English Teaching 56(2) 2001

Computer Training to Improve Word Recognition and Reading Speed

Everette Busbee

(Jeonju University)

Busbee, Everette. (2001). Computer Training to Improve Word Recognition and Reading Speed. English Teaching, 56(2). 

EFL learners tend to read very slowly, which diminishes the pleasure derived from reading, which in turn leads to less reading and thus even poorer reading skills. The failure to read English also shuts down an excellent source for the acquisition of vocabulary and of English itself. Nevertheless, EFL reading instruction focuses almost exclusively on comprehension. This paper reports on research into the effectiveness of computer-based speed-reading training, including word-recognition training, for Korean college EFL learners. Subjects made gains in reading speed, although simplified reading material was used to allow true reading rather than labored translation. The gains, while not supporting the use of computers over traditional paper-based speed-reading training, do suggest a way for English teachers to use the computers sometimes placed at their disposal.

I. INTRODUCTION

EFL students often read at an extremely slow speed. Jensen (1986, p. 106) concludes that “at the end of a reading course, even advanced ESL students may read only 100 words per minute or less.” Segalowitz, Poulsen, and Komoda (1991, p. 15) contend that the L2 reading rates of “highly bilingual”  readers are “30% or more slower than L1 reading rates.” Weber (1991) echoes this, noting that even highly-skilled bilinguals typically read more slowly in their second language.

Studies in Korea and Japan suggest that Jensen’s (1986) figure of 100 Words Per Minute (WPM) is an accurate figure. Song (1999), in research aimed at reading strategies rather than speeds, presents data on how first-year students at Seoul National University respond to simple English reading material appropriate for American high sophomores. These students read at what appears to be 75-100 WPM. Kitao and Kitao (1995) report that a group of Japanese college students who were asked to quickly read a text read 105 WPM. Even at these speeds the issue of comprehension looms large, with both Song and Kitao and Kitao reporting a comprehension rate of about 40%, if corrected for guessing by equating to zero the score predicted by chance. If these reading rates are multiplied by the comprehension rate to give what Jackson and McClelland (1979) call an Effective Reading Rate (ERR), it appears that the students in the two studies were effectively reading at perhaps 10% of the efficiency of the 300 WPM reported by Nuttall (1996) for Americans of average intelligence and education. Nevertheless, Kitao and Kitao (1995) report, Japanese feel strongly that although they cannot speak English well, they can read well, a belief that the researchers label as a myth, in that Japanese can neither speak nor read English well.

Such poor reading skills, while sufficient for doing well enough on a test to gain admission to a good college, have little practical significance. It could of course be that this level of reading among college freshmen is merely a step along the way to reading proficiency, but this does not appear to be the case. It is well known among Koreans who attend overseas universities that reading assignments are a great burden, and these students are usually Korea’s most successful English learners.

The importance of English text as a source of information is expanding, and Koreans in such areas as economics, technology, and medicine, particularly those who are specialists, rely a great deal on printed English from abroad to stay abreast of changes. English is taking on even greater importance as professional publications meant primarily for domestic consumption include more and more English articles. The quantity of English that Koreans are expected to read will continue to grow, and this quantity will require faster reading.

Therefore, what the pioneer in reading research Huey (1908, p. 8) wrote almost a century ago is even more important today: “Reading is the means by which the world does a large part of its work.... The slightest improvement either in the page or in the method of reading means a great service to the human race.” Relating this directly to our field, and focusing on the need to read ever larger amounts of English, we can paraphrase the second sentence as “Even a small improvement in the speed of reading English, as long as it is unaccompanied by a loss in comprehension, would be a great service to EFL students.”

Nevertheless, what little EFL reading instruction that exists on the college level in Korea is devoted almost exclusively to improving comprehension. Academic counselors at American colleges, however, are aware of the benefits of training for faster reading, and often suggest that students take a short speed-reading course, either paper-based or computer-based. Such training for Korean college students, the author hypothesized, would be an effective use of class time, especially given the training’s short duration. The native speaker goals for speed-reading training are speeding up word-recognition and breaking the habit of reading slowly. Training for Koreans would have two additional goals. One would be getting students used to reading pages rather than sentences, and the other, perhaps most important, would be fostering the awareness that faster reading is a worthy goal, thus improving the motivation to read faster, and one of the greatest factors in the success of speed-reading training is learner motivation.

To test the effectiveness of speed reading training for Korean college students, I undertook research on students at my university, and this paper reports on that research. Contemporary speed reading training normally involves either printed texts or computers, and I chose the latter for two reasons. One was the availability of a well-stocked College of Education computer lab and an administration desirous of its use. The other was the my interest in discovering new ways to utilize the computers we English teachers are nowadays being asked to use. Although I believe that the computer and the Internet will not revolutionize language teaching (Busbee, 2001), I have for seven years taught numerous computer courses, with one of those courses being a Web Design content course (described in Busbee, 2000) in which English benefits derive from using a great deal of English exclusively for communication (Busbee, 1998). I therefore chose computer-based speed-reading training.

