Asian Journal of English Language Teaching Vol. 8, 1998, pp. 19-40
© 1998 CUHK English Lanuage Teaching Unit

Reading Easy and Difficult Texts in English and Chinese: Strategy Use by Native Speakers of Chinese

Xiwu Feng
City University of New York
Kouider Mokhtari
Oklahoma State University

In this study, twenty adult native speakers of Chinese read and verbally reported their thinking processes while reading easy and difficult expository texts in English and Chinese. The subjects' think-aloud reports were collected as evidence of the strategies they used and subsequently analyzed to ascertain whether there were any differences in strategy use when the subjects read passages of differing difficulty in English and Chinese. Although all subjects demonstrated awareness of a variety of reading strategies, the pattern of strategy use in their think-aloud reports suggested that where differences existed, strategies were used more frequently when reading in English than in Chinese, and more frequently for difficult texts than for easy texts. These findings are discussed in the context of current research which has used verbal reports as a source of insights into first and second language reading.

The purpose of this study was to examine the strategies used by native speakers of Chinese while reading easy and difficult expository passages in English and Chinese. Specifically, we analyzed verbal report data to investigate (1) whether differences in strategy use existed among twenty adult native speakers of Chinese when they read English and Chinese, and (2) whether text difficulty affects strategy use when reading English and Chinese. These research questions are explored within the context of current research and practice relative to learner strategies in first and second language reading.

Reading is a complex process in which competent readers orchestrate a number of knowledge sources using a variety of strategies to comprehend what they read. These knowledge sources include what Smith (1994) refers to as "visual information" which involves awareness of linguistic and print conventions, as well as "non-visual information" which includes the knowledge that readers bring to the process of reading. In their attempts to make sense of what they read, readers resort to a number of strategies, which are deliberate, conscious plans readers execute when processing textual information. These strategies (e.g., previewing text, using headings and subheadings, re-reading, evaluating understanding, etc.) enable them to interpret printed information quickly and efficiently.

In recent years, research on the reading strategies of first and second language readers has become quite sophisticated thanks to the work of a number of researchers (e.g., Block, 1986, 1992; Carrell, 1984; Ericsson & Simon, 1993; Garner, 1987; Kletzien, 1991; Olshavsky, 1976 - 1977; Pressley & Afflerbach, 1995) who have found that the strategies which readers use when they interact with printed information play an important role in reading comprehension in both first and second language reading. These researchers and practitioners have concluded that successful readers are strategic readers and that non-successful readers often seem to be unaware of how and when to best use strategies when they read. There is also some consensus among the above researchers that different text types require different strategies and that efficient reading strategies are not acquired simply by reading, but that they should be learned through direct, formal instruction.

There appears to be widespread agreement among language and literacy researchers and practitioners that reading strategies developed in a first language tend to be transferred to a second language despite the differences that may exist between first and second languages (e.g., Alderson, 1984; Carrell, 1991; Clarke, 1980; Cohen, 1996). However, the variability which exists among readers in terms of both reading ability and language proficiency raises unanswered questions. For instance, the fact that a strategy appears -- or does not appear -- in first (L1) or second language (L2) does not indicate where it originated from. It cannot be presumed that the differences in the nature of languages account for the fact that some strategies appear more often in one language than in the other. A variety of factors may influence strategy use in L1 and L2. Clarke (1980), for instance, observed that "limited control over the language `short-circuits' the good reader's system, causing him/her to revert to poor reader strategies when confronted with a difficult or confusing task in the second language" (p. 206). A similar observation was made by Alderson (1984) who suggested that nonnative speakers cannot truly become skilled readers in a second language until they have reached "some sort of a threshold or language competence ceiling...before existing abilities begin to transfer" (p. 21). Carrell (1991) reported that "both first language reading and second language proficiency have significant effects on second language reading ability" (p. 167).

We concur with the above researchers that the level of language proficiency, the level of reading ability, and the type of language used (L1 or L2) are likely to play significant roles in determining the type and amount of strategy transfer from one language to another. While it cannot be assumed that English is always the weaker reading language of all second language learners, our teaching experience suggests that in many cases, nonnative speakers of English are often at a disadvantage using effective reading strategies in English because of potential linguistic, cultural, and other related problems. One of the languages used in the present study (Chinese) provides a prime example of a medium that is different enough from English that the transfer of strategies may not be straight- forward. For instance, unlike in English, there is no grapheme-phoneme relationship in Chinese. This means that the Chinese beginning reader of English must learn those correspondences, a strategy never needed in learning to read in Chinese. Conversely, the frequent repetition of about two hundred radicals in Chinese may not necessarily relate to morpho- logical and spelling constraints that are analogous to English. Therefore, some of the strategies that may be routinely used in one language may prove to be unnecessary in another language. According to Field (1984), the transfer of some reading strategies from Chinese to English is likely to be difficult for native speakers of Chinese, in part because of the adjustments which occur in the switch from reading an ideographic language to reading an alphabetic one.

Researchers studying language and reading strategies in first and second language have devised a variety of methodologies for researching, assessing, and teaching reading strategies (see for example Afflerbach & Johnson, 1984; Block, 1986, 1992; Carrell, 1984; Ericsson & Simon, 1993; Fawcett, 1993; Garner, 1987; Kletzien, 1991; Olson, Duffy, & Mack, 1984; Pressley & Afflerbach, 1995). One technique that has been used with varying degrees of success is the "think-aloud protocol" in which subjects verbally report their thinking as they perform a task such as reading. The think-aloud reports can be analyzed and used as the basis for understanding human thoughts and actions. In the case of language and reading comprehension, the data can also be used to assess strengths and needs which are ultimately used to inform instruction. However, like most of the research methodologies which have been used to examine human thoughts and actions, the uses of think-alouds in investigations of reading has been accompanied by a history of claims and challenges to the reliability and validity of such a methodology. These challenges have generally centered on issues of subject characteristics and task parameters which have affected the completeness, accuracy, and write-ups of verbal reports. For an excellent discussion of many of these problematic issues, see Cohen (1996) for some insightful ways of refining verbal report methods.

