EFL Extensive Reading Instruction:
Research and Procedure

Bernard Susser
Doshisha Womens Junior College

Thomas N. Robb
Kyoto Sangyo University

This article surveys the literature on extensive reading and establishes a working definition of extensive reading as a language teaching/learning procedure. It explores the main issues in extensive reading, including the role of graded readers and the transfer of L1 reading ability. A model of extensive reading is described, based on Richards and Rodgers' (1982) definition of "procedure."

1. Introduction

Reading has been the skill most emphasized in traditional FL teaching, and even today is the mainstay of EFL instruction in many countries. In Japan, for example, English instruction at the university level is usually the "intensive reading procedure," which implies close study of short passages, including syntactic, semantic, and lexical analyses and translation into the Ll to study meaning. This, as Alderson and Urquhart (1984) have argued, is not a reading but a language lesson:

Such a pedagogic practice--of focusing on the language of a text--may be justified as a language lesson, but it may very well be counterproductive as a reading lesson. Often what is known as "intensive reading" (as traditionally opposed to "extensive reading") is actually not reading at all: the lesson consists of a series of language points, using texts as points of departure. Reading texts, in other words, are sources of language exercises, rather than reading exercises. (pp. 246-247)

This view is supported by Brumfit (1984, p. 83), Hyland (1990, p. 14), Johns and Davies (1983, p. 2ff), and Yorio (1985, p. 157). See Greenwood (1988, pp. 5-9) for a dissenting view. Of course, no one would deny that language training is an essential part of any FL curriculum. It is just that this kind of lesson is not a reading lesson in the strict sense.

Today, FL/ESL/EFL reading instruction is moving increasingly, in Haas and Flower's phrase (1988, p. 169), from teaching texts to teaching readers. Specifically, we now teach learners reading skills/strategies for understanding such elements as content, textual features, rhetorical elements, and cultural background. "Skills building" emphasizes skills/strategies for text comprehension (Hamp-Lyons, 1985, p. 367; Hamp-Lyons and Proulx, 1982, p. 9; Mikulecky, 1985; Pakenham, 1984, p. 149). This has been the mainstay of L1 reading instruction in the United States, and ESL/EFL reading textbooks with words like "skills" or "strategies" in their titles are now common. Further, there has been much research on skills-based teaching procedures, including basic skills (finding the main idea, skimming, inferencing) and advanced skills (schema-building, metacognitive skills). The interested reader is referred to the bibliographies in Carrell, Devine, and Eskey, 1988; Devine, Carrell, and Eskey, 1987; and the Annual Summary of Investigations Relating to Reading, published by the International Reading Association.

A serious problem is that these so-called reading comprehension skills do not exist, or, as Rosenshine (1980) more cautiously states, "there is simply no clear evidence to support the naming of discrete skills in reading comprehension" (p. 552). Alderson and Urquhart repeated this in 1984 (p. xvii), and Alderson stated it again at TESOL '88 (1988) (see also Alderson & Lukmani, 1989; Berkoff, 1979, p. 97; Cooper, 1987, pp. 76-77; and Lee & Musumeci, 1988, pp. 175, 180). Barnett found that teaching FL students reading strategies "did not significantly improve their reading comprehension" (1988, p. 157). She calls these results "confusing," but they are consistent with the idea that skills do not exist.

If it is not meaningful to talk about discrete reading skills (and to our knowledge no one has published a refutation of Rosenshine's assertion), then what are researchers (e.g., Hosenfeld et al., 1981) and textbook authors talking about when they use the terms "skills" and "strategies" Gardner (1978) has suggested that these "skills" are better regarded as activities involving comprehension than as categories of abilities" (p. 72). In other words, when reading, we engage in activities such as recalling word meanings, inferring, drawing conclusions, and so on, but these are all aspects of the act of comprehending (i.e., reading). They cannot be separated into discrete skills, either statistically or by task-specific testing. Whatever problems there might be with "intensive reading" and "skills building" as procedures for teaching reading, they represent the mainstream of FL/ESL/EFL reading instruction today. The "extensive reading procedure," on the other hand, while often used, has attracted comparatively little research interest. As a result, we know little about either its pedagogical aspects or its effectiveness. In this paper we review the literature on extensive reading, examine its nature as a procedure for teaching ESL/EFL reading, and present a model for an extensive reading component of an ESL/EFL curriculum. While most of the specific examples are drawn from the EFL situation in Japan, but the argument applies generally.

