Claims that extensive reading could lead to significant improvements in learner's reading speeds date back thirty years, and the role of graded readers in programs to promote such reading has an even longer history. Studies that measure reading speeds have been relatively few and far between however, and those that do exist rarely evaluate reading speed in relation to the effect of different classroom methodologies in the teaching of reading. Early work on reading speed tended to focus on the development of techniques to help learners to read faster, and failed to recognize the importance of varying the speed according to the reader's purpose in approaching a text. Such techniques as have been employed on speed reading courses also tend to cause readers to suffer lower levels of reading comprehension. The study reported in this article was conducted in the Yemen Arab Republic on young adult students working in various government ministries. It measured both reading speeds and comprehension in two groups of learners exposed to "intensive" and "extensive" reading programs respectively. The "extensive" group was exposed to a regime of graded readers while the "intensive" group studied short texts followed by comprehension questions. Results indicate that subjects exposed to "extensive" reading achieved both significantly faster reading speeds and significantly higher scores on measures of reading comprehension.

Introduction

A widely recognized problem faced by learners throughout the ESL/EFL world is that of slow reading (Hamp-Lyons 1983; Cooper 1984). This has often been linked with classroom methodology in reading lessons, particularly where such lessons focus on language development rather than reading per se. A number of specialists have commented on a common practice in intensive reading lessons, where texts are often treated as vehicles for the presentation, practice, manipulation, and consolidation of language points, rather than the encouragement of reading itself (Nuttall 1982:20; Alderson & Urquhart 1984:246-247; Bartram & Parry 1989:7; Hyland 1990:14; Susser & Robb 1990:161-162). Slow reading as a problem for learners has been defined by Brown & Hirst (1983:140) as a "weakness independent of the purpose of reading", involving the processing of information at such a slow rate that the reader is unable to hold enough detail in short-term memory to permit decoding of the overall message of the text. In this context it should be noted that different reading purposes require different reading speeds, and though the development of adequate reading speed should receive a high priority in our learning programs, we should also recall, as Nuttall (1982) does that reading speed without comprehension is worthless. Therefore, the current study seeks to locate the twin issues of reading speed and reading comprehension within the framework of a comparison of intensive and extensive classroom reading procedures.

Background Studies

The use of graded readers dates back to the time of the writings of Michael West in the 1950s, and it was in the 1960s that interest in reading speed gained momentum through the writings of Fry (1963) and De Leeuw & De Leeuw (1965). Fry claimed that good readers achieve a speed of 350 words per minute, fair readers 250 words, and slow readers reach 150 words per minute. De Leeuw cited 230-250 words per minute as an average initial speed for the general public. These early insights led to growth in the development of speed reading courses, and to the belief that individuals requiring to read faster could be trained to do so through the use of paper-based techniques, and also by way of technological aids such as metronomes, and projectors or reading machines. Maddox (1963:85) criticized the use of such machines claiming that mechanical devices are "in no way superior to the method of timed practice", and Bright & McGregor (1970:96) wrote in similar vein that it is "the gimmick that stimulates interest and not the practice itself". Data-based evaluations of reading speed courses were offered by Hill (1981), and Richard (1982). Hill examined a course in 'rapid' or 'effective' reading with advanced students at the university of Leuven in Belgium. Using the program developed by De Leuw & De Leeuw (op cit), he showed that his advanced students could achieve an average increase in their reading speeds of 57% over a three year period. In terms of the speed categories used in the course, an average student therefore progressed from the 'slow' band (200 words per minute) through to the 'medium fast' category (314 wpm). Some of his subjects reached speeds of 600 words per minute or better leading him to claim that "students and others who read extensively for professional purposes should aim to cover routine material at speeds between 300 and 600 words per minute" (Hill 1981:271). Richard (1982), working in Japan, compared the reading speeds of students using traditional paper exercises with those using a reading machine (projector), and found that speeds in the latter group increased significantly more than in the former (p < .05).

Two areas of enquiry into reading speed seem to be suggested by the above. Comparisons of courses using traditional speed reading methodologies with programs emphasizing reading in quantity would appear to be the first. The second area would examine extensive reading and intensive reading and compare their relative effectiveness in developing basic reading speed. On the second area of enquiry, it is important to note that a number of researchers have warned of the possible negative consequences of intensive reading on reading speeds (Light 1970; Hamp-Lyons 1983; Cooper 1984; Kerecuk & Velloso Ribeiro 1984; Hino 1988; Brusch 1991). On extensive reading however, it was first claimed by Light (1970:122) that such reading would not only raise reading speeds, but importantly would reduce the negative affective consequences of slow, text-based, intensive approaches. More recently, Williams (1984:96) has argued for extensive reading as a way to develop adequate general reading speed, and Hill (1986:17) calls for the provision of class sets of graded readers as a means to the same end.

