The effectiveness and efficiency of extensive reading at developing reading rates

System 70 (2017) 92e106

Contents lists available at ScienceDirect

System journal homepage: www.elsevier.com/locate/system

The effectiveness and efficiency of extensive reading at developing reading rates Stuart McLean a, b, *, Greg Rouault c a

Osaka Jogakuin University, Japan Kansai University Graduate School, Japan c Tezukayama Gakuin University, Japan b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 23 May 2015 Received in revised form 1 September 2017 Accepted 5 September 2017

Few studies have examined the development of foreign language learners’ reading rates through extensive reading. The previous studies conducted have methodological limitations with regards to their research design or interpretation of results. To address these limitations, this study investigated the impact of extensive reading and grammartranslation on reading rate development using an experimental research design with evidence that time spent conducting the respective treatments was similar. First-year Japanese university students (N ¼ 50) were randomly assigned to one of two treatment groups. To measure reading rate improvements over an academic year, pre- and posttreatment reading rate measurements were used where comprehension was maintained above 70%. The between-groups analysis revealed that the extensive reading group participants (n ¼ 23) increased their reading rate significantly relative to the grammartranslation group participants (n ¼ 27). This study provides evidence of both the effectiveness and efficiency of developing reading rates through extensive reading relative to traditional reading instruction with grammar-translation exercises. Pedagogical implications include allocating more time for extensive reading and questioning the value of the grammar-translation approach. In addressing the call for stronger evidence than quasiexperimental studies, this research demonstrates that classroom-based experimental reading studies which control for time-on-task are feasible. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Extensive reading Intensive reading Grammar-translation Reading rate Comprehension Time-on-task Experimental design

1. Introduction In foreign language contexts, reading has been commonly conceptualized under two types: intensive and extensive reading. Palmer (1917) provided an initial, enduring definition: Reading may be intensive and extensive. In the former case each sentence is subjected to a careful scrutiny, and the more interesting may be paraphrased, translated, or learnt by heart. In the latter case book after book will be read through without giving more than a superficial and passing attention to the lexicological units of which it is composed. (p. 205)

* Corresponding author. Osaka Jogakuin University, Tamatsukuri, Chou-ku, Osaka-shi, Osaka, Japan. E-mail address: [email protected] (S. McLean). http://dx.doi.org/10.1016/j.system.2017.09.003 0346-251X/© 2017 Elsevier Ltd. All rights reserved.

S. McLean, G. Rouault / System 70 (2017) 92e106

93

Extensive reading (ER) is an increasingly popular approach to developing reading skills. However, in EFL contexts, the grammar-translation approach is still commonly used together with intensive reading instruction (e.g., Liu & Littlewood, 1997; Pazaver & Wang, 2009; Rao, 2001; Vardanjani, 2013). In Japan, although the use of ER has grown (Nakanishi & Ueda, 2011), grammar-translation is still the common instructional approach (Hino, 1988; Kikuchi & Browne, 2009; Tsukamoto & Tsujioka, 2013). While the number of studies investigating ER has increased greatly over the past 20 years, only 34 studies (up to 2012) met Nakanishi’s (2015) criteria for inclusion in his meta-analysis of empirical studies in which the overall effectiveness of extensive reading was investigated. Furthermore, only a limited number of published studies (Appendix) have investigated the influence of ER on reading rate. However, because of research design limitations, these studies can make only limited causal inferences between ER and reading rate improvement. By deploying an experimental research design and providing evidence that time-on-task was similar for the treatment groups, reading researchers can make a stronger case that ER is not only more effective but also more efficient than grammar-translation at developing learners' reading rates. 2. Literature review To understand the need for ER and the drive for more principled studies to support its incorporation into foreign language classrooms and curricula, it is important to examine the processes involved in reading development. 2.1. Reading development Fluent reading with appropriate levels of comprehension requires the efficient use of numerous cognitive processes. Because cognitive resources are finite, some if not all of these processes must become automatized (Grabe & Stoller, 2011). Readers must develop accurate and automatic sublexical and lexical processes; integrate the perceptual, phonological, orthographic, and morphological processes necessary during the reading of single words; and extract semantic meaning rapidly from orthographic forms. Readers also have to parse syntax and analyze text into phrasal and intersentential formations and interpret messages within the text through schematic activation (Wolf & Katzir-Cohen, 2001). First-language (L1) researchers have found a positive relationship between reading fluency and comprehension (e.g., De Soto & De Soto, 1983; Fuchs, Fuchs, & Maxwell, 1988; Nathan & Stanovich, 1991). Carver (1982) demonstrated this positive relationship by showing how restricting the L1 reading rate reduced readers’ text comprehension when compared with comprehension levels while reading at an unrestricted rate. McMillion and Shaw (2016) found that even advanced L2 readers typically read slower in their second language. Effective decoding and the development of automatic word recognition skills are the principal rationalizations for a positive relationship between comprehension and fluency (Grabe & Stoller, 2011). This stance is based on the information processing model of reading (LaBerge & Samuels, 1974). The automatization of lower-level reading processes reduces the cognitive effort required for word recognition. Automatization facilitates the application of cognitive resources to other lower-level processes metaphorically called bottom-up skills (e.g., syntactic parsing and semantic proposition coding) and also to top-down skills or higher-order processing in working memory. Empirical research into reading development posits that lower-level processes are of critical importance. Stanovich and Stanovich (1999) stated that “instructional programmes that emphasize spelling-sound decoding skills result in better reading outcomes because alphabetical coding is the critical subprocess that supports fluent reading” (p. 29). The label lowerlevel processes does not mean that they are easy or simple, but rather that they are elements more suited to become automatized. This automatization of lower-order skills is a requirement for fluent reading (Alderson, 2000; Koda, 2005). Many L2 learners, and especially those learning a language with a different writing system, face barriers to reading, even prior to developing any degree of reading fluency, as a result of limited lower-level processing abilities (Beglar, Hunt, & Kite, 2012; Grabe & Stoller, 2011). With languages that are orthographically different, little transfer of character recognition occurs. While related writing systems offer positive transfer, they can also lead to interference and slow L2 processing (Grabe & Stoller, 2002; McMillion & Shaw, 2016). Using spaced-reading tasks where participants pressed a key to move on to the next reading segment, Hoover and Dwivedi (1998) showed that faster L2 readers demonstrated faster word-recognition and parsing. The disparity between proficient and poor readers is unambiguous in the amounts of cognitive effort necessary for the two types of readers to engage in these skills. While proficient L2 readers identify words automatically, poor readers require greater cognitive resources and time to identify letters, common letter groupings, word morphology, and whole words (Grabe, 2009b; Koda, 2005). In addition to word recognition, accurate syntactic parsing is necessary to extract grammatical information to understand clause-level meanings. For L1 in general, Klauda and Guthrie (2008) found syntactic processing to be most strongly correlated with reading comprehension (r ¼ 0.75). More specifically they found that the meanings of polysemous words are more effectively and rapidly determined when syntactic aspects such as word order and phrasal groups within the text are interpreted in the way intended by the writer. Correspondingly, Alderson (1993) observed a high correlation (r ¼ 0.80) between syntactic knowledge and reading comprehension in an L2 setting, meaning that poorer readers face difficulty parsing the syntactic elements within written texts. However, in L2 contexts, “extended exposure to meaningful print” (Grabe & Stoller, 2011, p. 24) in ability-appropriate level texts can lead to more efficient processing as language learners develop the skill to parse text more fluently.

