surveys a large corpus of published researches on the question of whether reading difficulties in a second language are due to reading difficulties in the first language or due to reading abilities in a foreign language. Alderson (1984:4) summarizes the question as interference from the native language and inadequate knowledge of the target language in the following two hypotheses:

  1. Poor reading in a foreign language is due to poor reading ability in the first language. Poor first language readers will read poorly in the foreign language and good first language readers will read well in the foreign language.

  2. Poor reading in a foreign language is due to inadequate knowledge of the target language.

 Alderson finds supporters for these opposing viewpoints from linguists and researchers respectively.

  Jolly (1978) claims that success in reading a foreign language depends crucially upon one's first language reading ability rather than upon the student's level of English if this is identifiable. He asserts that reading in a foreign language requires “the transference of old skills, not the learning of new ones”(Alderson, 1984:2). Therefore, it follows that the students who fail to read adequately in the foreign language fail because they either do not possess the “old skills”, or because they have failed to transfer them.

This view is shared by Coady (1979:8), who holds the position that second language reading is a reading comprehension problem. Coady's psycholinguistic perspective is that reading is an interactive complex of abilities/skills and knowledge, some of which have a linguistic nature. Coady lays a heavy emphasis on the readers' store of textually relevant background knowledge and their attained reading comprehension proficiency in the first language, although he does concede that second language readers may lack attained competence in “process strategies which involve substantial knowledge of the target language”(Coady, 1979:8). Coady's position is summarized in the following quote (1979:12):

   We have only recently come to realize that many students have very poor reading habits to transfer from their first language, and thus in many cases, we must teach reading skills which should have been learned in first language instruction. 

Coady is indirectly supported by the “reading universals hypothesis” put forward by Goodman (1973), who claims “the reading process will be much the same for all languages”. This position has been strengthened by the work in EFL of people like Pat Rigg (1977:106-118), who found considerable similarities in reading miscues for EFL learners from a variety of different language background. If the reading process is the same or very similar in all languages, then one would expect reading ability to transfer across languages. That is, as Clark (1979:148) states, “If the reading process is basically the same in all languages we would logically expect good native language readers to be good readers to maintain their advantage over poor readers in the second language”.

  Yorio (1971:107-115) takes a different view that supports the second hypothesis. He claims that the difficulties of second language readers can be explained by second language interference and the readers' lack of sufficient second language competence.

Yorio's (1972:108) position is summarized in the following quote:

    The reader's knowledge of the foreign language is not like that of the native speaker; the guessing or predicting ability necessary to pick up the correct cues is hindered by the imperfect knowledge of the language; the wrong choice of cues or the uncertainty of the choice makes associations more difficult; due to unfamiliarity with the material and the lack of training, the memory span in a foreign language in the early stages of its acquisition is usually shorter than in our native language: recollection of previous cues then is more difficult in a foreign language than in the mother tongue; and at all levels, and at all times, there is interference of the native language.                                                                                                                               

In the summary that follows the research, Alderson reports that (1) in the bilingual studies, there is some evidence of transfer of reading ability from one language to another, although only moderate to low correlations have so far been established between reading ability in first language and reading ability in the foreign language when the same individuals are studied in both languages; (2) some evidence, however tentative, suggests that proficiency in the foreign language may be more closely associated with foreign-language reading ability; and (3) considerable support is found for the hypothesis that some sort of threshold or language competence ceiling has to be obtained before existing abilities in the first language can begin to transfer (Alderson,1984:20).

Part II

Reader Factors that Affect Reading

  In language communication, to make meaning clear, the language used should be grammatical or follow its rules in structure. At the same time, in communication, people need to abide by pragmatic principles. Language rules establish the superficial meaning of language signs, while pragmatic principles decide its practical meaning in communication. The knowledge about the world is another important component in reading ability. The readers usually use their background knowledge and personal experience to distinguish ambiguities. So, the factors that affect reading include the text, the reader, and the reading situation. Text-related factors are about reading materials, or features of language, including lexical items used, syntax, topic, etc. Reader factors or features of the comprehender refer to the reader’s motivation and interest in reading, his intelligence and proficiency, prior knowledge, etc. Situational factors refer to the social and physical situation of reading activity. Interactive model of comprehension is described as the ways in which these factors interact in the process.