2. The Extensive Reading Procedure

2.1 Background to Extensive Reading

There is a large body of research on L1 extensive reading, called "pleasure reading," "sustained silent reading" [SSR], or "uninterrupted sustained silent reading" [USSR] (see research cited in Krashen, 1985, p. 91; Krashen, 1988; Vaughan, 1982, p. 69). This corresponds more or less to FL/ESL/EFL extensive reading (see Bamford, 1987; Dubin & Olshtain, 1977, pp. 77ff; Grellet, 1981, p. 4; Krashen, 1982, pp. 164-167, 1985, pp. 89-94; Olshtain, 1976, pp. 39ff). However, in a survey of the literature on FL extensive reading, Brumfit (1978) noted: (a) the role of the extensive reader in the curriculum has been surprisingly little studied (p. 178); and (b) "the discussion of teaching methods is conducted at a low theoretical level if it is conducted at all" (p. 179). MacLean's (1985) bibliography of reading in a second or foreign language lists only four items on extensive reading. Zvetina's ( 1987) survey of research on L2 reading does not even mention extensive reading. None of the 99 items in ERIC computer search #200, "Reading Strategies in Second Languages" (October 1987) is about extensive reading. Swaffar's (1988) survey of FL reading research mentions only one article on extensive reading out of 221 items. Oddly enough, one conclusion Swaffar reaches after examining 220 items not about extensive reading is that teachers in the future "may well be asking students to do extensive reading on a longer text or in a particular field of study" (p. 141).

What literature there is on extensive reading is of limited value. Most general works on FL reading that discuss extensive reading do so in terms of book selection and course administration. Some even have detailed drawings showing how to display books (Nuttall, 1982, pp. 175ff; see also Bright & McGregor, 1970, pp. 65-80). More useful are the articles and reports on the pedagogical aspects of extensive reading (e.g., Boys, 1987; Hamrick, 1989; Kalb, 1986; Lipp, 1988; Lupardus, 1987; Marbe, 1979; Susser & Robb, 1989; Tangitau, 1973), including at least three at JALT '89 (Fox, 1989; Mason, 1989; Morimoto, 1989).

Only a few experimental studies of FL extensive reading exist: Elley and Mangubhai (1983); Hafiz and Tudor (1989) (see also Tudor & Hafiz, 1988, 1989; Hafiz and Tudor 1990; Hamp-Lyons 1983; Laufer-Dvorkin, 1981; Mason, 1987, forthcoming; Petrimoulx, 1988; Robb & Susser, 1989; and Saragi et al., 1978). These studies are limited for two reasons: First, methodological problems make many of their results worthless. Hafiz and Tudor (1990), for instance, report "significant post-treatment gains" (p. 36) for students who read extensively, but the gain could be the result solely of the 90 additional hours of exposure to English that the experimental group received, rather than from the extensive reading procedure itself. A second problem with studies on extensive reading is that the theoretical problems which underlie the extensive reading procedure--the definition of extensive reading, the nature of the materials to be read, and the transfer of L1 reading ability--have not been resolved.

2.2 Definition of Extensive Reading

Our working definition of "extensive reading" as a language teaching/learning procedure is that it is reading (a) of large quantities of material or long texts; (b) for global or general understanding; (c) with the intention of obtaining pleasure from the text. Further, because (d) reading is individualized, with students choosing the books they want to read, (e) the books are not discussed in class. (Based on Bamford, 1984a, pp. 4; Bamford, 1987; Barnett, 1989, p. 167; Brumfit, 1984, p. 84; Dawes, 1979; Dubin & Olshtain, 1977, pp. 77ff; Eskey, 1973, p. 173; Grellet, 1981, p. 4; Hedge, 1985, pp. vii, 68, 70; Krashen, 1982, pp. 164-167; Krashen & Terrell, 1983, p. 134; Norris, 1975, p. 208; Olshtain, 1976, pp. 39ff; Rivers, 1981, pp. 37, 278; Thompson, 1984, p. 21.)

