Prospect 16,3

I S P NATION – Victoria University of Wellington, New ZealandJEAN PAUL DEWEERDT – Free University, Belgium ABSTRACT
This paper argues that simplified or graded readers are an essential part of alanguage learning program if learners of all proficiency levels are to have theopportunity to do incidental language learning through reading, and to developfluency in reading. Unsimplified texts do not allow for this kind of learning atbeginning and intermediate levels because they contain too great a density of unknown words and too many different unknown words. Evidence is providedto support this from a corpus study of versions of Dracula. Many criticisms of simplified texts apply only to poorly simplified texts and to the poor use ofsuch texts in curriculum planning.
Introduction
Graded readers are books specially written for learners of English using a con-
trolled vocabulary and grammar. A typical graded reader series consists of
books written at five or six vocabulary levels, beginning at around 300–400
words and increasing in stages to around 2500 words. The Oxford Bookworms
series, for example, has six levels, as shown in Table 1.
Table 1: The Oxford Bookworms series
New words
Cumulative
Range of length
Examples of titles
total words
in running words
White Death, Mutiny on the Bounty Dracula, Dead Man’s Island The Bronte Story, Love Story A graded reader may be simplified from a text for native speakers, such as JaneEyre, Dracula or Lord Jim, or may be a book written solely for second languagelearners, such as White Death, One Way Ticket or The Mystery of Allegra.
Some teachers and researchers feel that there is something wrong with texts simplified from literary works in particular, but also more generally with any Prospect Vol. 16, No. 3 December 2001 text that involves controlled and simplified language use. This uneasiness aboutgraded readers may be expressed either indirectly, through not using them inextensive reading programs as in the Elley and Mangubhai (1981) study, ordirectly through criticism of the books and their use. Yano, Long and Ross(1994), for example, criticise simplified texts for being unauthentic, inhibitingcomprehension and providing poor conditions for learning. The purpose ofthis article is to show that graded readers are an essential part of a second orforeign language learning program. This will be done firstly by examining thecriticisms of graded readers, and secondly by showing the value of graded readersusing a comparison of a simplified version and the original unsimplified versionof Dracula.
Criticisms of simplified texts
Graded readers that are simplified from literary texts written for native speakersare criticised for being unauthentic (Honeyfield 1977). A similar criticism isapplied to all texts that are written within a deliberately restricted vocabulary.
There is no doubt that reading a simplified version of Dracula is not the sameas reading the original, but no writer of a graded reader would suggest that itis. There is still the feeling, however, that a text which has been altered in thedrastic way that a simplified text has is no longer an authentic text. This criticismdepends on the meaning of ‘authentic’. The meaning of ‘authentic’ that isbeing used in the criticism is ‘written for native speakers’. In a very perceptivearticle, Widdowson (1976) pointed out that ‘authentic’ or ‘authenticity’ canbe viewed in a different way. Authenticity is not a characteristic of texts, butis the result of the interaction between a reader and a text. If a learner reads atext, and responds to it in a way that we might expect of someone who compre-hends the text, then reading the text is authentic for that learner. This responsemight involve understanding the text, enjoying its message, seeing the strengthsand weaknesses in its content and expression, or seeing its contribution to awider field. As we shall see later, a text that is authentic, in that it was producedfor native speakers, may be too difficult for learners to respond to in an authenticway. This view of authenticity is similar to the modern view of validity. A testis not valid in its own right, but is valid when it is used for the purpose for whichit was designed and examined.
Simplified texts are also criticised because the restriction on writers to use short, simple sentences can result in choppy and unnatural discourse (Honeyfield1977; Yano et al 1994), and may result in poor cohesive reference and anoverreliance on implicit, rather than explicit, conjunction relationships. Thesecan make the texts difficult to comprehend. These criticisms may be true ofpoorly written simplifications, but there are many excellent simplifications Prospect Vol. 16, No. 3 December 2001 that are a joy to read. Hill (in Day and Bamford 1998) provides a very usefuland substantial list of these high quality texts. It is unfair and misleading tocondemn simplifications as a whole because some are poorly done.