  This thesis is a study of how reader factors affect reading comprehension. Reader factors involve reader’s language proficiency of both FL and NL, reading ability, background knowledge about the world and about the reading subject, intelligence, especially one’s ability to make inferences, and psychological quality, such as sense of participation. This part of the thesis will mainly discuss the relationship between reading comprehension and such factors as background knowledge, language proficiency and reading skills respectively. 

2.1 Background Knowledge and ESL/EFL Reading

The role of background knowledge in language comprehension has been formalized as schema theory. That theory holds that any text, either spoken or written, does not carry meaning by itself. Rather, a text only provides directions for listeners or readers as to how they should retrieve or construct meaning from their own previously acquired knowledge. Such knowledge is called the reader's background knowledge and the previously acquired knowledge structures are called schemata. According to schema theory, comprehending a text is an interactive process between the reader's background knowledge and the text. Efficient comprehension requires the ability to relate the textual material to one's own knowledge.

The process of interpretation is guided by the principle that every input is mapped against some existing schema and that all aspects of that schema must be compatible with the input information. This principle results in two basic modes of information processing, called bottom-up and top-down processing. Bottom-up processing is evoked by the incoming data; the features of the data enter the system through the best-fitting, bottom-level schemata. Schemata are hierarchically organized, from most specific at the bottom, to most general at the top. As these bottom-level schemata converge into higher-level, more general schemata, these too become activated. Bottom-up processing is, therefore, called data-driven. Top-down processing occurs as the system makes general predictions based on higher level, general schemata and then searches the input for information to fit into these partially satisfied, higher-order schemata. Top-down processing is, therefore, called “conceptually- driven”.

  An important aspect of top-down and bottom-up process is that it should be occurring at all levels simultaneously (Rumelhart, 1980). The data needed to fill out the schemata become available through bottom-up processing; top-down processing facilitates their assimilation if they are anticipated or consistent with the reader's conceptual expectations. Bottom-up processing insures that the reader will be sensitive to new information; top-down processing helps the reader resolve ambiguities, i.e., to select between alternative possible interpretations of the incoming data.   

  As an example of this process, consider the following three sentences (Hudson, 1982:8):

     He picked up his axe. He held it softly for a moment and blew sharp notes through its bell. The crowd listened and cheered him; they loved his saxophone playing.                                                 

When the first sentence is read, it is interpreted and the schema about a tool for chopping something is activated. However, when followed by the second sentence, the slots of the schemata are not filled; “blew, sharp notes, and bell” are not expected information or not reconciled. At this point the reader may either dismiss the initial schemata or become confused and continue to attempt reconciliation of the two sentences. Yet, when the third sentence is added, the reader may reconcile the schemata in reinterpreting the definitions of axe as a jazz instrument, namely a saxophone. Alternatively, the reader can retain the schemata from the first sentence and impose reconciliation from his or her background knowledge, a reconciliation which is not correct. Anderson and Ortony (1975) have indicated that the mental representation of the to-be-comprehended sentences is generally more elaborate and detailed than the words in the utterance might appear to entail.

  Thus, according to schema theory, readers activate an appropriate schema against which they try to give a text a consistent interpretation. To the extent that they are successful, we may say that they have comprehended the text.

 

2.2 Language proficiency and ESL/EFL reading

The ACTFL Proficiency Guidelines (1986) describes proficiency according to five scales, i.e. speaking, listening, reading, writing, and culture, in three dimensions: linguistic functions, content, and degree of accuracy. Hudson (1982:2) summarizes that proficiency is composed of basic elements, such as letter and word recognition, phoneme-grapheme correspondence, and recognition of the lexical syntactic, semantic, and discourse linguistic relationships which are presented through the text. Second or foreign language research in the past decades has attempted to determine the extent to which the psycholinguistic perspective of NL reading can explain SL/FL reading. Goodman (1971) hypothesizes that the reading process, a process in which proficient readers make generally successful predictions, will be much the same for all languages. Data based research has supported this psycholinguistic concept of reading universals for SL/FL reading, and has indicated that one's first language does not determine one's reading proficiency in SL/FL (Rigg 1977a, 1977b). However, Clarke (1979) finds that a language ceiling in SL/FL effectively prohibits the complete transfer of NL reading skills to SL/FL reading. The results of his study suggest that although the psycholinguistic assumptions of universals may be justified, the role of language proficiency in SL/FL may be greater than has previously been assumed by SL/FL researchers interested in the psycholinguistic perspective of reading. Thus, a "short-circuit" in the good reader's system is caused by a limited control over the language. Cziko (1978) suggests that syntactic, semantic, and discourse constraints serve as important sources of information for the fluent NL reader and that much of the difficulty in SL/FL reading may be due to an inability to make full use of those constraints because of low language proficiency.       