Research reports on computer speed-reading software for native speaker reading students are rare, probably because there exists no practical way to standardize text to a degree that allows quantitative cross-study comparisons. Wepner, Feeley and Wilde (1989) compared the commercial reading package Speed Reader with traditional paper-based speed-reading programs, but found no difference. As for computer-based speed-reading programs being used in EFL classes, I have found nothing at all in the literature.

1. EFL’s Lack of Concern for Reading Speed

To read well, a person must have two primary skills, the ability to comprehend what is being read, and the ability to do it at a reasonable speed. However, the field of EFL is almost exclusively concerned with comprehension. This is attested to by the  author’s failure in a search of 14 beginner to advanced EFL reading texts to find a single occurrence of “reading speed” in either the indices or the introductions. In addition, a careful look at chapters or sections on reading in seven texts for training language teachers found only one text expressing concern for reading speed (Nuttall, 1996).

Four possible explanations for English teaching’s disregard of reading speeds are: 1) English teachers in general view stress as an enemy of language learning, and pushing students to read faster when they are struggling for comprehension would tend to create considerable stress. 2) Measuring reading comprehension is simple, requiring only the administering of a multiple choice test, but measuring reading speed is complex. It involves the measurement of time required by an individual (not a class) to read a text that has been standardized for difficulty level of vocabulary and grammar, and in addition requires a comprehension test. 3) Teachers may feel that EFL students would in general not benefit from speed training because they lack the skills necessary for true reading, which is quite different from painstaking translating. 4) Whereas reading was once considered the most important EFL skill, it is now the poor cousin of the conversation skills, speaking and listening. Reading speed may be viewed as simply having little priority.

2. Causes of Slow Reading

In order for students to increase their reading speed, they should be aware of the causes of slow reading. The following causes have been summarized from Goodman (1996), Just and Carpenter (1992), and Carver (1990): 1) simply reading slowly because it is a habit, 2) deliberately reading slowly in the hopes of increasing comprehension, 3) recognizing words slowly, 4) reading word-by-word rather than in groups of words that fit together logically, as in prepositional phrases, 5) vocalizing, that is, quietly reading out loud, or subvocalizing, that is, saying the words in the mind, either simply as a habit or as a conscious attempt to increase comprehension, 6) often darting back (regressing) to get another look at a word or several words (due to simple habit or to a lack of the ability to maintain concentration, and 7) reading little or nothing on a regular basis. Computer-based speed reading training can counter these, especially if accompanied by lectures designed to alter reader attitudes.

However, comprehension must be maintained. Twining (1991) lists four levels at which comprehension can be lost if reading too fast: 1) failure to understand a word, 2) failure to understand a sentence, 3) failure to understand how sentences relate to one another, and 4) failure to understand how the information fits together in a meaningful way. The first two are mechanical decoding, or bottom-up skills, and the last two are conceptually-driven, or top-down skills. It is now being recognized in our field that teaching top-down reading skills has hardly been as effective as once hoped (Lee, 2000), and bottom-up skills are again being stressed. However, good readers make use of both top-down and bottom-up strategies, depending on the situation, task, and text (Bernhardt, 1991; Carrell, Devine, & Eskey, 1989). Grabe (1991) stresses the interaction of these two types of cognitive skills, which he terms identification (lower level skills that allow rapid and automatic recognition of words and grammatical forms) and interpretation (higher level skills that allow comprehension and interpretation).

The first element of decoding concerns vocabulary, which is thought to be the single most important factor in reading comprehension (Just & Carpenter, 1992), and the second element involves internal grammar. It is at this level of word recognition (of known words) and sentence comprehension that a speed-reading program comes into play. The goal is to increase decoding efficiency by presenting text for reading practice at a speed that pushes the reader to recognize words and comprehend sentences at a gradually faster rate. Top-down advocates will be pleased to note that improved decoding skills will provide more information for effective top-down processing, and will also free cognitive resources that can be allotted to higher-order activities.

3. The Effects of Slow Reading

In spite of the almost total focus of EFL reading instruction on comprehension, faster reading is a worthy goal, for the effects of slow reading are pernicious, affecting even the acquisition of English. One of the greatest challenges for English teachers is providing students with sufficient comprehensible input for acquisition to take place, and reading simplified texts for pleasure is an excellent source of such input (Krashen, 1983). In addition, reading is the best way to develop a long-lasting vocabulary that is understood on a level deeper than a vocabulary derived from memorizing simple L1-L2 equivalencies. Slow reading not only cuts down on the acquisition of vocabulary and of English, it diminishes the motivation to read. Nuttall (1996, p. 127) uses the term “vicious cycle of the weak reader” for the slowing down of reading by EFL students in an attempt to increase comprehension, only to find that they no longer enjoy reading because it takes so much time, and therefore read less, thus becoming even poorer readers.