Graded readers are also criticised because they provide poor conditions for learning (Yano et al 1994). These criticisms include the following: 1 Reading graded readers that are pitched lower than the learners’ level can 2 Removing difficult vocabulary denies learners access to what they need 3 Reading texts with little unknown vocabulary discourages the development of generalisable coping skills, such as guessing from context and dictionaryuse.
These are all criticisms of the ways in which graded readers are fitted into a language course. In the remainder of this article, we will look at how the useof graded readers needs to be managed through the matching of the level ofgraded readers to learners’ proficiency levels, and to the learning goals of thevarious strands of a course. It should then be clear that the criticisms listedabove are really criticisms of poor syllabus design, rather than criticisms of gradedreaders themselves. Let us now look at the various parts of a language courseto see how graded readers can fit into these parts, where unsimplified textswould not be suitable.
The strands of a language course
One way of ensuring that there is a balance of appropriate learning activitiesin a course is to see a well-balanced course as containing a roughly equal balanceof the four strands of meaning-focused input, language-focused learning,meaning-focused output and fluency development.
The strand of meaning-focused input involves incidental learning through listening and speaking where around 98 per cent of the running words arealready familiar or pose no learning burden to the learners (Hu and Nation2000). In this strand, learners’ focus is on the message of the texts. Gradedreaders have a very important part to play in this strand of a course, both assources of listening input as well as reading input, particularly for learners atthe beginning and intermediate levels of proficiency. Much of the justificationfor the meaning-focused input strand of a course comes partly from the workof Krashen (1985) and advocates of the comprehension approach (Nord 1980),and from research on ‘book floods’ (Elley 1991). The book flood experimentsinvolve dramatically increasing the amount of reading that language learners Prospect Vol. 16, No. 3 December 2001 do, and observing their increase in proficiency over a range of language skillsand aspects of language knowledge. Elley (1989) has also shown that listeningto stories can be a source of vocabulary learning. The conditions for learningfrom input involve being able to comprehend the input, even though it containsa few features that are just beyond the learners’ knowledge. The input shouldbe interesting and should engage the learners in following the message it con-tains. Advocates of learning from input (Krashen 1985; Elley 1991) want to seeit occupy the vast majority of the learning time. For reasons explained below,it seems wiser to see it occupying about one-quarter of the learning time.
The strand of language-focused learning involves the deliberate study of language features. This strand includes vocabulary study, grammar activities,pronunciation practice and intensive reading where learners work with difficult,often unsimplified, texts. There is now considerable research (Ellis 1994; Spada1997; Long and Robinson 1998) which shows that deliberate intentional learn-ing has positive benefits in language learning and learning from input, partlythrough consciousness raising. Research in vocabulary learning (Nation 2001:263–316) has shown that intentional learning produces faster and better resultsthan incidental learning. It is important in course design, however, that meaning-focused learning and language-focused learning are not seen as competitors, but,rather, are seen as complementing each other. The language-focused learningstrand (Ellis 1994 calls it ‘formal instruction’ and ‘form-focused learning’) shouldnot occupy too much class time at the expense of more message-focused learning,and so it seems useful to limit it to around 25 per cent of the learning time.
The strand of meaning-focused output involves incidental learning through speaking and writing. Learners’ focus is on communicating messages. The justi-fication for a meaning-focused output strand has a theoretical basis in the workof Swain (1985) and experimental support from research such as that of Joe(1998) and Newton (1995), which show that learning can occur, and is en-riched, when learners produce language. Swain argues that having to producemessages makes the learners look at language in a new way. Language features,particularly grammatical elements, that were given little attention in listeningand reading, gain a greater importance in production when learners are pushedto make choices in how to say something.
The strand of fluency development does not involve new language items but focuses on helping learners become fluent with what they already know.
This fluency development is meaning-focused and needs to occur across thefour skills of listening, speaking, reading and writing. Simplified texts have animportant role to play in this strand because they are the best means of providingmaterial that contains no unknown items. There has been increasing interest influency development (Schmidt 1992; Skehan 1996), particularly with findings Prospect Vol. 16, No. 3 December 2001 that fluency increases are typically accompanied by improvements in accuracyand complexity of language use (Arevart and Nation 1991). As well as improvingaccess to language features, fluency practice – in the form of speed reading andrepeated reading, timed speaking and writing activities, and extensive listeningactivities – can enrich and consolidate knowledge of language features. Althoughfluency increases seem to transfer well between languages (West 1941; Cramer1975), there seems to be a need to provide fluency practice in each of the fourskills of listening, speaking, reading and writing. Fast speakers are not necessarilyfast readers. Fluency improvement best occurs when learners work with materialthat contains no unknown language features, where there is some pressure toperform at a higher than usual rate, where there is quantity of practice, andwhere the focus is on comprehending or producing messages. Fluency develop-ment tends to be neglected in most language courses, but learners need it inorder to make the best use of what they already know. It is thus advisable toexplicitly devote about 25 per cent of learning time to fluency development.