  An additional implication for SL/FL reading relates to the short circuit and its causes across levels of language proficiency. Cziko (1978) indicates that the reading behaviors of advanced proficiency SL/FL readers parallel native speaker reading behaviors, while those of intermediate and low level SL/FL readers do not. While all three proficiency levels can make use of syntactic constraints, only the native speakers and the advanced SL/FL readers can make use of both syntactic and semantic constraints. Further, the native speakers and the advanced SL/FL readers are aided by the discourse constraints available. Intermediate SL/FL readers are not aided and no data are collected for beginning level SL/FL readers. Clarke (1979) indicates that low level SL/FL readers who are good NL readers show a reduced ability to utilize the good reader strategy of relying heavily more on semantic than on syntactic cues when reading in SL/FL, and thus indicate a reduced measured superiority over poor NL readers when both reading SL/FL materials. Despite this reduced ability to transfer NL skills, however, the good NL readers are still significantly better SL/FL readers than the poor readers at the same level of language proficiency.

  Thus, while the results from Cziko and Clarke indicate that basic language competence is involved in reading comprehension, the fact that the good NL readers maintain an advantage over poor NL readers at their proficiency level when both are reading SL/FL indicates that some non-text-related factors operate. That is, it seems likely that reading skills affect the degree to which the proficiency ceiling restricts comprehension, just as the proficiency ceiling may restrict the degree to which good reading skills are applied.

 

2.3 Reading Skills and ESL/EFL Reading

Language competence is obviously a major factor in SL/FL reading. Results of the research discussed above have indicated a breakdown in language proficiency processes by SL/FL readers, but there is a need to consider the part played by non-syntactic, non-semantic, and non-discourse elements in SL/FL reading, and to consider whether, and how, the role played by these elements changes across levels of language proficiency. It is necessary to look at what non-language factors are involved in the non-transference of good reading behaviors.

  According to psycholinguistic reading model, reading skills involve the reader’s hypothesis production and testing, guessing and identification of meaning, categorization, fitting new information to prior knowledge, reconciliation of assumptions to new possibilities of meaning, and the internalization of representations. In most cases, skill and ability are used interchangeably. Widdowson (1978:67) makes a distinction between linguistic skill and communicative ability that may be worth mentioning here. Linguistic skills are those skills such as speaking, hearing, composing and comprehending that refer to the way in which the language system is manifested as usage. Those skills such as saying, listening, writing and reading that refer to the way in which the system is realized as use are defined as communicative abilities. The acquisition of linguistic skills does not guarantee the consequent acquisition of communicative abilities in a language. It means, as far as reading is concerned, a reader can comprehend the signification of a written sentence without recognizing what it counts as in the context of a particular piece of written discourse. So according to Widdowson, reading skills discussed here refer to the ability to understand a discourse involving not simply the recognition of what words and sentences mean but also the recognition of the value they take on in association with each other as elements in a discourse. In this thesis, two of these skills will be discussed: reading speed and the ability to make inferences. The term “inference” is used in its broad sense here: it refers to the ability to make generalization, judgment, and conclusion.

Experts now agree that inference is a cognitive process used to construct meaning. Inference is, in other words, a thinking process that involves reasoning a step beyond the text, using generalization, judgment and explanation. For example, in the text: She went to the hairdresser’s and her hair is blond now, the reader usually reasons a step beyond the individual words on the page to surmise what happened in the hairdresser’s salon. To do so, the reader uses generalizations of typical events in the above setting and explanatory reasoning of how those typical events would apply to the words/text at hand (P. Afflerbach, 1990:31-46). The importance of inference to comprehension of even the simplest of texts has long been demonstrated in NL studies of comprehension (Anderson, 1984; Van Dijk, 1983). Some views of inference are mostly text-based and describe readers’ inference as bridges among propositions of the text (Kintsch, 1974; Frederiken, 1975). Other views of inference are mostly model-based and describe readers as quickly building whole models of understanding of texts that are subsequently refined based on the constraints imposed by the text. Much of the study of inference has focused on the identification of inference strategies used by different types of NL readers. For SL/FL study, J. Hammadou (1991:27-38) does research on the interaction among inference, background knowledge, and language proficiency for SL readers through recall protocols written in English about texts in French or Italian. The study provides evidence that the more proficient readers comprehend clearly more of the text than the less proficient readers. As comprehension proficiency improves, qualitative differences appear rather than just simply quantitative ones. Inferences remain as external evidence of the interaction between readers and texts. They document some of the encounter between readers and texts. They also show some of what the reader brings to the process and are part of the interaction between readers’ prior knowledge and the text. In the following experiment, inference is considered an important variable.