Large quantities are essential for this procedure to be "extensive," but there is no agreement on how much "extensive" is, as the following examples show: (a) thirty pages an hour (Hill and Thomas, 1988, p. 50); (b) three pages an hour (Matsumura, 1987, p. 120); (c) an hour per evening (Krashen, 1981, p. 105); (d) five hours by a specified date (Bowen, Madsen and Hilferty, 1985, p. 239); (e) an hour of extensive for every hour of intensive (Williams, 1986, p. 44); (f) one page per day and three pages per day during summer vacation (for Japanese high school students) (Matsumura, 1987, p. 179); (g) thirty minutes per day for five stories, poems, or essays per week (Dalle, 1988, p. 25); (h) at a rate of at least 200 words per minute and up to 250 words or more (Hill, 1986, p. 16); (i) at least 50 pages per week (Paulston and Bruder, 1976, p.202); (j) a chapter per week (Hansen, 1985, p. 161); (k) two hours per week of texts 10-20 pages in length (Newmark, 1971, p. 16); (l) one reader per week (Stoller, 1986, p. 65; Eskey, 1973, p. 176; Brumfit, 1979, cited in Bamford, 1984d, p. 260); (m) at least two books a week (Carroll, 1972, p. 180); (n) 60 hours over 3 months (Hafiz and Tudor, 1989, p. 7); (o) a minimum of 36 simplified readers per year (Hill, 1983); (p) 60 books a year (Bright & McGregor, 1970, p. 69); (q) a novel (Ferris, Kiyochi, and Kowal, 1988).

This variety suggests that quantity of reading is not an absolute number of hours or pages but depends on teacher and student perceptions of how extensive reading differs from other reading classes; this will vary according to type of program, level, and other variables. In an EFL situation such as Japan, a typical university "intensive reading" class might "read" fewer than 100 pages a year (in one 90-minute class weekly for 26 weeks). These students and teachers would perceive 1,000 pages as extensive. Of course, quantity by itself does not make the extensive reading procedure. It has specific techniques, practices, and activities (described below). Consequently, there is no hard and fast rule for the amount of reading to be done extensively, but a good rule of thumb is offered by Light (1970): the assignments should be of sufficient length "so that neither teacher nor pupils will fall for the temptation to talk them through in class" (p. 123).

By aiming at general comprehension, this procedure reduces both teacher demands on the student and student demands on the text to attain the objectives of fluency and speed as well as comprehension. Broughton et al.'s comment that extensive reading must imply a "relatively low degree of understanding" (1978, p. 92) must be taken in context: we want students to achieve a degree of understanding sufficient for pleasure reading. If the student finds the book too difficult to enjoy, the extensive reading procedure requires that the book be changed, and not that the student be made to study it more closely. The level of global understanding required varies with the student's language proficiency, the nature of the text, and other factors.

The third aspect, pleasure, may seem dubious, because, after all, the reading is an assignment, and most of us do not take pleasure in assignments. However, pleasure, like quantity, is relative. The procedure assumes that students will enjoy reading books that they have chosen on topics of interest to them more than they will enjoy assigned readings from a reader. If they have chosen correctly, the book should be easy to read for general understanding. Nell (1988), in his interesting study of the psychology of pleasure reading, argues (anecdotally) that one cannot read for pleasure in a foreign language before mastering it (p. 93), but any EFL reading teacher can supply anecdotal counter-evidence. Further, even learners who are far from fluent derive pleasure from the very experience of reading a book in a foreign language. Students in extensive reading courses regularly comment on their joy at having finished whole books in the target language.

A few other points on the definition of extensive reading should be clarified. Some writers (e.g., Field, 1985, p. 177; Munby, 1979, p. 143) see extensive reading as just another reading subskill such as skimming or scanning. This confuses the whole with its parts. We see extensive reading as a teaching/learning procedure, not a reading subskill. Further, the implication in many works that extensive reading is by definition the reading of graded readers (e.g., Bamford, 1984b, p. 3; 1987; Bright & McGregor, 1970, p. 65; Broughton et al., 1978, p. 110; Dubin & Olshtain, 1977, pp. 77ff; Livingstone et al., 1987, pp. 5-6; Stoller, 1986, p. 65) has no basis in theory or practice. There is no reason extensive reading should be confined to graded materials.