While the intent of this paper is not to review all of the studies which have used protocol analysis as an approach of investigating text processing, we refer the readers to the many excellent sources referenced in this paper. Perhaps the two most significant contributions made in this promising area of research to date are Ericsson and Simon's Protocol Analysis: Verbal Reports as Data (1993), which provides an excellent synthesis of protocol analysis research and offers a practical set of methodological recommendations for using protocol analysis as a research technique, and Pressley and Afflerbach's Verbal Protocols of Reading: The Nature of Constructively Responsive Reading (1995), which provides an outstanding summary of what is known about verbal reports from current research and practice.

A great deal has been learned about reading through protocol analysis despite its limitations as a research technique. Of the many studies conducted to investigate reading strategies using this technique, none, to our knowledge, has studied the use of reading strategies by native speakers of Chinese when reading English and Chinese. The purpose of this study was to investigate the reading strategies used by native speakers of Chinese while reading easy and difficult texts in both English and Chinese. Specifically, we analyzed think-aloud data to investigate (1) whether there were any differences in strategy use among adult native speakers of Chinese when they read English and Chinese, and (2) whether text difficulty affected strategy use when the subjects read in both English and Chinese.

Method

Subjects

Twenty native speakers of Chinese (15 males and 5 females) were selected for this study from a pool of approximately 200 potential candidates. The criteria for selection included willingness to participate in the study, commitment to spend a minimum of four hours for participating in all phases of the study, and evidence of adequate proficiency in both English and Chinese. Proficiency in English was demonstrated if the subjects had obtained a score of 550 or higher on the Test of English as a Foreign Language (TOEFL), while proficiency in Chinese was determined by the level of educational experience completed in the People's Republic of China which included at least a bachelor's degree. This information was verified by consulting available academic records and meeting privately with individual subjects about their proficiency in English and Chinese. As Table 1 shows, the sample consisted of well educated, relatively English-proficient subjects with a TOEFL score average of 593.15 (SD = 26.02).

Table 1 General Information about the Subjects (Age, Sex, Major, Degrees, Years of English, Years in U.S., and TOEFL Scores)
Subject Age Sex Major Degree in China Degree in U.S. Years of English Years in U.S. TOEFL Score
LSF 36 M Chemistry B.S. Ph.D. 10 4 580
TXS 39 M History M.A. Ph.D. 15 6 610
TXJ 35 M Engineering M.S. Ph.D. 16 7 630
JZW 39 M Engineering M.S. Ph.D. 16 4 580
WCL 38 M Marketing B.A. Ed.S. 16 7 610
QZB 28 M Math B.S. Ph.D. 13 8 600
CXG 35 M Medicine M.S. M.D. 16 5 607
LLS 35 F Engineering M.S. M.S. 16 2 617
LQ 33 M Statistics B.S. Ph.D. 15 6 596
ZWM 44 M Math M.S. Ph.D. 16 5 570
CDD 38 M Engineering B.S. M.S. 15 4 557
CJ 30 F Agronomy B.S. M.S. 15 5 560
GXF 36 M Engineering M.S. Ph.D. 6 3 580
GJ 38 F Sociology M.A. Ph.D. 12 3 567
QM 30 M Microbiology B.S. Ph.D. 18 4 610
ZMC 47 M Physics B.S. M.S. 8 5 570
QSN 32 M Engineering A.S. M.S. 8 5 563
XC 29 F English M.A. M.A. 15 2 633
MZY 25 F Math B.S. Ph.D. 15 5 643
LJ 48 F Management M.S. Ph.D. 9 7 580
M 35.75 -- -- -- -- 13.5 4.85 593.15
SD 6.04 -- -- -- -- 3.42 1.67 26.02

All selected subjects were either studying or working in the United States at the time of the study. They ranged in age from 25 to 48 years old (M = 35.75; SD = 6.04) and held college degrees (either bachelor's or master's) from China. In all, 14 subjects held doctorates and 6 were completing masters or doctoral degrees at various universities in the southwestern United States. The subjects' majors included business, physics, mathematics, chemistry, engineering, business, and liberal arts. Prior to taking part in this study, the subjects had lived in the United States for periods of time ranging from 2 to 8 years (MSD= 1.67) and reported having had several years of English language experience overseas ranging from 6 to 18 years (M = 13.5; SD = 3.42).

Materials

The materials used in the study consisted of expository passages varying in difficulty levels (easy/difficult) in both English and Chinese. The passages were typical of reading materials generally found in American basal readers. The easy passages ranged from 150 to 200 words and had a predetermined 7th grade readability level while the difficult passages were approximately 250 to 350 words long with a predetermined 12th grade readability level. The English passages were selected from SRA Reading Laboratory III (Parker, 1963) which are similar in structure to those typically found in basal readers. The Chinese passages were selected from a Chinese test study guide entitled A Study Guide to the TOEFL and GRE Tests (Xie, 1991). Appendix A includes a sample of the passages used in English and Chinese.