Giving roughly equal time to the four strands is an arbitrary decision aimed largely at counteracting spending too much time on direct language teachingand counteracting the neglect of fluency development and learning from trulycomprehensible input and output.
In the following comparison of a simplified and an unsimplified text, we will focus on extensive reading in the two strands of meaning-focused input andfluency development, because it is in these two areas particularly that simplifiedmaterial in the form of graded readers is essential.
Learning through meaning-focused input
Learning through meaning-focused input involves comprehension and con-ditions such as repetition that encourage learning. Table 2 lists some of theconditions affecting comprehension and learning. Three texts are analysed inthe table – the Oxford Bookworms version of Dracula (a level 2 simplified textof 7957 running words), the first 7965 running words of the unsimplifiedoriginal of Dracula, which is of roughly the same length as the simplified versionto control for the effects of text length, and the complete unsimplified originalof Dracula (161 952 running words).
Whenever the phrase ‘the first 2000 words’ or a similar phrase is used in this article, it refers to the 2000 word families of Michael West’s (1953) Ageneral service list of English words. Many of these 2000 words do not occur atall in the simplified and unsimplified versions of Dracula used in this study.
West’s list is now old and in some ways out of date. Unfortunately, in spite ofenormous advances in computing and the development of corpora, no com-parable carefully made list is available that is suitable for children or young Prospect Vol. 16, No. 3 December 2001 adults. Research on West’s list (Nation and Hwang 1995) indicates that the liststill works well and editing the list brings about only very small changes incoverage. In spite of this, a replacement is long overdue.
Table 2: Conditions for comprehension and learning in a simplified version
and the original version of Dracula
Conditions
Simplification
Short original
Complete original
Word families (Bauer and Nation 1993) have been used for most of thecounting in this study. A word family includes the stem form of a word, itsinflected forms and its regularly related, largely transparent, derived forms. Forexample, the word family mend includes the following members – mends,mended, mending, mendable, mender, unmendable. Once learners are familiarwith the inflectional system of English and the most frequent, regular and pro-ductive derivational affixes, the word family is the most suitable unit for countingthe receptive vocabulary load of texts. Otherwise, mend, mends, mending etcwould be counted as different words.
Research on vocabulary density and text comprehension (Hu and Nation 2000) indicates that for fiction texts, ideally, 98 per cent of the running wordsin the text should be familiar or should pose no comprehension burden for thereaders, if the text is to be read without consulting a dictionary. The followinglist of different coverage densities and average number of ten-word lines perunknown word shows the importance of coverage.
Table 2 shows that, for learners with a vocabulary of 2000 words or less, the simplified text (98.6 per cent coverage) would have 1 word outside thefirst 2000 words in every 7 lines. This provides a very supportive context foreach unknown word. The short unsimplified text (91.3 per cent coverage)would have 1 word outside the first 2000 in every 1.2 lines. This is a very Prospect Vol. 16, No. 3 December 2001 small context for the unknown words. Appendix 1 has an example of thesimplified and unsimplified texts with words outside the first 2000 markedin bold to show what this is like in practice.
The density of unknown words is one measure of difficulty. A related measure is the actual number of different unknown words outside the first 2000and that are not proper nouns. This affects difficulty because each new wordis another item to guess or look up, and if there are many of these, then com-prehension is difficult to achieve. Related to this are the repetitions of theseunknown words. If a reasonable proportion of the coverage of the running wordsoutside the first 2000 is made up of repeated words, then these have a very goodchance of being learned, and will quickly become known words that then addto the coverage of known words for the remainder of the text. The wordsvampire, wolves, howl and coffin are examples of repeated words outside the first2000 in Dracula.