The next issue to be examined is the relationship between reading speed and reading comprehension. While it is generally argued that the two are closely related (Broughton et al 1978; Nuttall op cit; Champeau de Lopez 1993), the precise nature of the link between them has been the focus of an on-going debate lasting more than half a century. They may be completely independent, or correlated, or cause and effect. What is thought clear however, is that a very slow reader is more likely to read with little understanding, as his memory is taxed by the inability to retain information in sufficiently large chunks to progress through a text with adequate retention of the content in the message. Before he reaches the end of a page, or even of a sentence, he has forgotten the beginning. Champeau de Lopez (1993:50/51) makes the useful distinction between 'timed readings', in which learners read at their own pace and then calculate their speeds in words per minute, and 'paced readings' where the teacher controls the time allowed and taps on the desk to indicate times when a certain marked place in the text should be reached. In her study, carried out in Venezuela, she found that students increased their reading speeds on average from 120 to 170 words per minute ( a 50% increase), after following a course based on a combination of timed and paced readings. However, she also noted a slight drop in comprehension over the same period, from 78% to 67%. This reminds us of the danger referred to earlier, of developing reading speed at the expense of comprehension (Berkoff 1979; Nuttall op cit). Coady's advice (1979:12) on this point appears salutary "....... comprehension is achieved by reading neither too fast nor too slow". In line with these warnings, Lai (1993), in a study carried out on students in Hong Kong secondary schools, found that although subjects' gains in reading speed were significant, gains in reading comprehension were not.

A more comprehensive review of studies into extensive reading will not be attempted here, as there are excellent reviews already in print (Susser & Robb 1990; Day & Bamford 1998). Numerous studies have measured reading comprehension, as these reviews indicate, but few of them have compared extensive reading with other classroom approaches to reading. Elley & Manghubai's (1983) book flood project remains by far the most convincing evidence of the value of reading books for pleasure and in quantity. Indeed, Anderson, Wilson & Fielding's (1988) study on fifth graders seems to confirm that gains in reading speed and comprehension appear to be most closely related to the number of books read. Growth in reading proficiency generally may be a function not only of reading interesting material for pleasure, but of doing so in quantity by reading a large number of books. As already indicated, few studies have actually related classroom reading methodology to the variables of reading speed and comprehension. One such study was Robb & Susser (1989). They compared extensive reading with a 'skills-building' approach and found that subjects in the former group made significantly greater gains in reading speeds and on some of their measures of reading comprehension. Measurements made on 'getting the main idea' and 'making inferences' did not, however, reach significant levels. Both this study, and those reviewed above seem to suggest that gains in reading speed may be easier to accomplish than advances in reading comprehension, and therefore that the former objective should not be prioritized at the expense of the latter, if we wish to serve the interests of ESL/EFL learners in reading development and improvement.

Method

Two groups of elementary level learners (n = 26) at the British Council English Language Centre in Sana'a, Yemen were exposed to differing reading programs. The experimental group (n= 14) received an extensive reading program consisting of class readers, a class library of books for students to borrow, and regular visits to the library providing access to a much larger collection of graded readers (up to 2000 titles). The study extended over a period of two semesters and the reading program covered one quarter of the total class time (36 out of 144 hours). The extensive reading program is described in detail in Bell (1998). An inventory of readers was compiled, and reading records maintained to record titles read and the time subjects spent reading each week. The control group (n = 12) received an entirely different reading program which was intensive in character, being based on the reading of short passages and the completion of tasks designed to 'milk' the texts for grammar, lexis, and rhetorical patterns. For this purpose, the control group made a detailed study of the title 'Basic Comprehension Passages' by Donn Byrne (Longman:1986). This text contains thirty short texts of around 300 words each, arranged in groups of ten, together with a wide variety of exercise types for intensive exploitation of the passages. These exercises included a variety of standard reading comprehension questions, referential questions, cloze, gap-fills, multiple choice and true/false items, and guided composition, together with word building exercises and dictation passages following each unit. The aim of the approach was to recycle and reinforce language items through intensive microlinguistic analysis of the texts. Taken in sequence, the units provided a carefully structured and graded course in reading comprehension. Subjects in the 'intensive' or control group were directed to read these passages and complete the accompanying exercises in a series of homework assignments over the duration of the study. For both subject groups therefore, a great deal of reading was done, both in class and for homework. Records of time spent reading each week were maintained for both classes, and there was no significant difference in the time the two classes spent on their reading (t = 0.32). The reading done by both groups was carefully monitored by checking on homework assignments, requiring subjects to write book reports and to give oral presentations in class on the books they had read.