Clearly, one goal EFL teachers should have for their students is faster reading. While faster reading is facilitated by long-term frequent reading and extensive vocabulary study, the time that must be devoted to these activities, if successful, is measured in years. This is by no means a dismissal of vocabulary study or long-term frequent reading. On the contrary, vocabulary is critical for comprehension, and one factor surely at play in slow reading among Korean and Japanese EFL students is that they spend little time if any reading English for pleasure or for getting information. It is highly likely that a sizable vocabulary and copious reading (not translating but true reading) are necessary for EFL students to become rapid readers with high comprehension.

II. SPEED-READING TRAINING

While vocabulary and frequent reading may be necessary for the development of rapid reading, they may not be sufficient: many native speakers who frequently read and have good vocabularies are slow readers. However, pushing slow readers to read faster without maintaining comprehension is not the answer: it would merely mean that a greater amount of text has been misunderstood.

Consequently, there have arisen courses in training slow readers to read faster. Increases in speed are gradual so as to minimize any sacrifice in comprehension. Such training is pursued by a number of college students in the United States, and colleges often offer such courses, at times even for credit. If native speakers must at times work on their reading speed, it seems reasonable that EFL learners might also need help with their reading speed, although EFL students whose grasp of English is insufficient for true reading would of course gain little or nothing from speed reading training.

1. Speed Reading and the Mechanics of Reading

Slow reading in a native speaker is looked upon largely as the result of habitually slow and disorganized eye mechanics that have been acquired over the years. Korean middle and high school EFL students have surely acquired similar bad habits during their six years of deciphering primarily sentences, at times a paragraph, and rarely more than a few short paragraphs. The results of inefficient reading will be encountered immediately by a student assigned to read, say, a ten-page simplified short story, but the small quantity of English actually encountered by middle and high school students requires little efficiency. As a result, poor reading habits have few immediate consequences. In addition, reading instruction tends to reinforce poor habits, in that it consists primarily of teacher-led translation focussing on vocabulary and grammar, with single sentences often discussed  two or three minutes or longer. Students develop a labored “translating” style of reading, and their eyes rarely acquire the discipline necessary to move smoothly and efficiently across a line of text and down a page without backtracking. Their eyes instead move back and forth through a sentence to identify the subject, the verb, and the object, and then again move back and forth through the sentence to associate modifiers with the subject, verb, and object. Finally, the eyes will likely move across the sentence yet again to confirm the translation. This habitual checking and rechecking becomes the way of reading every sentence, even straightforward S-V-O sentences with simple modifiers.

Native-speaker speed-reading courses are designed to overcome poor reading habits by increasing the efficiency of eye mechanics. Another goal, equally important in that it strongly affects the motivation to read faster, is instilling a deep awareness that faster reading is beneficial. Just as Japanese college EFL students, as reported by Kitao and Kitao (1995), are unaware that their slow reading clearly places them within the ranks of poor readers, Korean college students seem to be unaware that faster reading should be a goal. Awareness is often the first step toward change.

In addition, due to a paucity of reading experience, Korean college EFL students often lack the mental stamina required to store information in short-term memory long enough to make sense out of what is being read. This effectively precludes sustained reading in English. Speed reading training often uses texts that require five to ten minutes to read, and so build stamina.

Given the small investment in time and effort required, if a speed reading course could increase the reading speeds of upper intermediate and advanced EFL students by just 10 or 20%, it would be a worthwhile investment.

2. Questioning Speed Reading

Nuttall (1996), writing with a vagueness indicative of the problems of  discussing reading speeds objectively, states that a native speaker of English of average education and intelligence, when asked to read typical material for high comprehension, reads at about 300 WPM. Commercial speed-reading companies generally claim their training increases this to 1000 WPM, and claims of 3,000 WPM are not uncommon. Reading experts view these claims as unrealistic, and Carver (1990) even goes so far as to use speed reading and skimming interchangeably, because they both entail ignoring some words altogether, and so both lead to lower comprehension. Carver also notes that for trained speed readers and naturally fluent readers, comprehension suffers as rates pass 600 WPM. There is an obvious danger in reading too fast.

In addition, some material is more appropriate for fast reading. A complex example of a computer programming technique or an intricate philosophical argument are usually read slowly, pondered, and read again, perhaps several times, while novels, biographies, light history, and newspapers are good candidates for faster reading.

However, a reader can learn to match the speed to the material, and if speed reading is approached in a reasonable manner, there should be few concerns about loss of comprehension. In fact, because slower reading requires that text to be held in short term memory for longer periods, slow reading can lead to lower comprehension. Nevertheless, Kitao and Kitao (1995) report that Japanese students believe slower reading improves comprehension, a belief the authors label as a myth.