To control for the effect of prior knowledge, an attempt was made to select passages whose topics were generally familiar to the subjects. Given that that there are no established ways of accomplishing such a goal, and after much reflection, we devised a simple way of establishing familiarity by first generating a list of familiar topics. Examples of such topics include those frequently found in mainstream print media such as newspapers, magazines, and other leisure reading materials. With the assistance of a group of judges consisting of three reading specialists, we first generated a list of topics thought to be familiar to the subjects. After much discussion and deliberation, consensus was reached to include the following topics: culture, arts, famous people, animals, and popular science. Second, we selected several passages dealing with these topics at each level of diffi-culty and consistent with the criteria for passage length and readability described above. Third, we offered the subjects the option of self-selecting the topics they wished to read about. This last step was taken as a precautionary measure to avoid a potential counfounding source of variance having to do with prior knowledge.

Following established methodological recommendations for increasing the likelihood of obtaining reasonably complete and accurate self-reports (see Ericsson & Simon, 1993 and Pressley & Afflerbach, 1995), each passage was marked with intermittent red dots, placed after one to two sentences. The markings served as a constant reminder for the subjects to verbalize their thoughts while reading.

Design and Procedures

The design involved two languages (English and Chinese) and two levels of text difficulty (easy and difficult). The subjects read two passages (one easy and one difficult) in each language and verbally reported their thoughts while reading the passages. For consistency, the think-alouds were conducted in English when reading English texts and in Chinese when reading Chinese texts. The resulting protocols were tape recorded for purposes of accurate transcription and data analysis. The think-aloud protocol was conducted in accordance with established procedures used in the most current verbal report research literature (e.g., Afflerbach & Johnson, 1984; Block, 1986, 1992; Cohen, 1996; Ericsson, 1980; Ericsson & Simon, 1993; Fawcett, 1993; Garner, 1987; Kletzien, 1991; and Pressley & Afflerbach, 1995).

Prior to conducting the think-alouds, the subjects were briefed about the purpose of the study and trained to think aloud while reading. The instructions to the subjects were intentionally kept neutral to reduce the likelihood of biasing the subjects' processing in one way or another. They were simply asked to read and say everything out loud regardless of how trivial the thinking might seem. In addition, they were to make a special effort to do so consistently throughout the passages. Since our main goal was to learn as much as possible about the processes people naturally use when they read, we did not give the subjects any specific instructions about how the text might be processed. Pressley and Afflerbach (1995) note that "researcher silence about how the text might be processed is more defensible than directions that prompt particular processes..." (pp. 132 - 133). After the procedure was explained and modeled, the subjects practiced reading passages similar to the ones used in the actual study in both English and Chinese. The practice sessions, which lasted between 20 - 30 minutes, were tape recorded so that subjects could become accustomed to the use of the tape recording device and procedures. The tape recordings were helpful in reviewing the reports and checking for the completeness and accuracy of the reports. The subjects received feedback in reaction to their reports and assistance in refining their think-alouds until they felt comfortable with the procedure. After the practice sessions, a question and answer session took place and individual appointments were scheduled for carrying out the think-alouds analyzed in the present study.

All of the data collection sessions were individually scheduled and all were conducted within a few days immediately following the training sessions so that subjects would remember how to perform think-alouds. In addition, at the beginning of each session, the subjects were reminded of the basic steps for completing the think-aloud activity. While the sessions were in process, the researchers were available to answer questions or provide assistance. The subjects were allowed to choose whether they wanted to read first in Chinese and then in English or vice versa; however, the order of text difficulty was counterbalanced across the two languages. Because the sessions were conducted one-on-one, the duration of the entire experiment varied from 2 to 3 hours per subject.

Data Collection and Analysis

The tape recorded data were transcribed for analysis using a transcription system designed to preserve features of the verbal reports including pause time and repetition. In order to identify the strategies used while reading in Chinese and English, two groups of judges, representing each of the languages, were hired to work with the authors in identifying the reading strategies used and categorizing them in a meaningful way. The first group consisted of the first author (a native speaker of Chinese) and two reading specialists (both native speakers of English) while the second group consisted of the same author and two native speakers of Chinese who were completing their doctoral degrees in English.

After the purpose of the study was explained and an example discussed, the judges were instructed to independently identify and categorize the strategies using an agreed upon method of analysis for coding the reports and developing a list of strategies used by the subjects. The judges were also provided with a few examples of strategies that are typically made by readers and are commonly reported in the verbal report research literature such as previewing, re-reading, pausing and reflecting on what is read, questioning, making comments, etc. It was agreed that these were only examples of some of the many strategies that a reader might resort to during reading and that they should carefully analyze any other types of strategies that might be used. The following live example demonstrates the use of one such strategy at the beginning of a passage:

SAMPLE PROTOCOL
A MAMMAL OR BIRD THAT WEIGHED ONLY TWO AND A HALF GRAMS WOULD STARVE TO DEATH. [I don't know why it would starve to death. Maybe this is not true]. IT WOULD BURN UP ITS FOOD TOO RAPIDLY AND WOULD NOT BE ABLE TO EAT FAST ENOUGH TO SUPPLY MORE FUEL. [The small...smaller the animals the less they eat so they don't need to eat fast, very fast to...to live...].

Note: The verbal reports appear in lower case between square brackets. The strategy is labeled accordingly by the judges.

The method of analysis consisted of first reading the protocol transcripts and marking the parts of the reports containing the strategies. Once the strategies have been identified, the judges were to assign appro-priate labels and count the number of occurrences of those strategies. At the conclusion of this task, the judges compared their findings and discussed the strategies identified. The judges' generated lists of strategies were subsequently compared and discussed relative to the type and number of strategies identified across subjects and passages. Throughout this process, whenever a disagreement among the judges occurred with respect to either strategy type or number, a discussion ensued until consensus was reached. In general, a relatively high degree of agreement was reached among the judges with respect to the type and number of strategies identified even before discussion and consensus building took place, as indicated by the Kendall's Tau (Coefficient of Concordance for English: w = .88; Chinese: w = .89). The final list of the reading strategies obtained was analyzed for differences in strategy use across languages and passage difficulty levels. The results are presented below.