Part III

An Experiment

As an empirical research, the thesis takes such subjects’ factors as variables: language proficiency, reading speed, and the ability to make inferences standing for reading comprehension and it controls such variables as background knowledge and sense of participation. It is true that these are important factors in reading process, and Wang Chuming et al. (1988) evidence that knowledge about reading material has more effect on reading comprehension than difficulty of the material. But study done by Hammadou (1991: 30) shows that differences between the more and less proficient groups according to familiarity with topic are not evident. Nor does the study produce a significant relationship between background knowledge and amount of inferences in recall. And it seems hard to control background knowledge if combined with other factors. This wrier would like to work on this point in her further research on reading later.

This experiment is based on the work done by Perkins et al. (1989) and Wang Chuming et al. (1992), but it goes further. The research done by Perkins et al. (1989: 8) ascertains the question of whether readers at different proficiency levels covary significantly in reading comprehension tests on the first and second languages using Japanese students as subjects. They claim their results support both Alderson’s hypotheses: “poor first language readers will read poorly in the foreign language” and vice versa, and “poor reading in a foreign language is due to inadequate knowledge of the target language”(Alderson, 1984: 4). They find evidence that imperfect knowledge of the second language can cause difficulty in second language reading.

  There are two points the writer wants to mention about their study. At first, Perkins’ research is an ESL study and its results may not be suitable to Chinese EFL learners. Second, in their tests, the difficulties for items tested vary from simple fact-finding questions to complex high-level ones. This thesis writer considers that if the difficulties for tested items are different, it might require different reading abilities to answer these questions. Therefore, these items should be tested respectively. The same problem also exists in EFL study with Chinese as subjects done by Wang Chuming and Qi Luxia. In their empirical study (1992), they take English-major juniors as subjects. They claim that they have proved Alderson’s hypothesis that poor reading in an FL is due to poor reading ability in the first language. Their results can be summarized as follows:

For those with lower English proficiency and poor Chinese reading ability, their reading efficiency and reading speed show the tendency to transfer from Chinese to English reading. But this tendency is not clearly shown on reading comprehension. In addition, to those with high English proficiency and English reading ability, their English proficiency and English reading ability, especially reading comprehension are significantly correlated.

Again two points are mentioned here. Firstly, they have failed to mention the readability of the Chinese and English reading materials used in the tests. If these reading materials are not at the same readability level, they are not compatible with each other in difficulty. Secondly, they have failed to mention the types of items tested. As just mentioned in Perkins’ study, if the difficulties of items tested are not at the same level, the results may not be compatible. At the end of their report, they point out that reading contains several components, some easier to transfer than others from NL to FL. Researchers need further experiment to study these components respectively to reveal the secret in high-level reading comprehension-- the ability to make inferences-- so as to have a comparatively thorough concept of foreign language reading.

  Experiments on reading done by other Chinese researchers (Ji Menggang: 1992; Hei Yuqing: 2000; Xu yingjun: 2000) mainly take English major students as subjects, whose English level is quite different from that of the non-English majors. Therefore, the results and their indications may not be applicable to the majority of English learners in China.

  Another investigation on the factors that affect reading headed by Guo Haoru et al (1990) suggests that among reading abilities, Chinese college students are poor in high-level comprehension to make judgment, inference, and generalization, etc.. In their report, mean scores for items of making inference are lower than those for items of word understanding and fact-finding by 10%. The scores for items of making judgment are surprisingly low. English learners have difficulty in making inferences. What are the reasons? Are these abilities universal in NL and FL reading for the same subjects? To answer these questions, the writer makes the following hypothesis:

A.   The positive effects of Chinese language proficiency on English learning include: 1) The language proficiency and learning method of the subjects with good Chinese proficiency will make positive transfer and directly affect their English learning and help them become good English readers. 2) The mental faculties cultivated during Chinese learning, such as the abilities to make imagination, generalization and judgment, will help Chinese learners have a better understanding in English reading.