Another concern about speed reading is that gains may be fragile, and that bad habits may resurface. However, a goal of training is to help readers see the benefits of faster reading and the possibility of achieving it, and motivated readers would tend to keep pushing their reading speed.

A problem specifically addressing language learners is that for them training for faster reading is more complicated than for native speakers. While vocabulary limitations can at times be important in remedial reading programs, a typical American college student has a vocabulary and an internal grammar sufficient for rapid reading. For Korean college EFL students, however, vocabulary and grammar limitations are surely the most important factors in limiting reading speed, and of course Korean students can also have every bad habit an American student has.

In spite of these questions surrounding training in speed reading, people who take speed-reading courses evidently feel that their expectations are met to a reasonable degree, because many American colleges have for decades offered non-credit and even credit courses on how to read faster, some with computers, and these courses are popular enough to continue being offered.

One concern specific to computer-based reading needs to be addressed. The low quality of on-screen text at one time limited reading speeds, with Hansen and Haas (1988) stating that reading took 25% longer on screen than on paper.  However, within three years monitors had improved to such a degree that Muter and Maurutto (1991) were reporting that with a large high-resolution screen, reading from a computer could be as efficient as reading from a book. Adler-Kassner and Reynolds (1996) and Arroyo (1992) found no difference between computer and book reading.

III. WORD RECOGNITION TRAINING

A standard component of speed-reading software is practice in word recognition. Davis and Lyman-Hager (1997) break word recognition into two processes, decoding, which allows a word to be pronounced but gives it no meaning, and true word recognition, which matches the decoded word with a known word. Decoding appears to have a central role, because eye-movement studies show the eyes fixate on nearly every word rather than just scanning across a page (Rayner & Pollatsek, 1989). Gough (1984) feels that the unpredictable nature of words in context means context is not particularly important for good readers, that they probably read via a process involving the words. Lyon (1995), part of a growing consensus, maintains that reading problems in the L1 are primarily centered at the level of the single word, which is supported by Luppescu and Day’s (1993) conclusion that better readers tend to have larger vocabularies. The level of the word is also central to foreign language reading proficiency, in that Laufers (1997) found that L1 reading skills do not transfer to an L2 until about 3,000 headwords are known. Sparks et al. (1997) go so far as to say that one’s general level in a foreign language is highly correlated with one’s ability at the level of text decoding.

A good knowledge of English at the lowest level that carries meaning, the word level, is clearly necessary for good reading, and Korean and Japanese students are well-known for their large passive English vocabularies. Also required is a good working grasp of English grammar, and Korean and Japanese students tend to have quite a bit of grammar. However, the goal of decoding and recognition is automaticity, and few Korean or Japanese EFL readers seem to demonstrate this automaticity.

 Perfetti (1994, p. 849) refers to automaticity as “basic language ‘instincts,’” and maintains that they account for a substantial portion of those differences in reading ability that are not intellectual differences. Automaticity involves the highest level at which a word can be known. Automatic word recognition can come about through copious reading, which gives massive practice in recognition. It can also be assisted by training based on word-flashing software that presents word lists one word at a time, with the time on screen gradually decreased to force more rapid decoding/recognition.

 Research aimed at obtaining recognition-time data that has cross-study value is complex. Lamings (1968) reports that while simple reaction times average 220 msec., recognition reaction times average 384 msec, with the additional 164 msec. being the actual recognition times. However, word recognition research is complicated by the fact that, whereas simple reaction time to a stimulus such as a flashing light is easy to measure, it is difficult to determine exactly when a flashed word has been recognized, although it can be approximated by a complex order and timing of word presentation. In addition, common words such as thank are recognized more quickly than uncommon words such as plunge, and shorter words tend to be recognized more quickly than longer words, with commonness and length interacting in such a way that Saturday is likely to be recognized more quickly than quince. The combination of characters is also important, and queenly, though less common, may be recognized more rapidly than quickly. Even the shape of the word can be important, so that sudden is easier to recognize than success. A further complication is that when determining recognition time indirectly by the length of time a word is fixated during actual reading, length of fixation varies with the predictability of a word from its context (Rayner & Well, 1996). However, while these subtleties are of importance to cognitive scientists who need data with cross-study value, English teachers would be more interested in within-study changes that suggest a concrete way to teach in the classroom.

IV. MATERIALS AND METHODS

This research consists of two parts, speed-reading training with a word-flashing component, and word-flashing training that is self contained.

1. Speed-Reading Training

This research was conducted in what was officially a 2nd-semester (and therefore abbreviated) 4th-year English education major vocabulary class, but prior to enrolling, all members agreed to participate in the research, appearing genuinely interested in the possibility of increasing their reading speed. The class members, all who knew each other well, worked in pairs throughout the project, in order to help each other with technical problems and introduce a friendly competitiveness.