Results and Discusson

The subjects' think-aloud reports were examined to ascertain whether differences in strategy use existed when the subjects read easy and difficult passages in English and Chinese. The data were analyzed using a repeated measures analysis of variance (ANOVA) with two within-subjects variables: Language (English vs. Chinese) and difficulty level (easy vs. difficult). The dependent variables for each analysis were the average number of strategies invoked by the participants for each of the four conditions. The independent variables were language and text difficulty levels.

Overall Strategy Use. Table 2 presents an overview of the types of reading strategies used by the subjects when they read in English and Chinese. As the data show, 20 strategies were identified by the judges which could be categorized as text-based and reader-based strategies. The text-based strategies (italicized in Table 2) focused mainly on the micro- and macro-structural aspects of text including using syntax (e.g., "I put `taking' because it had to be a verb"), using context clues (e.g., "The following phrase gives the definition of the word gnomon"), and recognizing text structure (e.g., "I think the first is the statement and the author gives example in the second and third sentences"). The reader-based strategies focused primarily on the reader's reactions to text content including making predictions (e.g., "I guess the story will talk about the Chinese religion"), confirming (e.g., "I agree. That's correct"), and monitoring (e.g., "I am not quite clear what `wurning up' means"). These strategies seem to reflect the interactive nature of the reading process. To make sense of text, readers construct meaning by interacting with the reading materials using text-based and reader-based strategies.

Table 2 Strategies Used by the Subjects White Reading Easy and Difficult Texts in Chinese and English
Strategy Description Sample Responses
Adjusting Rate The response indicates that the reader is in control of reading and adjusts reading rate based on text difficulty. "This part is complicated and I have to read it very slowly."
"I need to go back to the first paragraph again."
Using Context Clues The subject uses context to understand text information. "The following phrase gives the definition of the word - 'gnomon.'"
Visualizing Information The subject makes a mental image of what he/she is reading. "I am thinking the time when I was on the plane changing my watch."
Using Key Words The subject reasons with text information using key vocabulary. "I am looking for the meaning of the word chariot in the folowing sentences."
Making Predictions The subject predicts what will happen in succeeding portions of the text. "I guess the story will talk about Chinese religion."
Monitoring Comprehenson The subject assesses his or here degree of understanding of the text, and attempts to make necessary repairs. "Now I see what it means."
"I am not quite clear what 'burning up' means."
Using Syntax The subject uses knowledge of language to make sense of printed information. "No, this is not something about philosophy as I thought earlier. It talks about science."
Translating The subject switches language codes to confirm understanding. (The subject translates from one language to another).
Integrating Information The subject connects new information with previously stated content. "This must have something to do with 'Moses' mentioned at the beginning of the passage."
Questioning Information The subject questions the significance of content read. "Why secondary important?"
"What is the first importance?"
Commenting The subject makes comments and/or evaluates content. "I don't think it is dragon. The old people said it was dog".
Reacting to Text Information The subject reacts emotionally to information in the text. "It is interesting."
Using Familiar Phrases The subject recognizes the use of previously known phrases. "Nine layers of the sky were supported by four columns."
"That's a phrase you hear from the legends."
Using Text Structure The subject demonstrates awareness of text structure. "I think the first is the statement and the author gives example in the second and third sentences."
Confirming Information The subject confirms or refutes understanding of text information. "I agree."
"That's correct."
Using Prior Knowledge The subject demonstrates prior knowledge and experience about content read. "I am familiar with it. Several years ago when I came to the U.S. we set our watch several times."
Using Main Ideas The subject identifies and uses main idea as a guide for understanding text information. "The main idea of the text is about Chinese religion and science."
Interpreting Information The subject assesses and interprets information in text. "This came to teh concluson of how to determine the rate at which animal grows up."
Paraphrasing Information The subject substitutes his own words for the original wording of the text. "The bird would die of hunger" substituted for "The bird would starve to death."

Before comparing individual strategies across languages and difficulty levels, the average number of strategies used per condition was computed to give the reader an overall sense of group differences. The typical number of strategies invoked overall by the participants for each of the four conditions shows that overall, for the Chinese passages, readers, on average, invoked strategies approximately 20 times regardless of the difficulty of the passage, whereas for English passages, readers invoked an average of 27.3 strategies for the easy passage and 36.15 strategies for the difficult passage (see overall means in Table 3). These data suggest that where differences in strategy use existed, more strategies were used in English than in Chinese, and more strategies were used for difficult than for easy passages in English.