94

S. McLean, G. Rouault / System 70 (2017) 92e106

The final lower-level reading process is the semantic level. At the semantic level, proficient readers rapidly and accurately conduct semantic proposition formation. This is the process of evaluating word meaning and then combining word meanings and structural information into basic clause-level meaning units. In contrast, poorer L2 readers form semantic propositions slowly and require large amounts of cognitive resources, or they simply fail to do so, thwarting comprehension (Grabe, 2009b; Koda, 2005). Carver (1990) introduced levels of reading including scanning or skimming, rauding, reading to learn, and reading for memorization. Raudingda combination of “reading” as the comprehension of written text, and “auding” as the comprehension of spoken textdis defined as “the fastest rate at which the individual can read relatively easy material and still comprehend accurately” (Carver, 2000, p. 76). Fraser (2007) used various types of tests to measure reading performance across Carver's range. The participants' reading rates were slower in L2 English than in L1 Chinese, although these rates did not correlate significantly with language proficiency. Such findings suggested that reading rate performance was a function of higher-order processing not simply vocabulary knowledge. Similar to Carver’s (1982) observation for L1 reading, fluent L2 reading depends on the rapid, accurate semantic processing of content because of the short duration and limited capacity of working memory, which is where new information is integrated with previously read information before it is forgotten (Grabe & Stoller, 2011). In summary, Koda (2005) suggested that the key differences between L1 and L2 reading include: (a) previous L1 reading experiences allow L2 readers to apply L1 strategies and skills; L1 learners who are building basic literacy and reading skills concurrently cannot do this; (b) the cross-linguistic nature of L2 reading inevitably involves both positive and negative transfer from the L1; and (c) L2 readers have a limited L2 vocabulary compared to literate L1 readers. Thus, as proposed by Grabe and Stoller (2011) having language learners read large amounts of text with a reduced cognitive load, as featured in ER with its controlled lexis, is a legitimate learning method for automatizing lower-order reading processes and developing reading strategies. 2.2. Grammar-translation and extensive reading Although there are theoretical reasons supporting the effectiveness of ER, its adoption in EFL programs and curricula has been slow and even resisted. The following section covers two approaches to reading instruction: (a) grammar-translation, the dominant teaching paradigm in Asian contexts, and (b) extensive reading. 2.2.1. Grammar-translation The grammar-translation method is “an approach to second language teaching characterized by the explicit teaching of grammar rules and translation exercises” (Lightbown & Spada, 2006, p. 200), during which the L2 is commonly translated into the L1, and the L1 is the predominant language of instruction. Clark (2003) noted that English language education in Asia follows the grammar-translation method, which was “originally designed to teach people how to read a language that they would never have to speak or write” (p. 325). From a study with over 1000 respondents, Hino (1988) reported that 70e80% of Japanese teachers of English were using yakudoku (the grammar-translation method) in their university and high school classes. In Japan, grammar-translation is currently the most common form of L2 instruction because of teachers’ previous language learning experiences (Tsukamoto & Tsujioka, 2013) and because of the amount and type of pre- and in-service teacher training experienced (Kizuka, 2006). While Grabe (2009b) advised that no single definition can capture the complexity of reading and its different purposes and processes, “most reading instruction occurs above the level of the beginning L2 student” (p. 216). Therefore, a link can be made between the grammar-translation approach and intensive reading used by EFL teachers. Song (1995) wrote of Asian students’ beliefs that the goal of reading is to decode words and recite the whole text fluently rather than to analyze a reading passage for its meaning. Nation (2009) described grammar-translation as the classic procedure used in teaching intensive reading focusing on the comprehension of a specific text by explaining it in the L1. In referencing Asian contexts, Nation and Waring (2011) explained that “many teachers believed the only way to teach reading was by working intensively with short difficult texts sentence by sentence, or word by word, and through translation if necessary” (p. 2). Because intensive reading including grammar-translation dominates reading instruction in the research setting, it was one treatment in this study. 2.2.2. Extensive reading ER was defined by Grabe and Stoller (2002) as “reading in which learners read large quantities of material that are written within their linguistic competence” (p. 259). In addition to automatizing the lower-level skills of word recognition and parsing processes, ER builds reading fluency, supports vocabulary learning, drives motivation to read more, and helps hone reading strategy use (Grabe, 2009a). Several studies have been conducted on the effectiveness of ER on reading fluency development. In a study with Saudi learners, Al-Homoud and Schmitt (2009) found that “learners who self-selected and read graded readers for pleasure … improved at least as much on every measure as students who received intense instruction in reading” (p. 399) and concluded that stress-free and popular extensive reading is a viable teaching approach. In his meta-analysis of 34 empirical studies on the effectiveness of ER, Nakanishi (2015) reported a medium effect size (d ¼ 0.46) for group contrasts when comparing an ER experimental group with a control group, and a medium effect size (d ¼ 0.71) for preepost contrasts when comparing students who received ER instruction and those who did not. These findings indicated that ER was more effective at developing the various language skills measured than the activities conducted by the control groups. For, example an effect

S. McLean, G. Rouault / System 70 (2017) 92e106

95

size of d ¼ 0.46 means that around 68% of the control group is below the average person in the experimental group. However, as Nakanishi pointed out, there are two major limitations in ER research: (a) In some studies, control group participants engaged in no activities while the ER group participants did ER, and (b) in some studies, the ER participants engaged in ER for much longer periods of time than the control group participants. Even in the face of considerable empirical evidence regarding the positive effects of ER, Grabe (2009b) offered five reasons why ER has not become more widely used in foreign language teaching and curriculum design. Each of these applies specifically to EFL teaching in Japan: (a) language skills, vocabulary building, grammar-translation, or study skills are often the goals of the reading course or curriculum, not fluent reading; (b) many reading teachers in EFL contexts believe that the path to becoming a fluent reader is to develop accurate comprehension abilities, even if it is a slow process, and further development is left up to initiative or need; (c) teachers without the language and reading skills themselves are not ready to rethink how reading should be taught; (d) administrators and teachers are uncomfortable beyond the direct, explicit, transmissionstyle of teaching and learning, and finally; (e) ER demands a significant number of resources in terms of reading materials and time. For these reasons, and to overcome the institutionalized barriers lying therein, robust studies of ER are required to provide stronger evidence of the effectiveness of ER in instructional settings. Such studies would justify the implementation and expansion of ER in EFL language learning curricula.

2.3. Previous research on reading rates Only seven studies (Appendix) are known to have investigated the influence of ER on reading rates. However, in each case, important design limitations have been present. Studies using a quasi-experimental design (e.g., Beglar et al., 2012; Bell, 2001; Huffman, 2014; Iwahori, 2008; Mason & Krashen, 1997; Taguchi, Takayasu-Maass, & Gorsuch, 2004) offer limited evidence of isolating the impact of ER on the development of reading rate. Thus, researchers are less able to argue that gains experienced by the non-randomized treatment groups relative to their control group are not the result of variables not controlled for in the study. For example, participants in control and treatment groups often experience different courses and lessons taught by different instructors. In contrast, in experimental research designs, participants’ experiences in other classes are more likely to be similar owing to the random assignment to groups even with a convenience sample from intact classes. L2 reading authorities have unanimously stated that reading fluency develops as readers read large amounts of text (e.g., Day & Bamford, 1998, p. 16; Grabe & Stoller, 2002, pp. 21, 23). However, while some studies (e.g., Beglar et al., 2012) have provided evidence that large amounts of text were read, other studies would have benefitted from doing so. For instance, in Taguchi et al. (2004), a mean of only 205 pages was read by participants. Further, Bell (2001) and Mason and Krashen (1997) did not state how much was read. As learners have different background knowledge, as well as gaps in their syntactic and lexical knowledge of English, the use of different reading rate passages as pre-treatment and post-treatment instruments is a limitation. Kramer and McLean (forthcoming) found significant differences in reading rate between texts of equivalent lexical coverage, syntactic difficulty, and word count, but which differed only in the number of standard words. A standard word unit is defined as six adjacent characters within a text, including all letters, numbers, punctuation marks, and spaces (Carver, 1982, 1990). Kramer and McLean accomplished this by creating sets of texts which only altered the proper nouns within the text to either longer or shorter forms. Thus, because different texts consist of words of different lengths, the measurement of reading rate, particularly for research purposes, would benefit from using the standard word unit. L2 reading rate studies using different English texts as pre- and post-reading rate instruments (e.g., Huffman, 2014; Robb & Susser, 1989) are limited in the degree to which comparisons can be made. This is because the rate at which a learner reads two different texts of the same length varies as learners have different gaps in their syntactic and lexical knowledge of English. Further, as background knowledge influences reading rate and comprehension, the use of different reading rate instruments is also a limitation. Thus, only researchers who have used the standard word unit and the same test as pre- and post-treatment instruments (e.g., Beglar et al., 2012) are able to argue that they have effectively measured changes in reading rate. Such standardization also facilitates the comparison of reading rate across studies. Finally, if researchers provide evidence that participants spent similar amounts of time completing ER and different treatments, it strengthens the argument for the greater efficiency of reading extensively as a learning activity. Of the previous ER reading rate studies, only Bell (2001) provided evidence that the participants engaged in treatments for similar amounts of time. The limitations of the research reviewed here reduce the validity with which researchers and proponents can argue that ER develops learners’ reading rate more effectively than grammar-translation, which is commonly used in Asian institutional settings. Consequently, this study was designed to address each of these limitations concurrently. Specifically, the following research questions are investigated: 1. Are post-treatment reading rates significantly higher than pre-treatment reading rates for (a) a group of Japanese undergraduates conducting weekly extensive reading and timed-reading practice, and (b) a group of Japanese undergraduates conducting weekly grammar-translation and timed-reading practice? 2. Were participants able to maintain adequate comprehension rates if their reading rates increased? 3. Which of the two treatments facilitated greater reading rate gains?