1) Subjects and Project Schedule

The 10 subjects, all computer-literate, were selected from the 15 students who enrolled in this class. I told the students that I needed 100% attendance and punctuality for the 14 class periods from the second through the eighth weeks. If they could not agree to this, I strongly requested they drop the course. Three students, however, were transfer students who were not competent in computers, and consequently their poor attendance excluded them from the study. In addition, two other students were excluded because they had scored over 550 on the TOEFL and the reading material for the training was too simple for them.

The training was conducted twice a week for seven weeks, the second through the eighth week of the semester, and took place in a computer lab that was the assigned classroom. The lab was free for the 30 minutes following this class, and because I had a close relationship with the students, and because I gave no homework and the semester ended quite early, the students agreed to include this time in their class. The research therefore consisted of 14 110-minute sessions with a ten-minute break.

2) Software

The speed reading program Reading Edge was used because it is full-featured and is free. It allows a teacher/researcher to use any text file for reading practice and tests, to produce sets of multiple-choice questions to test comprehension, and to give tests and record scores. It also allows student/subject control of the speed of text presentation during practice, as well as the duration of word flashing. The program is copyrighted 1996, but its Web site has disappeared, so Reading Edge has no support, nor does it have online help. Nevertheless, anyone familiar with other reading software will be able to figure it out in about an hour.

3) Text

Choice of text in reading research is determined by the goal of the research, and the goal of this research was to determine if speed-reading training can assist EFL readers in developing the skills to read simple text more quickly. This could be native speaker text that is simple in its entirety, or simple sentences or paragraphs scattered throughout more difficult native speaker material. That is, an entire book does not have to be read slowly just because some of it must be read slowly. EFL readers will probably always encounter some material that must be labored over. However, speed-reading training involves true reading rather than the labored deciphering of texts, and for EFL students other than the genuinely advanced, true reading demands simplified text―ungraded text would break up the rhythm of reading and so prevent the development of smooth and disciplined eye movement. Simplified texts were therefore used for this research.

Another concern is that pre-and post-tests be of equal difficulty, so a single long simplified novel was used for testing. It was scanned with the software OmniPro v.8 and edited, and then broken into 28 sections for the 14 training sessions, with an average of 390.8 words per reading. Other simplified novels were used for practice material. The class consumed a great deal of text, and in a questionnaire at the end of the project, all students replied that they had read  “much more” English for this class than for any other class, although this was a curtailed 2nd-semester senior course with no homework.

4) Presentation of Text

Tests involved the presentation of text that remained on screen as a page until the subject clicked “done,” but most practice involved reading a text that was presented one line at a time, with subjects in charge of the speed of presentation. Subjects were urged to keep pushing the speed higher.

Near the beginning of each session subjects took a practice speed test that also functioned as reading practice, and the score was not normally included in the study. If, however, on the real speed test at the end of the session the subject missed more than one of the four questions on the comprehension test, indicating reading was too fast, the practice test score was substituted. If students missed more than one question on both tests, which rarely happened, both tests from the following period were used.

5) Word-Recognition Training as Part of Speed-Reading Training

With the assumption being that improving the speed of word recognition will improve the speed of reading, word-flashing by Reading Edge was included in training. Each class had three 11-minute word flashing sessions, at the beginning, after the mid-class break, and just before the test that ended the session. This allowed approximately 1000 words to be viewed each class, or 14,000 for the entire training.

The words initially came from Dolch’s (1936) list of the 220 most common words in the English language, with the most common being the. Because these words comprise from 50-70% of all written English, Dolch felt they should be learned so well that they could be recognized as unique single units rather than being read letter-by-letter. I also produced a list of 95 common nouns. The subjects felt the lists were too simple until the I demonstrated I could easily recognize all the words when flashed for 50 msec. but many students could recognize fewer than 50% at 100 msec.

When students felt they could comfortably recognize most of the Dolch Sight Words when presented randomly at 70 msec, they switched to Hill’s list of 1,500 most-common English words, which includes the Dolch lists. This list is still quite basic in that it includes function words, the inflected forms of verbs, and plurals.

2. Word-Recognition Training

1) Subjects

For the preliminary study, nine English education majors from 1st to 4th year were divided into three equal groups. Three were genuinely advanced students who had all scored over 550 (580 for one) on the TOEFL test (Advanced Group), three were intermediate students who had done well in a required two-semester sequence in advanced computers taught in English (Intermediate Group), and three were 1st-year students who, although they had completed six years of English study in middle and high school, were functionally high beginner (Beginner Group).

For the in-depth study, six intermediate students were added to the Intermediate Group, giving a total of nine. The additional subjects were given the same short treatment as the original Intermediate Group so the entire group would enter their training with the same experience.

2) Software

Two programs were used. One, Reading Acceleration Machine, was used to present a sequential list of words to subjects at a predetermined rate and with a predetermined presentation time (flash). This program, primarily for flashing vocabulary lists, has controls for speed per line, speed per word, and the time the screen is blanked between words.