Table 3 Strategy Use by Language and by Text Difficulty Level
Strategy Chinese-Easy Chinese-Difficult English-Easy English-Difficult Univariable Analysis
Mean (SD) Mean (SD) Mean (SD) Mean (SD) F(1,79) Pr > F
Adjusting Reading Rate .30 (.57) .30 (.47) 1.50 (1.32) 2.40 (2.64) 8.15 .055*
Using context Clues .00 (.00) .10 (.31) .05 (.22) .90 (1.02) 34.62 .0001*
Visualizing Information .35 (.81) .10 (.31) .45 (.83) 1.20 (1.24) 41.04 .0001*
Using Key Words .10 (.31) .45 (.69) .50 (1.15) 2.40 (1.67) 36.45 .0001*
Making Predictions 1.10 (1.07) .80 (1.15) .70 (.80) 2.10 (1.94) 41.52 .0001*
Monitoring Comprehension 1.35 (1.57) 2.50 (2.46) 3.10 (2.17) 5.40 (3.63) 4.89 .0299*
Using Syntax .00 (.00) .10 (.31) .00 (.00) .40 (.82) 10.77 .0015*
Making corrections .25 (.44) .15 (.49) .30 (.57) .50 (.089) 7.42 .0079*
Translating .85 (1.46) .75 (1.52) .05 (.22) .55 (.89) 7.42 .0079*
Integrating Information 1.95 (1.43) 1.05 (1.00) 1.55 (1.64) 1.60 (1.73) 9.79 .0025*
Questioning Information 1.50 (1.47) .55 (.95) 1.30 (1.03) 1.45 (1.43) 16.44 .0001*
Commenting 3.25 (1.92) 2.00 (1.84) 2.15 (2.28) 2.30 (1.95) 8.24 .0053*
Reacting to Text Information .65 (.88) .35 (.67) .55 (.95) .60 (1.05) 4.03 .0482*
Using Familiar Phrases .05 (.22) 1.10 (1.29) .70 (1.59) 1.00 (1.21) 9.13 .0034*
Using Text Structure 1.25 (1.59) 1.60 (1.79) 2.25 (2.24) 1.30 (1.17) 17.55 .0001*
Confirming Information .15 (.37) .05 (.22) 1.30 (1.89) .30 (.66) 20.37 .0001*
Using Prior Knowledge 1.30 (1.34) 1.30 (1.03) 3.60 (3.78) 3.75 (2.36) .070 .7912
Using Main Ideas 1.75 (2.07) 2.05 (1.70) 2.10 (2.13) 1.95 (1.64) 1.15 .2862
Interpreting Information 3.70 (2.27) 4.20 (2.95) 3.80 (2.71) 4.15 (3.01) .120 .7248
Paraphrasing Information .40 (.82) .95 (1.10) 1.35 (1.35) 1.90 (1.89) .030 .8573
Overall Mean 20.25 20.45 27.30 36.15
The means represent the average number of strategies used per condition
* Significant Main Effects

Interaction of Language with Text Difficulty. Table 3 displays the means, standard deviations, and F-tests for the interaction of language and text difficulty. The results revealed significant interaction effects for sixteen of the twenty strategies invoked by the subjects when they read easy and difficult passages in English and Chinese. These significant interactions follow several different patterns as shown in Table 3. Overall, strategy use generally increased as a function of text difficulty for English passages and either increased, stayed the same, or decreased for Chinese passages.

Strategy Use by Language. Table 4 shows the results obtained when strategy use by language was examined. Significant differences were found for two of the four strategies with no interaction, namely: "using prior knowledge", F(1,79) = 73.70, p << .0001; and "paraphrasing", F(1,79) = 45.81, p << .0001. In both cases, the average number of times these strategies were used was, indeed, significantly higher in English than in Chinese.

Table 4 Strategy Use by Language
Strategy Chinese English Univariable Analysis
Mean (SD) Mean (SD) F(1,79) Pr > F
Adjusting Reading Rate .30 (.51) 1.95 (2.11) 96.77 .0001*
Using context Clues .05 (.22) .47 (.84) 51.48 .0001*
Visualizing Information .22 (.61) .82(1.10) 41.59 .0001*
Using Key Words .27 (.055) 1.45 (1.78) 87.56 .0001*
Making Predictions .95 (1.10) 1.40 (1.62) 13.01 .0005*
Monitoring Comprehension 1.92 (2.11) 4.25 (3.17) 53.72 .0001*
Using Syntax .05 (.22) .20 (.60) 10.77 .0015*
Making corrections .20 (.46) .40 (.74) 4.93 .0292*
Translating .80 (1.47) .30 (.68) 15.41 .0002*
Integrating Information 1.50 (1.30) 1.57 (1.66) .030 .6893
Questioning Information 1.02 (1.31) 1.37 (1.23) 6.38 .0136*
Commenting 2.62 (1.95) 2.22 (2.09) 5.56 .0209*
Reacting to Text Information .50 (.78) .57 (.98) .060 .4401
Using Familiar Phrases .57 (1.06) .85 (1.33) 3.91 .0516
Using Text Structure 1.42 (1.67) 1.77 (1.83) 3.49 .0653
Confirming Information .10 (.30) .80 (1.48) 27.61 .0001*
Using Prior Knowledge 1.30 (1.18) 3.67 (2.87) 73.70 .0001*
Using Main Ideas 1.90 (1.87) 2.02 (1.87) .14 .7076
Interpreting Information 3.95 (2.61) 3.97 (2.83) .05 .8216
Paraphrasing Information .67 (.99) 1.62 (1.64) 45.81 .0001*
Overall Mean 20.31 31.67
The means represent the average number of strategies used per condition
* Significant Main Effects

The results of this analysis show how strategy choice is affected by the language medium used. Of the sixteen strategies with a significant interaction between language and difficulty level, fourteen showed greater use of strategies in English than in Chinese. The subjects' reliance on these strategies when reading English passages appears to be consistent with prior L2 research (e.g., Block, 1986, 1992; Carrell, 1991; Clarke, 1979, 1980; Cziko, 1980), which has previously found that reading ability and strategy use tend to be dependent on language proficiency for many second language readers. Research indicates that most L2 readers (including some of our subjects) are likely to encounter difficulties reading English due, at least in part, to their level of language proficiency. We suspect that in an effort to compensate for underlying linguistic and content difficulties, the subjects resort to greater strategy use while reading in the target language. While reading a passage in English, one reader commented, "This part is complicated, and I have to read it very slowly...I need to go back to the first paragraph again." Comprehension monitoring, according to Casanave (1988) is of particular importance to some L2 readers. The process tends to operate rather automatically, particularly for the most proficient readers. Baker and Brown (1984) note that such monitoring is usually not readily observable until some triggering event -- confusion or failure to comprehend, occurs. Indeed, the readers in this study demonstrated awareness of strategy use, and achieved great success at detecting and "fixing" many of the inconsistencies they encountered while reading.