96

S. McLean, G. Rouault / System 70 (2017) 92e106

3. Method 3.1. Participants The participants (N ¼ 50) were a convenience sample of first-year Japanese EFL learners from five intact classes at a private university in western Japan. They had studied English formally for six years in secondary school. Standardized test scores were unavailable for these participants, but students not included in the study from a department within the same university who were placed in courses for students with similar English proficiency had a mean TOEIC score of 280. A baseline mean reading rate of 77 wpm was established from the number of words participants read during one hour of sustained silent reading in the first lesson of the year. All participants were enrolled in two compulsory courses, English A and English B, which provided a total of three hours of weekly instruction. This study was conducted in the English A courses, which met once a week for 90 min per session, for a total of 15 lessons per semester, over two semesters. The lead author taught all five English A courses, with class time concentrated on building spoken proficiency and with the treatments provided as homework. A questionnaire established that none of the participants were taking supplementary English classes outside of their university courses nor did they have any previous experience with English ER. Of the 59 students who started the year, nine did not complete one or both terms, and so they were excluded from this study, leaving a sample size of 50. 3.2. Design of study An experimental design was used where each student in the five classes was randomly assigned by the lead author to either the ER group or the grammar-translation group. This was done by randomly generating a number for each student and then creating an odd number group and an even number group. Before the groups were formed, it was decided that the odd numbered group would be the ER treatment group. The participants in the ER treatment group were set a minimum weekly reading goal that was calculated to take around 60 minutes to complete outside of class every week. Participants in the grammar-translation group were set grammar-translation and intensive reading activities for weekly homework calculated to take at least 60 minutes to complete. A homework time of 60 minutes was recommended by the university at which the research was conducted. All students received weekly feedback on their homework through feedback sheets. The ER group and grammar-translation group participants were each assigned marks for meeting their weekly reading target or completing their assigned homework, respectively. On the rare occasion that participants did not complete their assigned homework, they were encouraged to catch up by the following week. All participants recorded how long they spent completing the weekly homework activities in homework journals. As the treatment groups were formed through random assignment, the overall classroom experiences of the participants in their other classes were assumed to be evenly balanced across the two treatment groups. The university at which the research was conducted provided permission to conduct this research, and all of the participants agreed to participate in the study. The institution and participants understood that students in the same class would be experiencing different treatments for homework. The stance adopted by the institution was that different students learn in different ways and some people value reading fluency while others prioritize grammatical accuracy over fluency. 3.2.1. The ER group The ER group participants (n ¼ 23) self-selected graded readers predominantly from the Foundations Reading Library series published by Heinle Cengage and the Building Blocks Library series levels 5 to 7 (Scientific Education Group). A few participants read Oxford Bookworms Starters (Oxford University Press), Penguin Readers Easystarts (Pearson), and Cambridge Readers Starters (Cambridge University Press). Table 1 shows the lexical coverage of the first and second 1000-words of the British National Corpus/Corpus of Contemporary American English (BNC/COCA) (Nation, 2012) within the Foundations Reading Library series (McLean, 2014). Both the Foundations Reading Library series and levels 5 to 7 of the Building Blocks Library are at or below the 301e400 headword level of the Extensive Reading Foundation Graded Reading Scale (ERF, n.d.). These titles were therefore judged by the teacher/lead researcher to be within the students’ reading ability. The students were also regularly reminded of how to self-select reading materials at an appropriate level: They should (a) read a page and understand it, (b) find no more than two unknown words per page, and (c) enjoy the book. The selection of books read by the students indicated that the participants followed this advice. Steps were taken to increase the possibility for similar time-on-task to be spent by the ER and grammar-translation group participants on their respective treatments. In the first lesson of the year, a reading rate baseline for the ER group was established from the number of words participants read during one hour of sustained silent reading. During this time, there was no time pressure and enjoying easy, learner-level appropriate material was emphasized. A mean of 4624 words (SD ¼ 611) was read in 60 minutes, resulting in a sustained silent reading rate mean of 77 wpm. This is similar to what other researchers found measured among Japanese EFL students: 79 wpm in Robb and Susser (1989) and 82 wpm in Taguchi et al. (2004). However, as indicated by the standard deviation, individual students’ reading rates varied. Thus, while asking each ER group participant to read different numbers of words per week according to their reading rate is the ideal, it was unworkable pedagogically to do so. As a result, given the institutionally approved target of 60 minutes of homework per week, each ER group participant was required to read an average of 4000 running words a week over two academic semesters. A weekly

S. McLean, G. Rouault / System 70 (2017) 92e106

97

Table 1 Descriptive details of the Foundations Reading Library. Title

Headwords

Words

1k

1þ2k

Sarah's Surprise Goodbye, Hello Rain! Rain! Rain! Bad Dog? Good Dog! Get the Ball! The Tickets

75 75 75 75 75 75

526 568 504 622 521 620

94.5 96.6 94.6 95.6 96.0 92.9

96.9 98.6 99.6 99.1 98.1 97.8

FRL Level 1 Average SK8 for Jake The New Guitar Trouble at the Zoo Singer Wanted Old Boat, New Boat The Cave

560

95.0

98.3

100 100 100 100 100 100

724 716 757 648 713 726

94.9 90.0 92.1 92.0 97.3 95.9

96.0 93.5 94.7 98.3 98.2 97.5

714

93.7

96.4

150 150 150 150 150 150

771 762 980 693 681 895

96.6 97.2 96.3 95.8 93.3 95.1

97.9 99.5 98.7 98.3 100.0 97.2

FRL Level 2 Average Slam Dunk for Mark Kung Fu Kid A Good Friend Quick Thinking I Always Win! Quiz Night FRL Level 3 Average I Spy Go Jimmy Go! Do I Tell? Lost at Sea The Shipwreck Mystery on the Island

797

95.7

98.6

200 200 200 200 200 200

1280 1347 1330 1405 1301 1422

93.8 92.0 97.7 92.2 89.9 96.6

98.6 98.3 99.3 96.6 95.0 98.5

1347

93.7

97.7

250 250 250 250 250 250

1605 1127 1636 1710 1444 1684

97.1 96.1 95.3 92.2 91.6 91.3

98.5 99.1 98.3 95.7 96.7 95.8

1534

94.0

97.3

300 300 300 300 300 300

2590 2345 2633 2447 2301 2634

95.0 91.9 94.7 95.4 92.8 95.5

96.8 97.7 98.4 98.5 97.6 98.9

2492

94.2

98.0

2690 2471 2892 2285 2926 2852

97.1 95.9 95.7 92.9 96.0 91.7

98.8 97.4 98.8 96.8 98.2 96.0

FRL Level 4 Average Who's Best? The Big Test Where's Lorena? Boys vs. Girls Think Daniela! The Bear's Mouth FRL Level 5 Average The Lost Wallet No, You Can't! Does He Love Me? A Helping Hand Trouble at Sea The Old Promise FRL Level 6 Average Let's Party! Do it! My Mom, The Movie Star The Secret Tunnel Love Online The Golden Monkey