B.    The negative effect of Chinese language proficiency on English learning is that the language proficiency and learning method of the subjects with poor Chinese proficiency will transfer negatively to English learning and make them poor English readers.                                                                      To prove or dispute the above hypothesis, the following questions are supposed to be answered through experiment:

1)            Whether reading speed, and reading abilities (only the high-level abilities to make inferences are tested in this experiment) in Chinese reading transfer to English reading. If the answer is positive, what is the extent?

2)            Whether Chinese language proficiency affects Chinese high-level comprehension. If it does, to what extent?

3)            What is the correlation between English language proficiency and English reading speed and high-level comprehension as far as these subjects are concerned?

3.1 Subjects   

The subjects for this study are 120 non-English major students, 3 of the natural classes in Anhui University. They are representatives of a major group of English learners in China. The experiment was done in April 2002, in the mid of their second term on campus. Finally, 102 of them are considered qualified subjects, as each of them can provide six data required by the experiment for study. Their scores of Chinese and English in college entrance examination are used to represent their Chinese and English proficiency respectively, since the reliability and validity of the entrance exam are assured.                                                                                                     

3.2 Reading Material

Three passages are chosen respectively for Chinese and English reading tests. The Chinese passages contain 1604 characters altogether and the English passages 604 words. The passages are chosen by following criteria: the content of the passages in both tests reflects similar subject matter; to avoid the influence of familiarity factor, the topics are of general interest; the readability level for both tests is grade 10, estimated by the McLaughlin(1969) method and Chinese passages are tested after being translated into English; the same skills are assessed by both tests: both Chinese and English tests have 12 inference items; and finally, any bias in content and formal schemata, positive or negative for Chinese readers, remain constant in both tests. After the tests, a questionnaire is distributed to collect some information on familiarity of the reading materials, difficulties in reading, and on reading habits from the subjects.

3.3 Design and Procedures

The basic design of the study is a split-plot (repeated measures) one. The research performed all instruction during the class hour. Six reading test papers were distributed to three class subjects in different sequence. The instructional sequence was counterbalanced to account for possible maturation or weariness sequence effect.

  Before the test began, the subjects were told that they were to have a reading test for scientific study and the results would not be included in their academic record. But for the sake of science, they would do their best according to the requirement. Then, they were trained the way to note down their reading time. The subjects were required to begin their reading at the same time, and the assistant began to write down the time passed in minute and second (by tens) on the blackboard. After reading each passage at their normal speed, they would be reminded at the end of the passage to note down the reading time. They continued to do the comprehension exercises printed immediately after the passage. After finishing each passage, the subjects were required to note down the whole time used just to make sure they would do the exercises at their normal speed, without reading back too much.

The difference between this experiment and others is that one passage and its 4-comprehension questions and their possible choices (4 for each question) are put on the same piece of paper. The comprehension questions are all high-level ones, and the subjects are allowed to read the passage again when making choices. This is also a way to avoid the effect of memory factor on reading as it was pointed out as a defect in other’s (Wang, and Qi 1992:173) experiment. For the 12 items for both English and Chinese tests, each correct choice earns one point and the total scores are recorded as reading comprehension scores for each language. Their reading speed is converted into word or character per minute.

3. 4 Data collection

At first, data of such six variables as Chinese/English language proficiency, Chinese/English reading scores and Chinese/English reading speed are analyzed by Pearson Bivariate correlation procedure in SPSS software to gain the overall correlation matrix of the 6 variables. Secondly, data of above six variables are respectively put in order, to get 6 highest groups and 6 lowest groups (n=28) for further investigation into the changes of their correlation.

Part IV

Results and Implications

4.1 Overall Case Discussion

Table 1 presents the ranges, means, standard deviations, and skewness of the six variables mentioned above (all the data are included in Appendix II):

Table 1 Descriptive Statistics

	Range	Minimum	Maximum	Mean		Std. Deviation	Skewness
Statistic	Statistic	Statistic	Statistic	Std. Error	Statistic	Statistic	Std. Error
Chinese    proficiency
47.00	85.00	132.00	109.8824	.7998	8.0780	.111	.239
English proficiency
49.00	75.00	124.00	105.0490	1.0343	10.4459	-.327	.239
Chinese reading score
7.00	2.00	9.00	5.5294	.1556	1.5714	.195	.239
English reading scorn
10.00	2.00	12.00	6.7255	.1916	1.9354	-.262	.239
Chinese reading speed
894.00	175.00	1069.00	426.1471	15.3526	155.0539	1.391	.239
English reading speed
109.00	49.00	158.00	99.0294	2.7582	27.8566	.255	.239
Valid N (list wise)
102

?