The other software used was the word flasher of Reading Edge, which has three useful features not found in Reading Acceleration Machine. First, the user can at any time change the word’s time-on-screen. This allows a student to keep the speed on the edge of perception, but for a less ambitious student, it also allows words to be flashed at an undemanding speed. Second, the student decides when to flash the next word. Again, in the hands of a certain type of student, this is good, but the most efficient learning will take place if the students are pushed. Third, after seeing the word, the student is able to take another look at the word for immediate feedback. (Reading Edge’s lack of automatic advancement necessitated the use of Reading Acceleration Machine in this study.)

3) Word Lists

The vocabulary items came from three sources. The simple set of words (Beginner List) came from the General Service List of West (1953), and included all words of four or five letters from the 200th word (by frequency) through the 1000th. The intermediate set (Intermediate List) came from Black Beauty, by Anna Sewell (as a Project Gutenberg text file), and was processed as follows in less than an hour: In Concordancer for Windows v1.3 all words occurring from one to nine times were selected and alphabetized. In MS Word all inflected endings―s, ’s, es, ed, er, ers, ly, ing, and ings―were removed using find/replace, and the spell checker was used to eliminate duplicates. Words of 6, 7, or 8 characters were selected, and then irregular verb forms, words with a quaint or dated feel, and specialized words relating to horses were deleted.

This gave a list of 695 words that was divided into three working lists (Lists A, B, and C) of 200 words each, with the last 95 words discarded. Each list was put into five different random orders, giving a total of 15 files.

The third list, quite difficult, (Advanced List) was the 800 word University Word List produced by Xue and Nation (1984).

4) Word Presentation

During the preliminary study subjects were tested in my office one at a time. Words were presented with a two-second pause between them, during which subjects read the word. I sat by the subject, and on a printout of the word list recorded whether the answer was correct or incorrect.

Practice during the in-depth study involving the large Intermediate Group took place in a computer lab, but the subjects were also tested in my office one at a time.

The nine subjects in this group agreed to meet as a group in a computer lab for 11 30-minute sessions over a two-week period, which included some sessions twice a day, but these were always separated by at least two hours. Subjects practiced 400 words each session, two of the 200-word files from the 600-word list. The first file was viewed with Reading Edge, which allowed confirmation of recognition, but the second file was viewed with Reading Acceleration Machine, which presented the words non-stop. The files were viewed systematically, so the 600 word list was covered seven times during the research. Each subject was given a personal 100-word recognition test before training began and after it concluded.

Monitors were all set at 1024x768, and the words were presented as black 20-point Times New Roman against a white background.

V. RESULTS

1. Speed Reading

Figure 1 gives the results of the reading tests. The results appear to be quite impressive, with the reading speed nearly tripling, but the graph is misleading, because almost half the gain came from an increase between the first and second tests. This is likely the result of a combination of two factors other than the effective training of disciplined eye movement. The first is that the students had become more comfortable with the use of the program and with reading English text on a monitor, and as an artifact, this should be factored out.

The second is that students had become aware that a casual attitude toward reading would hurt their scores. That is, simply acquiring an awareness that fast reading is the goal leads to faster reading. The portion of the increase due to this shift in student attitudes is therefore a real improvement, but to increase the probability that gaining familarity with the process was factored out, data from the first period was not used in the statistical analysis. Consequently, the second period reading scores were used as the pre-test and the last period for the post-test, and a paired t-test was applied to the data. With a mean for the pre-test of 92.9 WPM and for the post-test 146 WPM, a paired t-test gave t=-8.93, with p<0.005. All students showed improvement, with the smallest gain being 21 WPM.

2. Word Recognition

1) Preliminary Study

(1) Advanced Group

A 20-word list from the Advanced List was presented to the Advanced Group (who all scored over 550 on the TOEFL) at 50 milliseconds on screen. All three students scored from 80 to 90% correct. With another 20 words presented at 75 msec, the scores reached 90-100%. Each of the advanced students was then given a series of four approximately 5-10-minute reading speed tests (with the commercial speed-reading software for adults Ultimate Speed Reader). This software’s reading material is on many different levels but is unranked, so the results are of little value as other than a general indication of reading skills, but the three advanced students all read at over 250 WPM with from 75-100% comprehension. 

FIGURE 1

Changes in mean reading speed during training^*

^*Initial scores were ignored as a possible artifact arising from increased familiarity with the training process, and 2nd-session and 14th-session scores were treated as pre- and post-tests. A paired t-test gave t=-8.93, p<0.005.