350 350 350 350 350 350

FRL Level 7 Average

2686

94.9

97.6

FRL series average

1447

94.5

97.7

Shimada et al. (2011a)

6193

83.4

93.8

Shimada et al. (2011b)

10529

82.7

92.1

83.1

93.0

Series average

Note. 1k ¼ the percentage of words from the 1000-word frequency level; 1þ2k ¼ the percentage of words from the 1000 and 2000-word frequency levels of the BNC/COCA corpus.

target of 4000 words per week, at a mean reading rate of 77 wpm, results in participants reading for an average of 52 minutes a week. However, it was expected that when answering the comprehension quizzes used to track the reading completion, some re-reading of the material would take place. Therefore, a weekly reading goal of a minimum of 4000 words was set for the ER group.

98

S. McLean, G. Rouault / System 70 (2017) 92e106

It should also be emphasized that while a weekly reading goal was set, the ER group participants were encouraged to read as much as possible beyond their weekly reading goal. However, experience conducting ER with non-English majors in compulsory university courses in Japan led the lead author to anticipate that only a small number of ER group participants would read significantly more than was required, even with encouragement. Thus, time-on-task for the two treatment groups was not controlled for by limiting the amount of ER conducted; instead, efforts were taken to encourage both ER and grammar-translation group participants to complete approximately one hour of homework each week. Evidence of ER group participants completing their graded reading task as homework outside of class came from participants completing quizzes on the Moodle Reader module (see https://moodle.org/plugins/mod_reader). The Moodle Reader module (now MReader) is an online program on which students take a short comprehension quiz of 10 questions specific to a single graded reader designed to demonstrate that a student has read the book. To control against non-reading and task apathy, it was necessary for students to correctly answer 7 out of 10 questions in order for the word count of a book to contribute towards their weekly word goal. The pass rate for reading comprehension in this study was set at 70%, as recommended by Nuttall (2005) and designated as “adequate comprehension” by Anderson (2008, p. 67). If a student failed a quiz for a particular graded reader, as was the case in only 2.39% of all instances, it was considered not possible to determine if the book was read. The book's word count was thereby not included in the student's weekly reading target, and the student was required to read other books to reach the weekly reading target. Students were closely monitored and encouraged weekly to complete their reading target or to catch up quickly. The lead author checked the Moodle Reader log to record the number of words each student read and awarded the points for them on weekly feedback sheets. For every 4000 words read, students received three points towards their final grade. Students were aware of this system and keenly checked their weekly progress. As part of their homework, participants in the ER group recorded the number of minutes spent reading and taking quizzes in homework journals. 3.2.2. The grammar-translation group The grammar-translation group participants (n ¼ 27) were assigned intensive reading and grammar-translation exercises to be completed from published materials (i.e., Shimada, Hisasue, & Yoneyama, 2011a, 2011b). The exercises consisted of selecting the meaning of pronouns, vocabulary items, and sentences; explaining the meaning of sentences in Japanese; and translating English sentences into Japanese. These materials were selected as they were (a) similar to those used by reading instructors in the institution where the research was conducted and (b) accompanied by recommended completion times for each unit (see Shimada, Hisasue, & Yoneyama, 2011a, 2011b). The lexical coverage of the reading sections of the two textbooks at the 1000-word (1k) and 1000 and 2000-word (1 þ 2k) frequency levels of the BNC/COCA are shown at the bottom of Table 1. Students in the grammar-translation group were assigned homework of two units per week with combined recommended completion times of between 60 and 70 minutes. The participants completed homework journals in which they noted when they started, paused, and completed homework assignments. Each week, the lead author marked and returned the participants’ homework with answer sheets including model translations. Weekly, students received three points towards their final grade for completed grammar-translation homework. Participants who were absent, did not submit their homework, or failed to complete their homework were required to submit missed homework by the following week. The students understood this system and actively tracked their progress. 3.3. Instruments Reading rates were measured with timed-reading passages followed by comprehension questions. Each timed-reading passage was a double-sided sheet with the 400-word text on the front and 10 multiple-choice comprehension questions on the back. All of the comprehension questions were factual questions with answers explicitly available in the text without any negative fact questions or higher-order inferencing required. As the class teacher, the lead author instructed the students to start reading upon a given signal, to read the whole text as quickly as possible while maintaining comprehension, and to then record their reading time at the bottom of the sheet. To facilitate timing, a large digital stopwatch was projected on to a screen at the front of the classroom. After recording their time, the students turned the sheet over and answered the comprehension questions without referring to the text. Upon finishing the questions, the students put their pens down and waited. The instructor monitored the students carefully to ensure that they did not refer back to the text nor confer with other students for assistance when answering the comprehension questions. Two 400-word timed-reading exercises were completed each week in class by members of the ER and grammartranslation groups. The students read texts written at the 2000-word level from New Zealand Speed Readings for ESL Learners, Book 1 (Millett, 2005) in sequence from Passage 1 to Passage 20, and from Passage 20 to Passage 1 of Asian and Pacific Speed Readings for ESL Learners (Quinn, Nation, & Millett, 2007) written at the 1000-word level. Although each passage contained around 400 words, the length of each passage was converted into standard word units (Carver, 1982, 1990), and the reading rate was measured in standard words per minute. The participants' mean comprehension score of 6.21 (SD ¼ 1.83) out of 10 on passages from New Zealand Speed Readings for ESL Learners, Book 1 was below the 70% figure adopted as evidence of comprehension in this study. Therefore, the data from these passages were not included in this study. In contrast, participants’ mean comprehension score of 7.21 (SD ¼ 1.09) out of 10 on passages from Asian and Pacific Speed Readings for ESL Learners was above the 70% target adopted for this study.

S. McLean, G. Rouault / System 70 (2017) 92e106

99

Passages 18, 17, and 16 of Asian and Pacific Speed Readings for ESL Learners from the third, fourth, and fifth weeks of timed reading made up the pre-treatment instrument. The reason was, as indicated in Fig. 1, the impact of any practice effect had ebbed by the third week of timed-reading practice. Participant interviews confirmed that in the first week of the timedreading practice, the participants were too cautious and read slower than usual. Subsequently, in the second week, participants overcompensated and read too fast to the detriment of comprehension. The post-treatment instrument used the same Passages 18, 17, and 16 from Asian and Pacific Speed Readings for ESL Learners to ensure equivalent reading difficulty. The post-treatment timed readings were conducted at the end of the second semester, eight months after they were completed as the pre-treatment instrument. It was believed that the participants had no background knowledge of the topics of the passages used for the pre- and post-treatment instruments (King Mongkut of Thailand, Abdullah, and Stamford Raffles) to correctly answer the comprehension questions without understanding the reading passages. Interviews with participants confirmed this assumption.

3.4. Procedures The participants’ sustained silent reading rates were established in the first week of the first semester, and the participants registered on the Moodle Reader module during the second week. Pre-treatment reading rate measurements were taken in weeks five, six and seven of the first semester. The ER and grammar-translation group-specific treatments were conducted between the third and fifteenth week of the first semester, and between the first and twelfth week of the second semester. All participants affirmed that they did not engage in ER or grammar-translation during their two-month summer break. Weekly timed-reading practice was conducted with both treatment groups between the third and twelfth week of the first semester, and between the first and tenth week of the second semester. Post-treatment reading rate measurements were taken during weeks 13, 14, and 15 of the second semester. Throughout the academic year, the participants in both treatment groups were interviewed in order to provide information to help explain the quantitative findings. 3.5. Data analysis Research Question 1 (Change in reading rates) was addressed through paired-sample t-tests, which compared pre- and post-treatment mean reading rates for each group separately.