The column of skewness shows that numerical values of 5 variables are in normal distribution (Skewness statistic<1) except for that of Chinese reading speed, which is in positively skew distribution (Skewness=1.39) indicating most subjects read Chinese passages at higher speed. For Chinese reading, its speed is distinctively higher than English reading while its mean value of reading comprehension is lower than that of English, which might be the result of its high speed. The standard deviation for Chinese reading speed is somewhat large, which indicates the considerable difference between subjects in Chinese reading speed.

Table 2 presents the correlation matrix for the above six variables in general.

Table 2

	Chinese proficiency	English proficiency	Chinese R. score	English R. score	Chinese R. speed	English R. speed
Chinese proficiency
1.00	.12	.08	-.02	.27**	.06
English proficiency
.12	1.00	-.05	.44**	.32**	.40**
Chinese R. score
.08	-.05	1.00	-.05	.04	.05
English R. score
-.02	.44**	-.05	1.00	.12	.26**
Chinese R. speed
.27**	.32**	.04	.12	1.00	.56**
English R. speed
.06	.40**	.05	.26**	.56**	1.00

** Correlation is significant at the 0.01 level (2-tailed).

Although they are just moderate or low, the correlation coefficients in Table 2 do reflect some important tendency in Chinese and English reading to a certain extent. Firstly, the reading speeds of Chinese and English are moderately related (r= .56**) while the reading scores of the two languages are not significantly correlated. That is, in the aspect of reading speed, there exists the tendency to transfer from Chinese to English reading-- fast Chinese readers are fast English readers and vice versa. But it is not the case in high-level comprehension of the two languages-- good comprehension in Chinese reading does not result in good comprehension in English. Secondly, the correlations between reading speed and reading comprehension are not significant or low, especially for Chinese reading (r= .04), which means quick readers do not comprehend better, nor do slow readers. We will come back to this point in group-discussion later. Thirdly, Chinese and English proficiency embodied by Chinese and English scores respectively in college entrance examination have low correlation coefficients with Chinese and English reading comprehension and reading speed in this experiment. Among them, the lowest coefficient is between Chinese proficiency and Chinese reading Comprehension (r= .08), the lowest but one is for Chinese proficiency and Chinese reading speed (r= .27**). So the reliability of Chinese entrance exam scores as Chinese proficiency is even doubted, because in entrance exam what is tested is comprehensive: except for modern Chinese literature and Chinese reading, it also involves knowledge on ancient Chinese literature, authors and their works and writing. However, considering the assured validity and reliability, the exam papers are going to meet the requirement of teaching syllabus for this subject. Finally, Chinese proficiency shows no significant correlation with English proficiency, English reading speed or comprehension. But English proficiency shows moderate correlation with English reading speed and comprehension(r= .40**, r=. 44**) respectively.

Have the above results refuted the hypothesis or answered our questions? To look into the problem further, the data should be analyzed from different aspects. We order the number of above six variables and get 6 highest groups and 6 lowest groups in which the correlation of the six variables makes some meaningful changes.

 

4.2 Group discussion

Table 3 presents the correlation coefficients between Chinese and English reading speed in each of the 12 groups.

Table 3   

Chinese reading		English reading		Chinese proficiency	English proficiency
speed	comp. score	speed	comp. score
High gr.  Low gr..	High gr.  Low gr. gr.	High gr.  Low gr.	High gr.  Low gr.	High gr.  Low gr.	High gr.  Low gr.
.16    .49**	.62**  .56**	.10     .05	.64** .52**	.57**  .57**	.38*   .54**