3. Issues in Extensive Reading

3.1 Graded Readers

Many, perhaps most, of the TESOL specialists who recommend extensive reading do so on the assumption that the students will be reading graded readers. In Japan, for example, both academics (Bamford, 1984b, 1984c, 1987; Kitao & Shimatani, 1988) and publishers' representatives (C. Thompson, 1984; M. Thompson, 1988; Tunnacliffe, 1983) have argued that reading can be studied more effectively and enjoyably when students use easy material that they can understand and enjoy, instead of being forced to decode and translate texts hopelessly beyond their abilities.

Exactly what are graded readers? Simensen (1987, pp. 42-43) distinguishes three types of graded reader: (a) authentic readers, not written for pedagogic purposes; (b) pedagogic readers, specially written for EFL/ESL students; and (c) adapted readers, which have been adapted from authentic texts. Strictly speaking, materials in her first category are "graded" after the fact; they include L1 children's literature and books for young people, known as high interest low vocabulary books" (the ERIC descriptor), "young adult literature" (Reed, 1985), High interest--easy reading (Matthews, 1988), or "easy read books" (Abrahamson and Conlon, 1988, p. 686). Hill and Thomas (1988, p. 44) define a graded reader as a book "written to a grading scheme," whether it is a simplified version of a previously written work or an original work written in simple language.

Although graded readers are widely used, research has pointed out some important problems with them. No one expects a simplified 70-page version of War and Peace to convey the richness and subtlety of the original, but we do expect that the graded version will be written in correct English. However, there are many examples of graded readers that are written in poor English or are empty of content (Davison, 1986, pp. 20-21; Wallace, 1988, pp. 153-154). More importantly, the process of simplification often leaves writing that is more difficult to understand than the original, because, for example, cohesion, coherence, and discourse structure are impaired (Anderson & Armbruster, 1986, pp. 154ff; Beck & McKeown, 1986, pp. 122ff; Davies & Widdowson, 1974, pp. 176ff; Harrison, 1980, pp. 134ff; Hedge, 1985, pp. 20-21; Honeyfield, 1977, pp. 434ff; Widdowson, 1978, pp. 88-89; Yorio, 1985, p. 160. Kelly [1969, pp. 140-42] shows that such objections have a long history, and Carrel [1987] surveys the role of readability formulas in creating this situation).

Some reading experts argue that denying FL learners access to complicated prose will prevent them from ever learning to comprehend it. Lautamatti ( 1978) says that we cannot use graded readers because

the reading process relies on a selective use of all possible levels of the text, and is based on the maximum use of minimum clues [so] it is only by giving the student material containing all the features naturally occurring in informative texts, that we can make it possible for him to learn to take advantage of these. (p. 104)

(See also Blau, 1982, p. 525; Dunning, 1988; Grellet, 1981, pp. 7-8; Hirvela, 1988; Krashen, 1985, p. 113, n. 9; Shook, 1977; Swaffar, 1985, p. 17; van Naerssen, 1985, p. 6.)

The literature presents many arguments for the use of authentic materials in teaching FL reading (e.g., Allen et al., 1988, p. 163, 170-71; Byrnes, 1987, p. 183), including some evidence that learners prefer them to commercial materials (Henner-Stanchina, 1985, p. 92). However, authenticity is not easy to pin down. Presumably graded readers, being written for pedagogic purposes, are not "authentic," and yet Simensen's first category of graded reader is "authentic readers." Even if we ignore this, we still must face the problem posed by Widdowson (1979), that authenticity does not reside in texts but is "a quality which is bestowed upon them" (p. 165). The "high interest low vocabulary books" correspond to Widdowson's "simple accounts": "genuine instance[s] of discourse, designed to meet a communicative purpose..." (Widdowson, 1978, p. 89; see also Davies, 1984, pp. 181ff). For example, when students read fiction as fiction for pleasure, it is not only genuine discourse but also authentic in Widdowson's terms (1979, p. 166; see also Grellet, 1981, pp. 7-8; van Naerssen, 1985, p. 6). In other words, reading is authentic when students read books for the purpose for which they were written rather than for language study.

3.2 Transfer of L1 Ability

Extensive reading as a teaching procedure cannot be considered without reference to the transfer of L1 reading ability. So far, the only explanation of why extensive reading is effective is that it replicates the process by which we learn to read in our native language, that is, "prolonged practice" (Nell, 1988, p. 84) or learning "to read by reading" (Smith, 1985, p. 88). If so, then an understanding of how and how much L1 reading ability transfers to L2 would help us build a model of extensive reading.