Fig. 1. Mean reading rate (standard word units per minute) during timed-reading practice.

100

S. McLean, G. Rouault / System 70 (2017) 92e106

Research Question 2 (Degree of comprehension) was addressed by establishing that the two groups’ mean comprehension scores during the three-week pre-treatment and post-treatment measurements remained above the 70% criterion set for adequate comprehension. This ensured that any increase in reading rate was not accompanied by a decrease in comprehension scores. Research Question 3 (The effect of treatments on reading rate) was addressed through first confirming that there was no significant difference between the two groups’ pre-treatment mean reading rates, and then conducting a t-test comparing the post-treatment reading rate gains by the two groups. As a result of these four statistical tests, and in order to reduce the chance of a Type 1 error, statistical significance was set at 0.0125 (0.05/4 ¼ 0.0125). In order to further investigate the efficacy of the two treatments, effect sizes were calculated. Effect size is a simple way of quantifying the difference between groups, or the influence of a treatment on a group. Effect size emphasizes the impact of a treatment rather than confounding this with sample size (Carver, 1993). This study adopted Plonsky and Oswald’s (2014) recommendations for small, medium, and large effect sizes for mean differences in within-group contrasts (g) of 0.6, 1.0, and 1.4, and 0.4, 0.7, and 1.0 for between-group contrasts (d). Before analyzing the results, it was necessary to confirm that time-on-task was similar for the two groups. Table 2 shows the self-declared mean weekly homework completion times in minutes. The difference between the amount of time spent conducting homework by the ER group and the grammar-translation group was not significantly different, t(48) ¼ 0.166, p ¼ 0.10, d ¼ 0.48. The Cronbach's alpha value for the self-reported mean weekly homework completion times was 0.90. While these measurements were self-declared, they are consistent with expectations. These expectations were based on the care taken in the research design and the minimum homework requirements set as per curricular guidelines along with the encouragement for students to complete their homework as part of their grade. An argument that the time spent conducting treatment activities by the two groups was not significantly different can be made because of three unique aspects of this study. First, the participants in the two groups reported similar homework completion times in their journals. Second, the ER group participants were set a weekly word target expected to take around 60 minutes based on their mean reading rates established in the first week of the term. Finally, the participants in the grammar-translation group completed homework activities based on the publisher's recommended completion times for each activity. Based on the evidence that the ER and grammar-translation group participants conducted their respective treatments for similar amounts of time, this research isolated the impact of ER on participants' reading rate to a greater degree than the reviewed studies (see Appendix). Despite the participants in both groups taking similar amounts of time to complete their respective treatments, the amount of text read by the two groups differed. The ER group participants’ word counts were taken from the Moodle Reader module. Table 3 shows that the ER group participants read substantially more words than the participants in the grammartranslation group. The ER group participants were set a goal of 112,000 words over the academic year. This was in line with their sustained silent reading rate established the first week of term and the 60-minute minimum target assigned as weekly homework. Of the 23 ER participants, 18 read over 112,000 words, two read over 100,000 words, one read more than 90,000 words, and the remaining two read over 60,000 words. While it might be argued that the two participants who read only over 60,000 words did not read sufficient amounts for their reading to be considered ER, these are respectable amounts for nonEnglish majors taking compulsory English classes in Japan. Data from each participant are also relevant to preserve the ecological validity of the sample and support the external validity for the generalizability of the inferences.

Table 2 Descriptive statistics for self-declared mean weekly time spent on homework (minutes).

M SE 95% CI lower bound 95% CI upper bound SD

Extensive Reading

Grammar-translation

61.91 1.47 58.86 64.96 7.05

56.91 2.49 51.80 62.02 12.92

Table 3 Descriptive statistics for the number of words read.

M SE 95% CI lower bound 95% CI upper bound SD

Extensive Reading

Grammar-translation

107,964.04 3257.90 101,207.59 114,720.50 15,624.30

16,464 0

0

S. McLean, G. Rouault / System 70 (2017) 92e106

101

Table 4 Descriptive statistics for the mean pre-treatment reading rate. Reading rate (standard words)

M SE 95% CI lower bound 95% CI upper bound SD

Extensive Reading

Grammar-translation

99.38 3.20 92.74 106.02 15.35

97.79 3.80 89.90 105.68 18.23

The grammar-translation group participants read 16,464 words when completing the intensive reading and grammartranslation activities in the assigned textbooks. However, since the grammar-translation group participants were required to complete tasks related to the passages, how much re-reading was done is unknown. Nevertheless, the self-reported homework times of the two treatment groups were similar. Prior to comparing changes in reading rate between the two treatment groups (Research Question 3), it was necessary to establish that there was no significant difference in the pre-treatment reading rates (see Table 4). In a paired sample t-test using the pre-treatment reading rate measurement data, none was found, t(48) ¼ 0.804, p ¼ 0.43 d ¼ 0.09.

4. Results Research Question 1 asked whether post-treatment reading rates were significantly higher than pre-treatment reading rates for either treatment group. A paired sample t-test found that both the ER group (t(22) ¼ 7.75, p ¼ 0.000, g ¼ 1.32) and the grammar-translation group (t(26) ¼ 3.027, p ¼ 0.006; g ¼ 0.76) participants significantly increased their reading rate with medium and small effect sizes, respectively (see Tables 4 and 5). Possible causes for the significant gains in reading rate experienced by both groups could come from performing timed-reading activities each week, activities conducted in other English lessons, as well as the respective treatments. Research Question 2 asked whether the participants were able to maintain adequate comprehension rates as their reading rates increased between the pre- and post-treatment measurements. Table 6 shows that both the ER and grammartranslation group participants' pre- and post-treatment comprehension rates remained above the 70% threshold for comprehension. The difference between the pre-treatment comprehension scores of the two randomized groups was not significant, t(48) ¼ 1.318, p ¼ 0.194, d ¼ 0.36. The difference between the two groups’ post-treatment comprehension scores was not significant, t(48) ¼ 0.756, p ¼ 0.453, d ¼ 0.22. Carver’s (1990) and Breznitz’s (2006) L1 reading models provide a possible explanation for this finding. According to rauding theory (Carver, 1990), learners' most efficient reading comprehension is achieved at their optimal reading rate or when rauding. If learners read above or below their optimal rate, they comprehend less effectively because reading too slowly

Table 5 Descriptive statistics for the mean post-treatment reading rate. Reading rate (standard words)

M SE 95% CI lower bound 95% CI upper bound SD

Extensive Reading

Grammar-translation

130.34 5.84 118.24 142.44 27.99

103.05 3.69 95.39 110.70 17.70

Table 6 Descriptive statistics for the mean initial reading comprehension scores (K ¼ 10). Pre-treatment

M SE 95% CI lower bound 95% CI upper bound SD

Post-treatment

ER

GT

ER

GT

7.36 0.18 6.99 7.74 0.87

7.13 0.24 6.64 7.62 0.24

7.31 0.22 6.85 7.78 1.08

7.07 0.22 6.62 7.53 1.15

Note. ER ¼ extensive reading; GT ¼ grammar-translation.