Strategy Use by Text Difficulty Level. Table 5 shows the results of overall strategy use while reading easy and difficult texts in English and Chinese. The results revealed significant differences for two of the four strategies which did not show a significant interaction effect, namely interpreting, F(1,79) = 4.63, p << .0345; and paraphrasing, F(1,79) = 19.26, p << .0001. In both of these cases, strategy use was significantly greater for difficult than for easy passages.

Table 5 Strategy Use by Text Difficulty Level
Strategy Easy Difficult Univariable Analysis
Mean (SD) Mean (SD) F(1,79) Pr > F
Adjusting Reading Rate .90 (1.17) 1.35 (2.15) 9.03 .0036*
Using context Clues 1.75 (1.98) 1.45 (1.50) 4.46 .0380*
Visualizing Information .40 (.81) .65(1.05) 13.39 .0005*
Using Key Words .30 (.85) 1.43 (1.60) 95.74 .0001*
Making Predictions .90 (.96) 1.45 (1.71) 15.21 .0002*
Monitoring Comprehension 2.23 (2.07) 3.95 (3.40) 42.28 .0001*
Using Syntax .00 (.00) .25 (.63) 24.69 .0001*
Making corrections .28 (.51) .33 (.73) .630 .4280
Translating .45 (1.11) .65 (1.23) 2.90 .0927
Integrating Information 1.75 (1.53) 1.33 (1.42) 6.41 .0133*
Questioning Information 1.40 (1.26) 1.00 (1.28) 15.28 .0002*
Commenting 2.70 (2.15) 2.15 (1.87) 6.85 .0106*
Reacting to Text Information .60 (.90) .48 (.88) 2.01 .1607
Using Familiar Phrases .38 (1.27) 1.05(1.15) 23.33 .0001*
Using Text Structure .73 (1.47) .18 (.50) 33.21 .0001*
Confirming Information .03 (.16) .50 (.85) 53.25 .0001*
Using Prior Knowledge 2.45 (2.79) 2.53 (2.18) .110 .7444
Using Main Ideas 1.93 (2.08) 2.00 (1.65) .700 .4039
Interpreting Information 3.75 (2.47) 4.18 (2.94) 4.63 .0345*
Paraphrasing Information .88 (1.20) 1.43 (1.60) 19.26 .0001*
Overall Mean 23.81 28.34
The means represent the average number of strategies used per condition
* Significant Main Effects

Of the sixteen strategies which showed a significant interaction between language and difficulty level, ten were used more frequently for difficult than for easy passages (see Table 3). This finding is not too surprising in light of the fact that difficult texts are more linguistically and cognitively demanding than easy texts. Indeed, the subjects invoked a variety of monitoring and "fix-up" strategies in an attempt to control and regulate their comprehension processes. They also tended to use certain strategies flexibly and adjust their use to the difficulty level of the passage they read. One reader, who recognized the use of a familiar phrase while reading, commented, "That's a phrase you hear from the legends." When reading a difficult text in English, another reader used context clues as a strategy to understand the meaning of a word and said, "The following phrase gives the meaning of the word `gnomon.'" Overall, the results make good common sense: Subjects resorted more often to explicit strategy use when reading English and when reading difficult texts. These findings are consistent with prior research by Pressley and Afflerbach (1995) who have stated that "active and strategic efforts at meaning construction only occur in reaction to more challenging texts" (p. 14) resulting in "substantial verbalization of information not explicitly given in the text" (Ericsson & Simon, cited in Pressley & Afflerbach, 1995, p. 13).

Conclusion

The results of this study have shown that the subjects demonstrated awareness of a variety of reading strategies while reading English and Chinese. The overall pattern of strategy use in their think-aloud reports suggest that where differences existed, reading strategies were used more frequently in English than in Chinese, and more frequently for difficult texts than for easy texts. In the majority of cases, the presence of significant interaction effects simply reflects the fact that for the Chinese passages, the number of strategies used did not dramatically change between the easy and difficult passages, while for English, strategy use generally increased when going from easy to difficult passages. However, while the majority of the strategies identified were used by the subjects in both English and Chinese, it is unclear whether the strategies learned in one language may operate alongside those learned in a second language, as previously maintained by L2 researchers (e.g., Carrell, 1991; Cohen, 1996; Cziko, 1980). Whether the use of specific strategies can be transferred across languages or originate from one language or another is an important research question which should be systematically investigated.

The results obtained provide additional research support to the findings of several L1 and L2 researchers (e.g., Afflerbach, 1990; Baker & Brown, 1984; Block, 1992; Carrell, 1991; Clarke, 1979, 1980; Cziko, 1980; Kletzien, 1991; Olshavsky, 1976-1977) who have investigated the use of reading strategies by first and second language learners. These researchers, and others, have found that good readers are more aware of the strategies they use than poor readers. Good readers also tend to invoke strategies more flexibly, adjust strategy use to text type and purpose for reading, (e.g., Olshavsky, 1976-1977; Olson, Duffy, & Mack, 1984), are able to successfully detect and correct inconsistencies encountered while reading (e.g., Garner, 1987), and are more careful at monitoring their comprehension processes (Baumann, Jones, Sieffert-Kessel, 1993; Block, 1992; Wade, 1990). However, while the readers in this study appear to fit the profile of "good readers" described in most L2 research, we should stress the fact that the subjects in this study were generally well educated, relatively fluent speakers of English. Because of the type of participants used in this study, no definitive conclusions can be made with respect to strategy use by less proficient L2 readers.