Significant correlation between Chinese and English reading speeds does not occur in the highest Chinese reading speed group and lowest English reading speed group, which seems to contradict the results of other researchers (Alderson, 1984; Wang and Qi, 1992). But the subjects in this experiment are different from those in others in their English proficiency, as it has been mentioned. Therefore, two views are put forward here: Firstly, the transfer of Chinese reading speed to English reading occurs within limits and under restraint of English proficiency. It is not that the faster a reader reads in Chinese, the faster he will read in English. But a fast Chinese reader may not necessarily be a fast English reader while a slow Chinese reader has the tendency to be a slow English reader. Secondly, the reason for the slow reader of English is not caused by their slow Chinese reading speed only. For our subjects, the largest obstacle in English reading is language factor. In the Questionnaire (see Appendix I) which was done immediately after the reading tests, when asked the most important factor that affected their English reading, 59% of the subjects ticked the choice item “lots of new words”. As a matter of fact, the words in the three English passages have been checked to make sure they all have appeared in text books and are not new to the subjects beforehand except for some specialized terms. 25% consider the reason to be “complex sentence structures. For the question “in what way have you read the English passage”, 64% chose “word by word, read it again when the passage is not understood”. Because of their poor English proficiency and reading method, the FL readers spend much time tackling some unfamiliar words and sentence structures that will inevitably make their reading speed slow and comprehension poor, especially in high-level comprehension. Therefore, in college English teaching one of the important parts in teaching practice is to help students to improve language proficiency, without which it would be impossible to improve student’s reading speed and comprehension.

The general tendency shows that there exists a moderate correlation between Chinese reading and English reading in the aspect of reading speed (r=56**). The transfer of reading speed occurs more easily for slow Chinese readers, whose word-for-word reading habit has affected their English reading speed. From the view of English language proficiency, the subjects with lower English proficiency have more significant tendency to transfer than those with high English proficiency in reading speed from Chinese reading to English reading. This is identical with the result of Alderson and Wang, i.e. the poorer the reader’s FL proficiency is, the more likely he is to read FL with NL reading habit.

West (1978) divides reading skills into different levels and reading skills are regarded as a combination of many sub-skills. He considers reading a huge psychological process composed of many component processes，which will coordinate with each other. Theoretically speaking, it is certain for reading ability to transfer from NL to FL. But the transfer has its limitations. To Chinese students, English is considerably different from their NL in word formation, phonological and grammatical rules and pragmatic principles. These differences inevitably make obstacles in their English reading. Foreign language proficiency, reading skills and NL reading habits work together to have an effect on the transfer of reading speed. On the one hand, those with word-for-word reading habit in NL may apply this habit to their FL reading and make their FL reading slow. Those who can use some reading skills in NL reading such as skimming and scanning may read Chinese quite fast. But restrained by their FL proficiency, their grasped reading skills can hardly be applied to FL reading and they are still slow FL readers. On the other hand, those who are both poor in FL proficiency and NL reading will be likely to depend on their NL reading habits to read FL. With the improvement of their FL proficiency, this dependence may become less, but for the subjects in this experiment, their FL proficiency has not been up to the level so as to be free from the effect of their NL on FL reading.

From the view of reading comprehension, what the case will be for transfer in Chinese and English reading? Table 4 presents the correlation coefficients between Chinese reading comprehension and English reading comprehension in the 12 groups.

Table 4

Chinese reading				English  reading			Chinese proficiency	English proficiency
speed		comp. score		speed		comp. score
High gr.   Low gr.		High gr.  Low gr.		High gr.  Low gr.		High gr.  Low gr.	High gr.  Low gr.	High gr.  Low gr.
-2.4   .07	.27    -.25		.04     .01		-.11     -.12		-.15     .03	.00      .15

Table 4 shows no significant correlation between Chinese and English reading in high-level comprehension. The high or low language proficiency cannot make any difference in their reading comprehension for both languages. To illustrate this point more clearly, the data of this group are analyzed again with Independent-Samples T-test. Independent-Samples T-test is a statistic method used to compare the mean values of two samples in one test. The result is presented in Table 5.

Table 5

	Chinese reading compr. high group	Chinese reading compr. low group	T- value	Sig. (2-tailed)	95% confidence interval lower  upper
English R. comprehen-  sion	Mean  6.5 Std. error  2.0	Mean  6.9 Std. error 1.7	-.85	.40	-1.2    0.49

Table 5 indicates the subjects’ mean values of English reading comprehension do not make any difference in the highest or lowest group of Chinese reading comprehension (p=.40>.001). That is, whether the subjects are good or poor in Chinese reading comprehension will not affect their English reading comprehension.