102

S. McLean, G. Rouault / System 70 (2017) 92e106

Table 7 Within-group reading rate gains and between-groups effect sizes. Reading rate change

Robb and Susser (1989)a Bell (2001)a Beglar et al. (2012)*b Huffman (2014) b McLean and Rouault (this study) McLean and Rouault (this study)

Effect size (d)

ER

GT

þ7.241 þ59.431 þ16.482 þ20.732 þ33.741 þ30.962

1.751 þ14.091 þ2.972 0.622 þ5.681 þ5.152

0.96 1.40 1.12 1.47 1.73 1.73

Note. Effect sizes are for differences between ER ¼ extensive reading group and GT ¼ grammar-translation group. 1 ¼ Words per minute, 2 ¼ Standard Word Units per minute. * Out of the three ER groups, Group 3 was chosen because it had the largest sample size. a Reported in Nakanishi (2015). b Calculated by the authors of this paper.

means that they are not reading efficiently, and reading too quickly means that they are sacrificing comprehension for reading rate. Thus, in Carver's model of reading efficiency, gains in reading rate with the same or better comprehension signal an improvement in optimal rauding rate. Although Carver's model proposes an optimal balance between reading rate and comprehension, it does not attempt to explain the mental processes underlying reading rate gains. Breznitz hypothesizes that the synchronization of multiple modalities and brain systems provides an explanation of the mental processes underlying improved reading fluency with maintained or improved comprehension. Breznitz (2006) argues that “successful synchronization is based not only on the content of the information, but also on the speed at which the information is processed and transferred within and between the various systems activated in the process” (p. 211). Thus, increased reading rates, while maintaining high levels of comprehension, might result from a combination of improved processing speed plus the coordination of such autonomous processes as orthographic decoding and phonological recoding. Finally, Research Question 3 asked which of the two treatments facilitated greater gains in reading rate. The within-groups comparison showed post-treatment versus pre-treatment mean reading rate gains for both the ER treatment group (þ30.96 standard words per minute) and the grammar-translation group (þ5.15 standard words per minute). However, an independent sample t-test found that ER group participants experienced significantly greater gains (t(48) ¼ 6.271, p ¼ 0.000, d ¼ 1.76) in reading rate than the participants in the grammar-translation group, with a large effect size. These gains are believed to be the result of ER group participants having read substantially more words (Table 3) within the same time-on-task, and thus having developed their lower-level reading processes more than the grammar-translation group participants. This interpretation is supported by previous research and interviews with the participants. Reading authorities (e.g., Grabe, 2009b; Grabe & Stoller, 2011; Samuels, 2006; Stanovich & Stanovich, 1999) have argued that it is through the reading and processing of large amounts of text that the lower-level reading processes of orthographic decoding, syntactic processing, and semantic proposition formation are developed. Participants in this study from both groups expressed feelings of greater confidence regarding the ability to conduct timed readings. However, this was much more the case for participants of the ER group. Only the ER participants stated that when conducting timed reading and ER they read increasingly smoothly in a way similar to L1 reading (e.g., not looking first for the subject of the sentence and then the verb and object, a tactic from grammar-translation practice taught in Japanese high schools and universities). 5. Discussion In this study, post-treatment reading rates were significantly higher than pre-treatment reading rates for both the ER and grammar-translation groups. As a result, the gains by the ER group participants were not due to engaging in ER alone. One explanation for reading rate improvements for both treatment groups is that through conducting timed-reading practice, reading rates increased, as shown by Bismoko and Nation (1974), Chang and Millet (2013), Chung and Nation (2006), Cramer (1975), Macalister (2008; 2010), and Yen (2012). This study showed that the participants maintained adequate comprehension rates as their reading rates increased between pre- and post-treatment measurements. These results contrast with Taguchi et al. (2004), who provided evidence that the materials were actually not comprehended. Mason and Krashen (1997), Bell (2001), and Iwahori (2008) would have benefited from tracking evidence of comprehension when measuring and reporting changes in reading rate. Sustained comprehension rates were reported in Beglar et al. (2012), Huffman (2014), and Robb and Susser (1989). These results support the hypothesis that ER is an effective means for developing higher reading rates while sustaining comprehension. Table 7 presents the change between the initial and final reading rates (in words per minute or standard words per minute) for studies promoting the effectiveness of ER in improving reading rates. The reading rate gains found in this study are less than those reported by Bell (2001), who calculated gains by comparing the post-treatment reading rates with the first timed-reading rates. However, using the initial timed-reading rate confounds development in reading rates due to the practice effect experienced when subsequently conducting the same task. The greater gains reported in the present study could be higher than those in Beglar et al. (2012), as the participants of the present study completed timed-reading exercises

S. McLean, G. Rouault / System 70 (2017) 92e106

103

weekly. Similarly, the 15 timed-reading passages completed between the pre- and post-treatment measurements in the present study may help explain the greater reading rate gains experienced (þ30.96 standard words per minute) relative to the þ20.73 standard words per minute gains in Huffman (2014) when only six timed readings were conducted between measures. Finally, the small reading rate gain (þ7.24 wpm) in Robb and Susser (1989) might have resulted from the difficulty of the unsimplified treatment materials not having developed participants’ ability to read fluently, or the methodological limitation in having used different passages as pre-treatment and post-treatment instruments. In this study, the effect size of the difference in reading rate gains experienced by the ER group and the grammartranslation group was large at d ¼ 1.73 Comparatively, Robb and Susser (1989), Bell (2001), Beglar et al. (2012), and Huffman (2014), reported between-group contrast effect sizes of d ¼ 0.96, 1.40, 1.12, and 1.47, respectively. This study provides evidence for the positive effects of large quantities of ER within an autonomous, monitored out-of-class setting. As this study measured the pre-treatment reading rate after the practice effect ebbed, it reduced the degree to which reading rate development was confounded by any such effect. By employing the standard word unit, and in using the same texts for the pre-treatment and post-treatment measurements, this study avoided these specific limitations in previous studies. Additionally, this study answers the calls for experimental research investigations (see Beglar et al., 2012; Huffman, 2014; Nakanishi, 2015) into the impact of ER on reading rate. Finally, under the design conditions that the two treatment groups conducted their respective treatments for similar amounts of time, the significantly greater increases in reading rate experienced by the ER group participants provide evidence of greater efficiency for ER over grammar-translation. The findings from this research should be viewed in the context of the study's limitations. Generalization of this study's results to other learning settings should be conducted with care. The participants had not engaged in ER previously and initial reading rates were very low. Thus, it can be expected that more proficient readers might not experience reading rate gains as large as in this study. Furthermore, L2 learners whose native languages are orthographically similar to English experience less difficulty in decoding English (Hamada & Koda, 2008). Another limitation concerns the potential practice effect arising from using the same three timed-reading passages as pre-treatment and post-treatment instruments. However, given the eightmonth gap between pre- and post-treatment reading rate measurements and the confirmed lack of background knowledge of the passage contents, it is believed that any such influence would be insubstantial. A final limitation of this study is the small number of participants comprising each treatment group. Future researchers should preferably conduct similar studies with larger sample sizes. 6. Conclusion This study investigated the impact of ER and grammar-translation treatments on reading rate development among Japanese undergraduates conducting timed-reading practice. Although post-treatment reading rates were significantly higher than pre-treatment reading rates in both conditions, the ER and timed-reading treatment facilitated significantly greater gains in reading rate than grammar-translation and timed reading. The research design also sought to avoid the limitations present in previous studies. Researchers have stated the need for experimental designs investigating the effectiveness of ER relative to grammar-translation at developing reading rates. This study demonstrated that it is possible to conduct classroom-based experimental research on reading with students randomly assigned to different treatment groups for homework. This investigation also tracked the amount of reading completed and provided stronger evidence than some previous reading rate studies that the ER reading materials were comprehended. Additionally, this study showed that the comprehension of timed-reading passages was not sacrificed at the cost of increased reading rate. Most critically, evidence that the participants in the two groups spent similar amounts of time conducting the two treatments supports the argument that ER is not merely more effective, but also more efficient at increasing reading rate than grammar-translation. Reviews of research into ER have found persuasive evidence that extensive reading conducted consistently over a sustained period will positively impact students’ reading abilities. Such findings, together with the results from this study into the efficiency and effectiveness of ER over more traditional grammar-translation methods, can serve to guide policy and practice decisions in the Asian context.