Finally, consistent with prior research, it appears that the process thinking aloud while reading increases one's awareness of oneself as a reader and of how one interacts with text information (Cohen, 1996) as demonstrated by the subjects in this study. Becoming aware of the strategies one uses while reading can lead to positive outcomes for first and second language readers. Such information should allow readers to gain insights into their reading strengths and weaknesses and to plan appropriate instructional intervention. However, while the ultimate goal of studying reading strategies is an applied one, it is difficult to speculate about the teacher's role in helping students become strategic readers. We concur with Gu (1996) who has called for further study of the effects of strategy training on first and second language reading, and for teachers to become aware of the importance of reading strategies before such training can be incorporated in the classroom.

References

Afflerbach, P. (1990). The influence of prior knowledge on expert reader's main idea construction strategies. Reading Research Quarterly, 25, 31-46.

Afflerbach, P., & Johnson, P. (1984). Research methodology on the use of verbal reports in reading research. Journal of Reading Behavior, 16, 307-321.

Alderson, J. C. (1984). Reading in a foreign language: A reading problem or a language problem? In J. C. Alderson & A. H. Urquhart (Eds.), Reading in a foreign language (pp. 1-24). New York: Longman.

Baker, L., & Brown, A. (1984). Metacognitive skills and reading. In P.D. Pearson (Ed.), Handbook of reading research (pp. 353-394). New York: Longman.

Baumann, J. F., Jones, L. A., & Seifert-Kessell, N. (1993). Using think alouds to enhance children's comprehension monitoring abilities. The Reading Teacher, 47, 184-193.

Block, E. (1986). The comprehension strategies of second language readers. TESOL Quarterly, 26, 463-494.

Block, E. (1992). See how they read: Comprehension monitoring of native and second language readers. TESOL Quarterly, 26, 319-343.

Carrell, P. L. (1984). The effects of rhetorical organization on ESL readers. TESOL Quarterly, 18, 441-469.

Carrell, P. L. (1991). Second language reading: Reading ability or language proficiency? Applied Linguistics, 12, 159-179.

Casanave, C. P. (1988). Comprehension monitoring in ESL reading: A neglected essential. TESOL Quarterly, 22(2), 283-302.

Clarke, M. (1979). Reading in Spanish and English: Evidence from adult ESL students. Applied Language Learning, 29, 121-150.

Clarke, M. (1980). The short-circuit hypothesis of ESL reading -- or when language competence interferes with reading performance. Modern Language Journal, 64(2), 203-209.

Cohen, A. (1996). Verbal reports as a source of insights into second language learner strategies. Applied language Learning, 7(1 & 2), 5-24.

Cziko, G. A. (1980). Language competence and reading strategies: A comparison of first-and second-language oral reading errors. Language Learning, 30, 101-114.

Ericsson, K. A. (1980). Verbal reports as data. Psychological Review, 87, 215-251.

Ericsson, K. A. & Simon, H., A. (1993). Protocol analysis: Verbal reports as data. Cambridge, MA: MIT Press.

Fawcett, G. (1993). Using students as think-aloud models. Reading Research and Instruction, 33, 95-104

Field, M. L. (1984). A psycholinguistic model of the Chinese ESL reader. In On TESOL '84. A brave new world for TESOL. Selected papers from the 18th Annual TESOL Convention Houston, Texas, March 6-11.

Garner, P. (1987). Metacognition and reading comprehension. Norwood, New Jersey: Ablex.

Gu, P.Y. (1996). Robin Hood in SLA: What has the learning strategy researcher taught us? Asian Journal of English Language Teaching, 6, 1-29.

Kletzien, S. B. (1991). Strategy use by good and poor comprehenders reading expository text of differing levels. Reading Research Quarterly, 26, 67-86.

Olshavsky, J. E. (1976-1977). Reading as problem solving: An investigation of strategies. Reading Research Quarterly, 12, 654-675.

Olson, G. M., Duffy, S. A., & Mack, R. I., (1984). Thinking-out-aloud as a method for studying real-time comprehension processes. In D. Kieras & M.A. Just (Eds.), New methods in reading comprehension research (pp. 253-286). Hillsdale, NJ: Lawrence Erlbaum Associates.

Parker, D. (1963). SRA Reading Laboratory III B. Chicago, IL: Science Research Associates, Inc.

Pressley, M. & Afflerbach, P. (1995). Verbal reports of reading: The nature of constructively responsive reading. Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Smith, F. (1994). Understanding Reading. New York: Holt, Reinhart, & Winston.

Wade, S. (1990). Using think-alouds to assess comprehension. The Reading Teacher, 43(2), 442-451.

Xie, Z. Y. (1991). A Study guide to the TOEFL and GRE. Fuxinmenwai, Beijing: Ocean Publishing.

Appendix A
ENGLISH PASSAGE
EASY VERSION
Every animal is a living radiator. Heat formed in its cells is given off through its skin. Warm-blooded animals maintain a steady temperature by constantly replacing lost surface heat. Smaller animals, which have more skin for every ounce of body weight, must produce heat faster than bigger ones. Because smaller animals burn fuel faster, scientists say they live faster.

The speed at which an animal lives is determined by measuring the rate at which it uses oxygen. A chicken, for example, uses one-half cubic centimeter of oxygen every hour of each gram it weighs. The tiny shrew uses four cubic centimeters of oxygen every hour for each gram it weighs. Because it uses oxygen eight times as fast, it is said that the mouse-like shrew is living eight times as fast as the chicken. The smallest of the warm-blooded creatures, the hummingbird, lives a hundred times as fast as an elephant.

There is no limit to how small a warm-blooded animal can be. A mammal or bird that weighed only two and a half grams would starve to death. It would burn up its food too rapidly and would not be able to eat fast enough to supply more fuel.