It is supposed that a reader’s mental faculties to make inference would be the same or similar in both NL and FL reading. But this supposition is not proved in this experiment.

A study by Aron (1978) offers a similar result—there is little relation between reading ability in the first language and that in the foreign language. In Aron’s test, low correlations are found between skills in the two languages: for detail, r= .50; for understanding the main idea, r= .37; and for inference, r= .28. This lack of relationship between reading ability in the NL and that in the FL is attributed by the author to “possible differences between the differing models of the world which are held by differing language/culture groups” (Alderson, 1984:13). Unfortunately no evidence is presented to support this interesting assumption. As Alderson (1984:14) points out, this lack of relationship between the reading abilities in the two languages might as well be due to inadequate language knowledge.

Then, what is the factor that affects high-level comprehension in both Chinese and English reading, is it language proficiency or reading skills? Let us have a look at Chinese reading at first. The coefficients between Chinese language proficiency and Chinese reading comprehension in the related 6 groups are presented in table 6.

 

Table 6

Chinese reading		Chinese proficiency
speed high gr.  low gr. .48*        -.18	compr. score high gr.    low gr. -.04        .28	high gr.  low gr. -.08       -.30

Table 6 shows at least two points. Firstly, high-level comprehension in Chinese reading is not related with Chinese language proficiency in general. That is, the difference in Chinese language proficiency is not the reason for the good or poor reading comprehension. Secondly, the language proficiency of the subjects with some reading skills (what can be tested in this experiment is reading speed) is positively correlated with their reading comprehension. That is, the subjects with fast reading speed are more able to grasp the global idea of the material and have a better performance in high-level comprehension.

Accordingly, the main problem for Chinese NL reading might not be a language problem, but rather a reading problem—the problem of reading ability and skills to which not much attention is paid in Chinese teaching in middle school education. Much more time and energy are put on the teaching of language knowledge and moral education because of the traditional belief that writings are to convey truth. Even the explanation of language knowledge is limited to the usage of words and rhetorical devices and overall arrangement in writings. What lacks is the training in reading ability and skills so that Chinese students are in need of a solid base to cultivate FL reading ability (Shu Dingfang, et al., 1996:131).

Then, let us take a view at the factor that might affect high-level comprehension in English reading: language proficiency or reading skills. Table 7 presents the correlation coefficients of English language proficiency and English reading comprehension in the related 6 groups.

Table 7

English reading		English proficiency
speed high gr.     low gr. .38*        .16	compr. score high gr.      low gr. .40*        .13	high gr.     low gr. .17        .54**

Table 7 indicates English reading and English language proficiency are positively correlated though the coefficients are low. The language proficiency of those faster readers or those with good comprehension significantly affects their reading comprehension. But what should be noted is that the correlation exists only in the lowest English proficiency group. It implies that those with higher language proficiency may not be better at high-level comprehension. It seems abnormal. But one fact is stressed again here: the comprehension exercises are to test the subjects’ ability to make inferences that requires the readers to have certain reading skills. The readers should be able to read between the lines, and understanding on word level is not enough. Therefore, different from NL reading, FL reading is both a reading problem and a language problem. What is showed in the highest English proficiency group may reflect another problem in middle school English teaching-- some teachers and English learners pay much attention to what may probably be tested while neglecting the cultivation in comprehensive intelligence and the training in reading skills. If it is true that the main purpose for English teaching in middle school is to lay language foundation, it would also be true that in collegiate teaching period, there is an urgent need for training of reading skills in addition to the improvement of language proficiency. For a clearer view of the effect of English language proficiency on English high-level comprehension, Independent-Sample T-test is applied to analyze the related data of English reading speed and reading comprehension in the highest and lowest English language proficiency. The result is presented in Table 8.

Table 8

	English proficiency highest group	English proficiency lowest group	T value	Sig. (2-tailed)
	Mean     Std.error	Mean     Std.error
English r. speed	112.73     24.85	86.52     19.78	3.79	.000
English r. compr.	7.77      1.43	6.04      1.74	4.17	.000

Table 8 shows that English language proficiency makes significant difference to English reading speed and high-level comprehension in general (p=.000 <.001). That is to say, a reader’s English language proficiency affects not only his reading speed but also his reading comprehension. But language proficiency involved in reading comprehension on high level is just one of the factors-- it requires reading skills simultaneously.