Acknowledgements The authors would like to thank the editor and reviewers for their valuable comments and suggestions. The authors would also like to thank Paul Nation and Brandon Kramer for their feedback on earlier versions of the paper. The authors have special thanks for David Beglar whose observations greatly improved the quality of the article. Any remaining faults are the sole responsibility of the authors. This work was supported by Japan Society for the Promotion of Science KAKENHI (grant number JP 16K16890). Appendix Summary of studies which investigated the influence of ER on reading rate.

104

S. McLean, G. Rouault / System 70 (2017) 92e106

Study

Participants/Context/Treatment/Materials

Japanese university students (N ¼ 125) were randomly assigned to an ER group which read an average of 641 pages of unsimplified text for American teenagers, and SRA cards, or to IR skill-building groups which worked through one or two 2e4 page readings in a textbook with one more section assigned for homework. Mason and Japanese university students. ER þ L1 Krashen summary (n ¼ 40), ER þ L2 summary (n ¼ 36). ER group participants conducted ER in class (1997) and outside of class (3e4 h) and wrote summaries in Japanese or English. Control group (n ¼ 38) conducted cloze tasks, read 32 four-page stories and took weekly vocabulary tests. Bell (2001) Yemeni English learners at the British Council. IR (n ¼ 12) read short texts and completed tasks focused on grammar and vocabulary. ER (n ¼ 14) conducted ER. Timeon-task was similar during the two-semester treatment period. Japanese university student volunteers. IR Taguchi, Takayasu- (n ¼ 10) conducted 24 sessions using 42 Maass, & segments taken from The Missing Madonna Gorsuch and Away Match. 57 pages were read 5 times (2004) each. ER (n ¼ 10) 733e901 min, between 3 and 6 books (mean of 205 pages). Iwahori Japanese high school students (n ¼ 33) read (2008) toward a goal of 28 graded readers during their term and summer vacation. Beglar, Hunt, Japanese university students. IR (n ¼ 17) in and Kite and out of class IR. Three ER (pleasure reading ¼ PR) treatment groups: PR1 (n ¼ 23) (2012) in class IR and out of class ER (136,000 standard words). PR2 (n ¼ 22) (158,000 standard words) and PR3 (n ¼ 35) (200,000 standard words) read six simplified texts in class and out of class ER. Robb and Susser (1989)

Huffman (2014)

Note.

1

Measurement method

Results

Words read within 1 min. and passage reading time. Pretests and posttests varied. Comprehension measures were taken.

The ER group read faster than the skills groups over a minute mean ER 86.55 (þ7.24 wpm1) vs IR 76.75 (1.75 wpm) F ¼ 4.580, p < 0.05) and read a passage faster (mean 336.39 seconds vs. 411.90 seconds, F ¼ 28.921, p < 0.001). Comprehension rates were sustained.

Participants read a required book before and ER þ L1 summary group (t (68) ¼ 12.76, p < 0.001) and ER þ L2 summary group after the treatment. Comprehension not (t(67) ¼ 5.70, p < 0.001) increased reading measured. rate gains more than the control group.

Pretest and posttest consisting of two texts. Words read over 3 min. Comprehension assessed through 3 different passages.

Nonequivalent pretests and posttests were read 5 times before and after treatments. First, third and fifth readings were used. Comprehension scores were collected.

1-minute reading probe with the same text for pretest and posttest

Reading rate: IR þ 14.09 wpm ER þ 59.43 wpm Adjusted comprehension scores: IR þ 11.36% ER þ 33.72% Rate: *RR: 3 wpm (85e82) ER: 17 wpm (81e64) Comprehension score very low (K ¼ 16): *RR: þ (1.60e3.90) ER: þ (1.90e4.50) * ¼ Repeated reading Significant increase in reading rate, Pre 84.18 to Post 112.82 ¼ þ28.64 wpm.

While sustaining comprehension with no significant increase, the ER (pleasure reading) groups experienced significantly greater reading rate gains relative to the IR Group (t(38.77) ¼ 4.18, p < 0.001, r ¼ 0.56) PR1 þ8.02 std wpm2 PR2 þ12.84 std wpm PR3 þ16.84 std wpm IR þ2.97 std wpm Japanese university students. IR (n ¼ 32) read 3 pretests and 3 posttests reading passages While sustaining comprehension with no 9682 standard words. ER (n ¼ 34) read 80,202 were different. 1 week prior to pretest only significant decrease, the ER groups standard words. ER group spent significantly the ER students conducted 1 practice reading. experienced significantly greater reading rate gains relative to the IR Group longer reading than IR group. Only ER (t(64) ¼ 5.97, p ¼ 0.000) students conducted weekly timed-reading ER þ20.73 std wpm practice. IR -0.62 std wpm

¼ words per minute,

2

4 passages over 2 weeks as pretests and posttests. Prior to the pretest students conducted 4 timed readings to reduce the practice effect.

¼ standard words per minute.

References Al-Homoud, F., & Schmitt, N. (2009). Extensive reading and a challenging environment: A comparison of extensive and intensive reading approaches in Saudi Arabia. Language Teaching Research, 13, 383e401. https://doi.org/10.1177/1362168809341508. Alderson, J. C. (1993). The relationship between grammar and reading in an English for academic purposes test battery. In D. Douglas, & C. Chapelle (Eds.), A new decade of language testing research (pp. 203e214). Washington, DC: TESOL. Alderson, J. C. (2000). Assessing reading. Cambridge, England: Cambridge University Press. Anderson, N. J. (2008). Practical english language teaching: Reading. New York, NY: McGraw-Hill. Beglar, D., Hunt, A., & Kite, Y. (2012). The effect of pleasure reading on Japanese university EFL learners' reading rates. Language Learning, 62, 665e703. https://doi.org/10.1111/j.1467-9922.2011.00651.x. Bell, T. (2001). Extensive reading: Speed and comprehension. The Reading Matrix, 1(1). Retrieved from http://www.readingmatrix.com/articles/bell/index. html. Bismoko, J., & Nation, I. S. P. (1974). English reading speed and the mother-tongue or national language. RELC Journal, 5, 86e89. https://doi.org/10.1177/ 003368827400500109. Breznitz, Z. (2006). Fluency in reading: Synchronization of processes. Mahwah, NJ: Erlbaum. Carver, R. P. (1982). Optimal rate of reading prose. Reading Research Quarterly, 18, 56e88. https://doi.org/10.2307/747538. Carver, R. P. (1990). Reading rate: A review of research and theory. New York, NY: Academic Press.