ENGLISH PASSAGE
DIFFICULT VERSION
The Chinese of 3500 years ago believed that the earth was a chariot, and the sky a curved canopy stretched above it. The canopy was nine layers thick, and it sloped slightly to the northwest, as a cataclysm had broken one of its supporting columns. This gentle slope explained the movement of the stars from east to west.

According to these ancient Chinese beliefs, the sun spent the night on earth and ascended to the sky each morning from the luminous valley of the east by climbing the branches of an immensely tall sacred tree. To the Chinese people, the sun was the incarnation of goodness, beauty, and truth. In popular imagination, the sun was represented as a cock that little by little assumed human form. His battles with the dragons, which personified evil in their beliefs, accounted for the momentary disappearances of the sun that men now call eclipses. Many of the Chinese people worshiped the sun, but in the vast and complicated organization of the Chinese gods, the sun was of only secondary importance.

Along with these unsophisticated beliefs about the sun, the Chinese evolved a science of astronomy based upon observation -- through essentially religious -- which enabled them to predict the eclipses of the sun and the movement of the stars. Such predictions were based on calculations made by using a gnomon -- an object whose shadow could be used as a measure, as with a sundial or simpler shadow pointers. Moreover, with the naked eye, the Chinese observed sunspots, a phenomenon not then known to their contemporaries.

CHINESE PASSAGE
EASY VERSION
摩西祖母已躋身於20世紀美國最著名的畫家行列, 然而她是70多歲以後才開始作畫的, 如同她曾說過的, "我絕不坐回到搖掎里, 等著什麼人來幫助我," 無人能擁有一個 (比她) 更多產的老年時期.

安娜.瑪麗.羅伯遜出生在紐約州的一個家庭, 是 (家庭中) 5男5女中的一個 ("我們像小夢卜一樣成串地出生") . 她12歲時離開家去當家庭幫工, 27歲時她和雇主的助手托馬斯.摩西結婚. 他們一生中絕大部分時間都在家, 先在弗吉尼州, 後來在紐約州的英格布里奇. 她生了10個孩子, 其中的5個活了下來, 她的丈夫在1927年去世.

摩西祖母在孩提時期很少畫畫, 作為愛好, 她給制繡花圖案, 直到老年, 手變得太僵硬無法縫制 (東西) 了, 她才轉移到油畫上來. 她樂意忙著趕點活兒來打發時間, 她的畫開始是在當地一家葯店和定期集市上出售, 不久 (她的畫) 開始為一位商人所注意, 他購買她畫的每一件作品, 其中的3幅在現代藝術博物館展出. 1940年, 她在紐約舉辦第一次個人畫展. 從1930年直到她去世為止, 她創作了2000多幅畫, 以詳細而充滿活力 (的筆觸) 伴之以奇跡般的色彩意識和形式, 去描繪她多年來所熟知的田園生活.

她說 : "我構思很艱難, 只有我想起一些真正美麗的事物, 我才開始作畫."

CHINESE PASSAGE
DIFFICULT VERSION
早期統計學方法的發展受到兩種大相徑庭的影響. 統計學有一位 "母親", 她用於保存政府單位有條理的紀錄 (State 和 Statistics 來源於相同的拉丁語詞根Status), 和一位彬彬有禮的 "父親", 他依靠數學計算, 測量, 描述, 列表, 整理及摘要, 所有這些導致了現代描述統計學 (的出現). 從這位 "父親" 的影響中產生了現代推論統計學, 它堅實地基於概率原理之上.

描述統計學包括資料的列表, 敘述和制圖集成. 這些資料可以是數學變量, 諸如高度的測量, 智力或等級水平, 由基礎的連續統一體表示其特性, 或者可以是性質的變量, 諸如性別,學院或人格類型. 大量的資料在能夠被領會之前, 通常需要經過一個摘要或縮減的過程, 描述統計學是用於描述,概括或縮減浩瀚資料使之成為可理解形式的一種工具.

推論統計學是一種解決其他種類問題方法的定形主體, 這些問題給人們無助的頭腦帶來極多的困難, 這類綜合的問題包括使用觀察例証去作出預的. 例如, 一位學校主管人想在一個大範圍的學校系統內確定來吃早餐的學生比例, 或是接種了流感疫苗學生的比例, 以及其他諸如此類的事; (如果) 有一些統計學知識, (那麼) 這位主管人就會知道, 去查問每個學生是不必要和低效率的, 全部管區的比率能夠從一次100個孩子的典型調查中相當精確地估算出來. 這樣, 推論統計學的意圖是通過部分人口調查的特點信息去預全體居民的特點.

Xiwu Feng received his Ed.D. degree from Oklahoma State University in December 1995, and is currently an assistant professor in the Communications Skills Department at Laguardia Community College-CUNY, where he teaches college reading, writing, and English as a second language.

Kouider Mokhtari, Ph.D., is associate professor of Curriculum and Instruction at Oklahoma State University, where he teaches graduate and undergraduate reading education courses. His research interests focus on the acquisition of language and literacy by first and second language learners.


1. Some of the subjects had developed their English language abilities beyond the reported TOEFL scores, having resided in the US for several years and having earned advanced degrees in English. Adequate proficiency in the language was important to help ensure complete and accurate verbal reports.

2. Consistent with methodological recommendations from prior protocol analysis research (e.g., Ericsson & Simon, 1993), an attempt was made to limit all analyses to approximately the same text length (approximately 150 words per passage). This was done in order to control for passage length and therefore avoid a possible confound between passage length and number of strategies used.

3. The passage difficulty levels were predetermined by the authors of the passages using established readability formulas for English and Chinese.

4. We found that most of the strategies identified had occurred at least 3 times across passage and subjects.


HOME CONTENTS ARTICLES REPORTS REVIEWS