For research purposes reading speed was defined as 'speeds measured in words per minute on selected texts at a level appropriate to the learners'. Likewise, reading comprehension was defined as 'scores on a test of reading comprehension with three texts accompanied by questions containing modified cloze, true/false, and multiple choice items respectively'. Two texts were selected for the measurement of reading speed (see Appendix 1) based on considerations of teacher preference, readability evaluations (Fry 1977), and the length of the texts. For the measurement of reading comprehension, three texts were selected (see Appendix 2), again with reference to teacher's views, length of texts, and readability measurements. In measuring reading comprehension, it was felt important to include a range of task types and test items, so as to achieve validity and reliability in the tests designed and selected. In reporting final results for reading comprehension it was decided to adjust the weightings of the three components of the reading comprehension test in order to reduce the effect of the multiple choice and true/false items on the overall result. This was because of the negative effect of guessing on the reliability of these test types (Weir 1990:43-44).

The following hypothesis were tested in the study:
Learners in the 'extensive' group will achieve significantly faster reading speeds than those in the 'intensive' group as measured on relatively easy, non-problematic texts.

Learners in the 'extensive' group will achieve significantly higher scores on a test of reading comprehension containing texts at an appropriate level, than those in the 'intensive' group.

In measuring reading speed, subjects were first assured that the exercise would not form part of their assessment and they were then told to read at normal speed. They were given a time limit of three minutes and told that when the researcher banged on the desk they were to mark the word they had reached with a cross (x). Measurements were taken twice using selected texts (see appendix 1). Speeds were then calculated in words per minute and tabulated. For the reading comprehension tests, a time limit of 30 minutes was set for each test. Brief examples of how to complete test items were provided, examination conditions established, and the tests administered. All tests were administered prior to the study in September 1993, and again on completion in February 1994.

Results

Results are presented as both raw test scores and as gains for both reading speed and reading comprehension. Statistical comparisons were made using the 't' test for correlated samples to compare scores prior to, and after the study on the same group. To compare performance between the groups, the 't' test for independent samples was used. Results for reading speed are presented in tables 4.1 to 4.5, and for reading comprehension, results are presented in tables 4.6 to 4.13.

Discussion

The hypothesis that learners in the extensive group would achieve significantly faster reading speeds than subjects in the intensive group is very strongly supported by the data; the large and significant differences between the reading speeds of the two groups at the end of the study, the much greater gains in speed achieved by the extensive group, and the fact that the intensive group were faster readers at the start of the program (the direction of the difference being reversed by the end of the study) all support this conclusion. The first important observation on these results is that the reading program based on grade readers, to which the extensive group were exposed, has brought about much more substantial gains in basic reading speed than the traditional close reading techniques applied to the intensive group. The numerical comparisons presented in tables 4.1 to 4.5 show this to be the case. In particular, the large and significant difference on gains in reading speed strongly suggests that an extensive reading program based on graded readers is much more beneficial to the development of reading speed than traditional reading lessons based on the close study of short texts. We may conclude therefore that these results imply that reading speed will develop naturally if learners are motivated to read interesting simplified material like graded readers that are accessible linguistically. They would also appear to question the need for courses in rapid or speed reading, except perhaps for learners in EAP situations, where the need to develop faster reading speeds to cope with large volumes of material might call for more urgent surgery. These results certainly also suggest that certain task types used in intensive reading lessons (eg. referential questions, re-ordering sentences, matching, gap-filling, etc) may impede the development of learners' reading speeds, even if they do not actually bring about slower reading speeds. It is important in this regard not to overlook the fact that subjects in both groups in this study did actually increase their reading speeds.