S. McLean, G. Rouault / System 70 (2017) 92e106

105

Carver, R. P. (1993). The case against statistical significance testing revisited. Journal of Experimental Education, 61, 287e292. https://doi.org/10.1080/ 00220973.1993.10806591. Carver, R. P. (2000). The causes of high and low reading achievement. Mahwah, NJ: Routledge. Chang, A., & Millet, S. (2013). Improving reading rates and comprehension through timed repeated reading. Reading in a Foreign Language, 25, 126e148. Chung, M., & Nation, I. S. P. (2006). The effect of a speed reading course. English Teaching, 61, 181e204. Clark, I. L. (2003). Concepts in composition: Theory and practice in the teaching of writing. Mahwah, NJ: Erlbaum. Cramer, S. (1975). Increasing reading speed in English or in the national language. RELC Journal, 6, 19e23. https://doi.org/10.1177/003368827500600203. Day, R., & Bamford, J. (1998). Extensive reading in the second language classroom. Cambridge, England: Cambridge University Press. De Soto, J. L., & De Soto, C. B. (1983). Relationship of reading achievement to verbal processing abilities. Journal of Educational Psychology, 75, 116e127. https://doi.org/10.1037/0022-0663.75.1.116. Extensive Reading Foundation. (n.d.). ERF Graded Reader Scale. Retrieved May 19, 2016 from http://erfoundation.org/wordpress/graded-readers/erfgraded-reader-scale/. Fraser, C. A. (2007). Reading rate in L1 Mandarin Chinese and L2 English across five reading tasks. Modern Language Journal, 91, 372e394. https://doi.org/10. 1111/j.1540-4781.2007.00587.x. Fuchs, S., Fuchs, D., & Maxwell, L. (1988). The validity of informal reading comprehension measures. RASE: Remedial and Special Education, 9(2), 20e28. https://doi.org/10.1177/074193258800900206. Grabe, W. (2009a). Epilogue: Reflections on second language reading research and instruction. In Z. H. Han, & N. J. Anderson (Eds.), Second language reading research and instruction: Crossing the boundaries (pp. 192e205). Ann Arbor: University of Michigan Press. Grabe, W. (2009b). Reading in a second language: Moving from theory to practice. New York, NY: Cambridge University Press. Grabe, W., & Stoller, F. L. (2002). Teaching and researching reading. Harlow, England: Pearson Education. Grabe, W., & Stoller, F. L. (2011). Teaching and researching reading (2nd ed.). Harlow, England: Pearson. Hamada, M., & Koda, K. (2008). Influence of first language orthographic experience on second language decoding and word learning. Language Learning, 58, 1e31. https://doi.org/: 10.1111/j.1467e9922.2007.00433.x. Hino, N. (1988). Yakudoku, Japan's dominant tradition in foreign language learning. JALT Journal, 10(1, 2), 45e55. Hoover, M. L., & Dwivedi, V. D. (1998). Syntactic processing by skilled bilinguals. Language Learning, 48, 1e29. https://doi.org/:10.1111/1467-9922.00031. Huffman, J. (2014). Reading rate gains during a one-semester extensive reading course. Reading in a Foreign Language, 26(2), 17e33. Iwahori, Y. (2008). Developing reading fluency: A study of extensive reading in EFL. Reading in a Foreign Language, 20, 70e91. Kikuchi, K., & Browne, C. (2009). English educational policy for high school in Japan- Ideals vs Reality. RELC Journal, 40, 172e191. https://doi.org/:10.1177/ 0033688209105865. Kizuka, M. (2006). Professionalism in English language education in Japan. English Language Teacher Education and Development, 9(1), 55e62. Retrieved from http://www2.warwick.ac.uk/fac/soc/celte/research/eltd/volume_9_final_version.pdf. Klauda, S. L., & Guthrie, J. T. (2008). Relationships of three components of reading fluency to reading comprehension. Journal of Educational Psychology, 100, 310e321. https://doi.org/:10.1037/0022-0663.100.2.310. Koda, K. (2005). Insights into second language reading: A cross-linguistic approach. New York, NY: Cambridge University Press. Kramer, B., & McLean, S. L2 reading rate and word length: The necessity of character-based measurement. Manuscript in progress. LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 293e323. https://doi.org/:10. 1016/0010-0285(74)90015-2. Lightbown, P. M., & Spada, N. (2006). How languages are learned (3rd ed.). Oxford, England: Oxford University Press. Liu, N. F., & Littlewood, W. (1997). Why do many students appear reluctant to participate in classroom learning discourse? System, 25, 371e384. https://doi. org/:10.1016/S0346-251X(97)00029-8. Macalister, J. (2008). The effect of a speed reading course in an English as a Second Language environment. The TESOLANZ Journal, 16, 23e33. Macalister, J. (2010). Speed reading courses and their effect on reading authentic texts: A preliminary investigation. Reading in a Foreign Language, 22, 104e116. Mason, B., & Krashen, S. (1997). Extensive reading in English as a foreign language. System, 25, 91e102. https://doi.org/:10.1016/S0346-251X(96)00063-2. McLean, S. (2014). Evaluation of the cognitive and affective advantages of the Foundations Reading Library series. Journal of Extensive Reading, 2, 1e14. McMillion, A., & Shaw, P. (2016). Reading proficiency in advanced L2 users. In K. Hyltenstam (Ed.), Advanced proficiency and exceptional ability in second languages (pp. 149e184). Berlin, Germany: De Gruyter. Millett, S. (2005). New Zealand speed reading for ESL learners. Book 1. English Language Institute Occasional Publication. No.19. Moodle Reader Module [Computer software]. Available at: https://moodle.org/plugins/mod_reader Nakanishi, T. (2015). A meta-analysis of extensive reading research. TESOL Quarterly, 49, 6e37. https://doi.org/10.1002/tesq.157. Nakanishi, T., & Ueda, A. (2011). Extensive reading and the effect of shadowing. Reading in a Foreign Language, 23, 1e16. Nathan, R. G., & Stanovich, K. E. (1991). The causes and consequences of differences in reading fluency. Theory Into Practice, 30, 176e184. https://doi.org/:10. 1080/00405849109543498. Nation, I. S. P. (2009). Teaching ESL/EFL reading and writing. London, England: Routledge. Nation, I. S. P. (2012). The BNC/COCA word family lists (17 September 2012). Unpublished paper. Available at: www.victoria.ac.nz/lals/about/staff/paul-nation. Nation, P., & Waring, R. (2011). The Compass media guide to extensive reading. Seoul, Korea: Compass. Nuttall, C. (2005). Teaching reading skills in a foreign language (2nd ed.). Oxford, England: Heinemann Educational Books. Palmer, H. E. (1917). The scientific study and teaching of languages. London, England: Harrap. Pazaver, A., & Wang, H. (2009). Asian students' perceptions of grammar teaching in the ESL classroom. The International Journal of Language, Society and Culture, 27, 27e35. Plonsky, L., & Oswald, F. L. (2014). How big is “big”? Interpreting effect sizes in L2 research. Language Learning, 64, 878e912. https://doi.org/:10.1111/lang. 12079. Quinn, E., Nation, I. S. P., & Millett, S. (2007). Asian and Pacific speed readings for ESL learners. English Language Institute Occasional Publication, No.24. Rao, Z. H. (2001). Matching teaching styles with learning styles in East Asian contexts. The Internet TESL Journal, VII(7). Robb, T. N., & Susser, B. (1989). Extensive reading vs. skills building in an EFL context. Reading in a Foreign Language, 5, 239e251. Samuels, S. J. (2006). Reading fluency: Its past, present, and future. In T. V. Rasinski, C. Blachowicz, & K. Lems (Eds.), Fluency instruction: Research-based best practices (pp. 7e20). New York, NY: The Guilford Press. Shimada, H., Hisasue, A., & Yoneyama, T. (2011a). 英語長文出題パターン演習 1 [Practice common question patterns in long-English-passage type questions 1]. Nagoya, Japan: Kawai. Shimada, H., Hisasue, A., & Yoneyama, T. (2011b). 英語長文出題パターン演習 2 [Practice common question patterns in long-English-passage type questions 2]. Nagoya, Japan: Kawai. Song, B. (1995). What does reading mean for East Asian students? College ESL, 5(2), 35e48. Stanovich, K., & Stanovich, P. (1999). How research might inform the debate about early reading acquisition. Journal of Research in Reading, 18, 87e105. https://doi.org/:10.1111/j.1467-9817.1995.tb00075.x. Taguchi, E., Takayasu-Maass, M., & Gorsuch, G. J. (2004). Developing reading fluency in EFL: How assisted repeated reading and extensive reading affect fluency development. Reading in a Foreign Language, 16, 70e94. Tsukamoto, M., & Tsujioka, N. (2013). Teaching English through English to senior high school students in Japan: Towards the implementation of the new course of study. Shitennoji University Bulletin, 55, 309e324.

106

S. McLean, G. Rouault / System 70 (2017) 92e106

Vardanjani, A. M. (2013). A critical study of Iranian EFL environment. International Journal of English Language & Translation Studies, 1(3), 4e19. Retrieved from http://www.eltsjournal.org. Wolf, M., & Katzir-Cohen, T. (2001). Reading fluency and its intervention. Scientific Studies of Reading, 5, 211e240. https://doi.org/10.1207/ S1532799XSSR0503_2. Yen, T. T.-N. (2012). The effects of a speed reading course and speed transfer to other types of texts. RELC Journal, 43, 23e37. https://doi.org/10.1177/ 0033688212439996.