Table 2 shows that there are only 30 word families outside the first 2000 and not proper nouns in the simplified text, and several of these words arerepeated many times. In the short unsimplified text, on the other hand, 530of the total 1416 word families (37.4 per cent) are not in the first 2000. In thecomplete unsimplified version, this rises to 67.8 per cent of the word families.
Moreover, in the short unsimplified text, 391 of the 530 word families occuronly once. This means that each one is potentially a comprehension and learn-ing problem. Thus, there are around 530 possible interruptions in the shortunsimplified text and relatively few in the simplified text. These ‘one-timers’make up a significant proportion of the total different words in the unsimplifiedtext, and this proportion increases with the length of the text. These one-timersrequire effort for comprehension, and this effort is not repaid by the oppor-tunity to meet them again when they would be less of a burden and wouldhave a chance of being learned.
Even in the simplified text there are many words, particularly within the first 2000 words, that occur only once in the text. However, because these wordsare from the most frequent 2000 words of English, they are worth learning.
It is a reasonable piece of advice to give to a learner of English that any wordin a graded reader is worth learning. This piece of advice is certainly not truefor an unsimplified text. Here are some of the uncommon words in the shortunsimplified text from Dracula – alacrity, aquiline, baying, crags, diligence(a type of stagecoach), engendered, goitre, hospadars, oleander, polyglot. Someof these words will eventually be useful additions to a broad and rich vocab-ulary, but there are thousands of more useful words to learn before these.
We have seen that unsimplified text can place a very heavy vocabulary burden on second language readers. Unsimplified text has too many of its runningwords outside most second language learners’ vocabulary knowledge, so that Prospect Vol. 16, No. 3 December 2001 readers with a limited vocabulary meet an unknown word in every few runningwords. These unknown words make up a very large group of words, most ofwhich occur only once in the text. A large number of texts would need to beread before many of them were met again. For learners of English with a vocab-ulary smaller than 2000 words, most unsimplified text is just too difficult anddoes not provide the conditions necessary for learning through meaning-focusedinput. Simplified texts, as in graded readers, can provide these conditions andare thus essential if there is to be a substantial meaning-focused input strandto a course. Struggling with difficult text can be a useful part of a course, butit is part of the language-focused learning strand.
Fluency development
Fluency development requires practice across the four skills of listening, speaking,reading and writing. The material used to develop fluency should contain nounknown vocabulary, so that learners can focus on making the best and mostfluent use of what they already know. For learners with a vocabulary of lessthan 2000 words, unsimplified text is completely unsuitable because it will notonly contain unknown words, but will contain large numbers of them. Todevelop fluency in reading, learners should be working with graded readersthat are well within their proficiency level. They need to read many of them,and they need to push themselves to read as fast as they can. Nation and Wang(1999) have shown that reading several graded readers provides excellent oppor-tunities to meet the same vocabulary many times. Reading unsimplified textdoes not provide suitable conditions for fluency development for either begin-ning or intermediate learners. Research on developing speed reading withlanguage learners (West 1941; Cramer 1975) has shown that, with the use ofsimplified material, substantial increases in speed are possible within reasonablyshort periods of time. As most second language learners read at speeds far belowa normal untrained reading rate of around 250 words per minute, developingfluency in reading should be a major goal of most English courses. It is a goalthat can be easily achieved with the use of simplified material.
Other novels
This article has compared the controlled text of a graded reader with uncon-trolled unsimplified text. It might be argued that a novel such as Dracula is anunusual kind of text in that it was written over 100 years ago and contains alot of foreign vocabulary because of its setting in a non-English-speaking country.
Table 3 presents the coverage figures, the number of word types not found inthe most frequent 2000 words of English, and the percentage of one-timersin a range of novels, to show that Dracula compares with other unsimplified Prospect Vol. 16, No. 3 December 2001 fiction texts and that the arguments presented here about the unsuitability ofunsimplified texts for learning from meaning-focused input and fluency devel-opment are generally true.
Table 3: Vocabulary statistics for a range of unsimplified fiction texts
Dracula
39 Steps
Washington
Something
Square
New
outside first 2000 occurringonly once, including propernouns Table 3 contains data largely from novels which usually provide the mostfavourable vocabulary load for learners, while newspapers and academictexts provide a much heavier vocabulary load.