(2) Intermediate Group

A list of 20 words from the Advanced List was presented to the Intermediate Group at 50 msec. on screen. They all complained about the speed, and generally were frustrated, saying all but the shortest words were hopeless. Correct responses were all 35% or less. When the on-screen time was increased to 75 msec, there was some improvement, and more improvement at 100 msec. It was not until 150 msec. that all students scored 75% or higher, although it is likely that some of this was “empty” word recognition, or reading a word without knowing its meaning. Presenting the word on screen for a longer duration did not help, with the students reporting that if the word was unfamiliar and long, they could not read it aloud in the allotted 2,000 msec. Their reading speeds were less than 90 WPM, with comprehension from 38-75%.

The Intermediate Group, as the scores suggest, was limited by both vocabulary and reading skills. I asked each member how often in their lives they had read English for either pleasure or to get information that would be of practical use in their lives. All answered never.　The interesting thing about this group, and the most important result of the preliminary study, is that it was the only group that increased its score appreciably as time on screen was increased. This suggests that members of this group knew many of the words, but were reading them letter by letter, suggesting that intermediate EFL learners might benefit most from training in word recognition. This group, enlarged with an additional six students with similar language skills, was therefore selected for training in basic word recognition.

(3) Beginner group

A 20-word list from the Advanced List was presented to the Beginner Group at 50 msec. on screen. Two students could recognize nothing, and the other one scored 20%. When the time on screen was finally increased to 200 msec., the correct responses were still 40% or less. The reading speed test was too difficult for the results to have meaning. The Beginner Group had so few reading skills and such little vocabulary that they were not ready for training in word recognition.

2) Experimental Study

While tests were easily administered during the speed reading training and in fact were part of the training, I had to administer word recognition tests to students one at a time, which precluded daily testing. I had planned to graph progress by recording the time-on-screen chosen by subjects for their training sessions However, students often asked to interrupt the flashing of a file to slow it down or speed it up, so I gave up on recording flashing duration.

Pre- and post-tests consisted of 100 words from the intermediate list flashed for a duration of 75 msec. Mean correct for the pre-test was 34.6 and for the post-test was 90.0. On a paired t-test t=-18.1, p<0.005.

VI. CONCLUSIONS

The broadest interpretation of the results is that a relatively small amount of training can help Korean EFL students learn to read significantly faster, at least if the material’s level allows true reading to occur. More specifically, about 25 hours of computer-based speed-reading training can help Korean college students increase computer reading speeds for simplified materials. Further, about five hours of computer-based training in word recognition can lead to much faster recognition, and good readers recognize words more quickly.

This in no way endorses computers as the best means of training students to read faster. An equal amount of paper-based training by a skilled teacher might well bring about similar gains, for Wepner, Feeley and Wilde (1989) found no difference between a commercial speed-reading computer package and a traditional paper-based speed-reading training program. In addition, it is likely that an equal amount of time spent intensely reading, unaccompanied by training, would have lead to noticeable gains. However, an advantage derived from computer-based training over individual reading is that the former holds students accountable, and accountability is often at the heart of academic achievement.

The results seem impressive, but it would have been quite surprising if such gains had not occurred. Reading speeds among Korean college students are so slow that improvement might come from such a simple thing as telling students to read a story and write down when they started and finished. In addition, computer speed-reading training teaches a very specific set of skills, and 25 hours of intensive training would likely improve performance in that specific set of skills: Words are flashed for ever briefer durations; text is presented line-by-line at a rate that is on the edge of comprehension, and then questions concerning that text are asked. It is like a computer game, and 25 hours on a game can lead to dramatic improvements.

Expanding the results into the actual reading speeds of EFL students leads to three questions that suggest further research: 1) How well do these computer-based gains transfer to paper reading? 2) How durable are the gains? 3) To what extent do gains for simplified material affect speeds for ungraded material?

 Whatever the answers to these questions turn out to be, they will in no way negate the broad conclusion of this research, that Korean college students have the potential to read faster than they do. Further, with the assumption that an entire book does not have to be read slowly just because some of it must be read slowly, this potential to read faster likely includes ungraded material. As to durability and transferability to paper reading, motivation to continue reading faster on paper is central, and the first step in acquiring motivation is seeing the possibilities. It sometimes seems that an English teacher’s main job is to get students to feel the possibilities strongly, thus arousing motivation sufficient for independent action. Students who participated in this study have gained the awareness that faster reading is possible with just a few hours of work, and although some of the gain in reading speed may decay, it is doubtful whether these students will ever again read as slowly as they once did.

REFERENCES

Adler-Kassner, L., & Reynolds, T. (1996). Computers, reading and basic writers: Online strategies for helping students with academic texts. Teaching English in the Two Year College, 23(3), 170-178.

Arroyo, C. (1992). What is the effect of extensive use of computers on the reading achievement scores of seventh grade students? (ERIC Document Reproduction Service No. ED 353544).

Bernhardt, E. (1991). Reading development in a second language. Norwood, NJ: Ablex.

Busbee, E. (1998). A small “content course” in phonology. English Teaching, 53(1), 41-71.