The results of the reading comprehension tests also provide very strong support for the hypothesis that learners in the extensive group would achieve significantly higher scores than learners in the intensive group. Large and significant differences between the groups were recorded on all three tests, with the extensive group obtaining significantly higher post-study scores and large and significantly greater gains in reading speed. The results presented in tables 4.6 to 4.13 clearly confirm this. It can be concluded therefore that the extensive reading program based on graded readers has led to much greater improvement in learners' reading comprehension than traditional text-based, intensive language exploitation activities. In spite of the fact that one of the stated objectives of the material used by subjects in the intensive group was a 'close and detailed understanding of the text', it appears that this approach is much less successful in developing comprehension than providing learners with attractive, high-interest story books, which learners are well-motivated to read and understand.

Given that the gains in reading speed have been accompanied by similar large and significant differences in the performance of the two groups on the reading comprehension tests (the extensive group clearly outperforming the intensive group on both sets of tests), these results point to a powerful role for graded readers and extensive reading in stimulating reading improvement with elementary level learners. With freedom to select material according to their interests, and with associated high motivation, these learners are not only achieving substantial improvements in their reading speeds, they are importantly achieving a greater understanding of the material. In contrast with the speed reading approaches discussed in our background review, there is no retardation of reading comprehension ability when simplified and motivating reading materials are used.

A number of important limitations to these findings need to be highlighted. The first is that the number of subjects on which these results were obtained is small (a total of only 26 across the two groups). With a larger group it would have been possible to include another control group and possibly another treatment group exposed to a different variation of extensive reading. Practical realities, unfortunately, precluded such endeavors. A second important limitation is that the validity and reliability of the instruments used to measure reading speed and comprehension need to be established by correlating them with standard tests of reading comprehension. As indicated earlier, the validity and reliability of the multiple choice and true/false tests is highly questionable, which is why the reading comprehension test scores were modified as described. It is possible that the Hawthorne Effect, Halo Effect, and Subject Expectancy (Brown 1988:33-34) all exerted some influence over the results. Subjects in the extensive group were certainly aware of being involved in a separate and special reading program, and the high profile of the research could have led subjects to 'assist' the researcher by modifying the motivation of learners in the extensive group, and therefore their performance on the tests. One drawback of this type of data collection is that in order to motivate learners to read, it is necessary to discuss with them the potential benefits of such reading. This may be considered incompatible with the objective measurement of the performance variables of reading speed and comprehension. However, such limitations affect all studies of extensive reading, and there is no reason to suppose that such issues have influenced the results of this investigation more than in any other. While the above factors certainly account for part of the performance gap between the groups, they clearly cannot account for very large differences in reading speeds and comprehension test scores uncovered by this study. The measured gains in reading speeds were four times greater in the extensive group, and the gains in reading comprehension were three times greater in this group. Clearly, this data points to the importance and effectiveness of extensive reading, and to the inherent problems with intensive reading approaches in the classroom. Intensive approaches, because they focus on language manipulation rather than developing reading, tend to inhibit reading improvement among learners at low proficiency levels. Extensive reading in contrast, seems to liberate the learner from slow reading speeds, and lead to genuine comprehension of what is being read.

There has been much discussion recently about the role of extensive reading in developing automaticity of word recognition and in promoting lexical access skills. Further research will certainly be needed if we are to eventually understand exactly how simplified, pleasure-driven, high-interest materials like graded readers contribute to the development of bottom-up decoding processes. Future studies need to examine the relationship between reading speed and reading comprehension more closely. Possible questions include:

Is there an optimum speed for the processing of a particular text and the extraction of meaning from it? 2) Is there a threshold speed below which processing and comprehension becomes impossible? Longitudinal studies on the comparative effects of different methodologies on reading speeds and comprehension in a variety of ESL/EFL settings are needed, along with attempts to discover what linguistic and rhetorical features of graded materials promote the most efficient decoding by readers.

Note: Reproduction of these texts was not possible on the website, however, the following information is provided.

Appendix 1: Reading Speed Texts

1. Inspector Holt - The Bridge, John Tully, Nelson, Collins/Nelson English Library, Level Two.
2. So You Want To Change Roles, Catch 6, 1976, CATCH magazine, Mary Glasgow Publications.

Appendix 2: Reading Comprehension Texts (Selected/Adapted from)

1. Modified Cloze
Wullen, T.L., (1978). Roland and Harriet, Practice in Comprehension and English Usage. Hulton Educational.
2. Multiple Choice
Mainwaring, F.M., (1964). Looking at Life London:Longman.
3. True-False
Packer, J. (1975). Sinking For Your Supper, Current 1, CURRENT magazine, Mary Glasgow Publications.

References