Conclusion
Unsimplified texts do have important roles to play in language courses, and, formost learners, being able to comprehend them represents a major goal of alanguage course. But the end is not usefully the means. Good pedagogy involveshelping learners reach their goals through suitably staged steps. In the beginningand intermediate stages of language learning, controlled texts are among themost suitable means of bringing the important strands of learning from meaning-focused input and fluency development into play. When learners’ vocabularyis large enough, then unsimplified texts can fill these roles. At the beginningand intermediate stages, unsimplified texts may have a role to play in intensivereading and strategy development in the language-focused learning strandof the course.
Learners of English are extremely fortunate in that there is this tremendous resource of hundreds of graded readers at a variety of levels. This resourceshould be used as fully as possible and should not be rejected because the textsare not ‘authentic’ unsimplified texts. To reject this resource is to effectivelyeliminate many of the essential strands of meaning-focused input and fluencydevelopment from language courses.
REFERENCES
Arevart, S and I S P Nation 1991. ‘Fluency improvement in a second language’. RELC Prospect Vol. 16, No. 3 December 2001 Bauer, L and I S P Nation 1993. ‘Word families’. International Journal of Lexicography, Cramer, S 1975. ‘Increasing reading speed in English and in the national language’.
Day, R R and J Bamford 1998. Extensive reading in the second language classroom. Cambridge: Elley, W B 1989. ‘Vocabulary acquisition from listening to stories’. Reading Research Elley, W B 1991. ‘Acquiring literacy in a second language: The effect of book-based programs’. Language Learning, 41, 3: 375–411 Elley, W B and F Mangubhai 1981. The impact of a book flood in Fiji primary schools. Ellis, R 1994. The study of second language acquisition. Oxford: Oxford University PressJoe, A 1998. ‘What effects do text-based tasks promoting generation have on incidental vocabulary acquisition?’ Applied Linguistics, 19, 3: 357–77 Honeyfield, J 1977. ‘Simplification’. TESOL Quarterly, 11, 4: 431–40Hu, M and I S P Nation 2000. ‘Unknown vocabulary density and reading comprehension’.
Reading in a Foreign Language, 13, 1: 403–30 Krashen, S 1985. The input hypothesis: Issues and implications. London: LongmanLong, M and P Robinson 1998. Focus on form: Theory, research and practice. In C Doughty and J Williams (eds). Focus on form in classroom second language acquisition. Cambridge:Cambridge University Press Nation, I S P 2001. Learning vocabulary in another language. Cambridge: Cambridge Nation, I S P and K Hwang 1995. ‘Where would general service vocabulary stop and special purposes vocabulary begin?’ System, 23, 1: 35–41 Nation, P and K Wang 1999. ‘Graded readers and vocabulary’. Reading in a Foreign Newton, J 1995. ‘Task-based interaction and incidental vocabulary learning: A case study’.
Second Language Research, 11, 2: 159–77 Nord, J R 1980. ‘Developing listening fluency before speaking: An alternative paradigm’.
Schmidt, R W 1992. ‘Psychological mechanisms underlying second language fluency’.
Studies in Second Language Acquisition, 14: 357–85 Skehan, P 1996. ‘A framework for the implementation of task-based instruction’. Applied Spada, N 1997. ‘Form-focused instruction and second language acquisition: A review of classroom and laboratory research’. Language Teaching, 30, 2: 73–87 Swain, M 1985. Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S M Gass and C G Madden (eds). Inputin second language acquisition. Rowley, Mass: Newbury House Prospect Vol. 16, No. 3 December 2001 West, M 1941. Learning to read a foreign language. London: LongmanWest, M 1953. A general service list of English words. London: Longman, Green & CoWiddowson, H G 1976. On the authenticity of language data. In J Fanselow and R Crymes (eds). On TESOL ’76. Washington: TESOL Yano, Y, M H Long and S Ross 1994. ‘The effects of simplified and elaborated texts on foreign language comprehension’. Language Learning, 44, 2: 189–219 Prospect Vol. 16, No. 3 December 2001

Source: http://www.ameprc.mq.edu.au/docs/prospect_journal/volume_16_no_3/Prospect_16,3_Article_5.pdf