Busbee, E. (2000). A Web design content course that holds students accountable for classroom English. English Teaching 55(3), 247-267.

Busbee, E. (2001). The computer and the Internet: Are they really destined to play a major role in English teaching? English Teaching 56(1), 201-225.

Carrell, P., Pharis, B., & Liberto, J. (1989). Metacognitive strategy training for ESL reading. TESOL Quarterly, 23(4), 647-678.

Carver, R. (1990). Reading rate: A review of research and theory. San Diego, CA: Academic Press.

Davis, J., & Lyman-Hager, M. (1997). Computers and L2 reading: Student performance, student attitudes. Foreign Language Annals, 30(1), 58-72.

Dolch, E. (1936). A basic sight vocabulary. The Elementary School Journal, 36, 456-460.

Goodman, K. (1996). Ken Goodman on reading: A common-sense look at the nature of language and the science of reading. Portsmouth, NH: Heinemann.

Gough, P. (1984). Word recognition. In P. Pearson, R. Barr, M. Kamil, & P. Mosenthal (Eds.), Handbook of Reading Research, Vol. 1 (pp. 225-253). White Plains, NY: Longman.

Grabe, W. (1991). Current developments in second language reading research. TESOL Quarterly, 25(3), 375-406.

Hansen, W., & Haas, C. (1988). Reading and writing with computers: A framework for explaining differences in performance. Communications of the ACM, 31(9), 1080-1089.

Huey, E. (1908). The psychology and pedagogy of reading. Boston, MS: MIT Press.

Jackson, M., & McClelland, J. (1979). Processing determinants of reading speed. Journal of Verbal Learning and Verbal Behavior, 108, 151-181.

Jensen, L. (1986). Advanced reading skills in a comprehensive course. In F. Dubin, D. Eskey, and W. Grabe. (Eds.) Teaching second language reading for academic purpose (pp. 103-124). Reading, MA: Addison-Wesley.

Just, M., & Carpenter, P. (1992). A capacity theory of comprehension: Individual differences in working memory. Psychological Review, 99, 122-149.

Kitao, K., & Kitao, S. (1995). English teaching: Theory, research, practice. Tokyo: Eichosha.

Krashen, S. (1983). Applications of psychological research for the classroom. In C. James (Ed.), Practical applications of research in foreigh language teaching (pp. 51-66). Skokie, IL: National Textbook Co.

Laming, D. (1968). Information Theory of Choice-Reaction Times. London: Academic Press.

Laufer, B. (1997). The lexical plight in second language reading. In J. Coady & T. Huckin (Eds.), Second language vocabulary acquisition (20-34). Cambridge: Cambridge Univ. Press.

Lee, J. W. (2000). Reading methods revisited: Are we going back to bottom-up models? English Teaching, 55(4), 191-211.

Luppescu, S., & Day, R. (1993). Reading, dictionaries, and vocabulary learning. Language Learning, 43, 263-287.

Lyon, G. (1995). Toward a definition of dyslexia. Annals of Dyslexia, 45, 3-27.

Muter, P., & Maurutto, P. (1991). Reading and skimming from computer screens and books. The paper office revisited? Behaviour and Information Technology, 10, 257-266.

Nuttall, C. (1996). Teaching reading skills in a foreign language. Hong Kong: Heinemann.

Perfetti, C. (1995). Cognitive research can inform reading education. Journal of Research in Reading, 18, 106-115.

Raynor, K., & Pollatsek, A. (1989). The psychology of reading. Hillsdale, NJ: Erlbaum.

Rayner, K., & Well, A. (1996). Effects of contextual constraint on eye movements in reading: A further examination. Psychonomic Bulletin & Review, 3(4), 504-509.

Segalowitz, N., Poulsen, C., & Komoda, M. (1991). Lower level components of reading skill in higher level bilinguals: Implications for reading instruction.  AILA Review, 8(1), 158-169.

Song, M. J. (1999). Reading strategies and second language reading ability: The magnitude of the relationship. English Teaching, 54(3), 73-95.

Sparks, R., Ganschow, L., Patton, J., Artzer, M., Siebenhar, D., & Plageman, M. (1997). Prediction of Foreign Language Proficiency. Journal of Educational Psychology, 89(3), 549-561.

Twining, J. (1991). Strategies for active learning. Boston, MS: Allyn and Bacon.

Weber, R. (1991). Linguistic diversity and reading in an American society. In R. Barr, M. Kamil, P. Mosenthal, & D. Pearson. (Eds.) Handbook of reading research, Vol. 2 (97-119). New York: Longman.

West, M. (1953). A general service list of English words. London: Longman.

Wepner, S., Feeley, J., & Wilde, S. (1989). Using computers in college reading courses. Journal of Developmental Education, 13(1), 6-8.

Xue, G., & Nation, I. (1984). A university word list. Language Learning and Communication, 3, 215-229.