The Impact of Reading on Second Language Learning

Author(s): Warwick B. Elley and Francis Mangubhai

Source: Reading Research Quarterly , Autumn, 1983, Vol. 19, No. 1 (Autumn, 1983), pp.
53-67
Published by: International Literacy Association and Wiley

Stable URL: https://www.jstor.org/stable/747337

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/747337?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Wiley and International Literacy Association are collaborating with JSTOR to digitize, preserve
and extend access to Reading Research Quarterly

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
The impact of reading on second language learning
WARWICK B. ELLEY
University of Canterbury, New Zealand
FRANCIS MANGUBHAI
University of the South Pacific

FIVE CRITICAL differences between first and second language learning were ide
discussed. It was hypothesized that the effect of these differences in forma
could be virtually eliminated by means of a reading program based on th
abundance of high-interest illustrated story books. A sample of 380 Class 4
from eight rural Fijian schools with very few books was selected, and eac
provided with 250 high-interest story books in English. The 16 participating te
given directions in two different methods of encouraging the pupils to rea
Pre- and posttests were given to all pupils and to matched control groups of
who followed the normal structured English language program, which
emphasis on reading. Posttest results after eight months showed that pupils
many stories progressed in reading and listening corfiprehension at twice the n
and confirmed the hypothesis that high-interest story reading has an impor
play in second language learning. After 20 months, the gains had increased
spread to related language skills.

L'impact de la lecture sur I'dtude d'une deuxikme langue

ON A IDENTIFIEP et discut6 cinq diff6rences critiques entre l'6tude d'une
deuxieme langue. On a pos6 comme hypoth6se que l'effet de ces diff6rences e
formelle pourrait &tre pratiquement 61imin6 au moyen d'un programme de l
sur l'usage de nombreux livres d'histoires illustr6es a grand int6r&t. On a sel
6chantillon de 380 e61ves de classe de huitieme et septieme d partir de huit 6col
de zone rurale avec tres peu de livres, on a donc donn6 a chaque classe
d'histoires a grand int6rAt en anglais. On a fourni aux seize enseignants par
directions avec deux m6thodes diff6rentes pour encourager les l66ves a lire les l
donn6 des pre- et post-tests a tous les l66ves et a des groupes de contr6le de
conprennant 234 e61ves qui ont suivi le programme normal et structure de lan
lequel accentue peu la lecture. Les r6sultats des post-tests apr6s huit mois ont
les e61ves mis en contact avec de nombreuses histoires ont progress6 en l
compr6hension d'&coute a un taux double du taux normal, et ont confirm6 l
que la lecture d'histoires a grand int6rAt joue un r6le important dans l'6
deuxieme langue. Apr6s 20 mois, les b6n6fices avaient augment6 davantage e
et6ndus a des comp6tences ayant rapport a la langue.

El impacto de la lectura en el aprendizaje de un segundo lenguaje

SE IDENTIFICARON y se discuten 5 diferencias criticas entre el aprendizaje d
segundo lenguaje. Se plante6 la hip6tesis que el efecto de estas diferencias en e
instrucci6n formal podia virtualmente ser eliminado por medio de un pr
lectura, basado en el uso de una abundancia de libros de cuentos ilustrado
inter6s. Se seleccion6 una muestra de 380 alumnos de las cuartas y quintas
escuelas rurales de las Islas Fiyi que tenian muy poco libros, y se provey6 a ca
libros de cuentos de gran inter6s en ingl6s. Los 16 maestros participantes
instrucci6n en dos m6todos, para estimular a los alumnos a leer los libros. Se
de comprobaci6n experimental y final a todos los alumnos y a grupos a
control, compuestos por 234 alumnos que seguian el curso regular de idioma

53

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
 pone poco 6nfasis en la lectura. Los resultados de comprobaci6n final despu6s de 8 meses
mostraron que los alumnos que fueron expuestos a muchos cuentos progresaron a un
nivel doble del normal en lectura y comprensi6n auditiva, y confirm6 la hip6tesis de que
la lectura de cuentos de mucho interns juega un importante papel en el aprendizaje de un
segundo lenguaje. Despu6s de 2Q meses, el incremento en el aprendizaje fue todavia
mayor y se habia extendido a pttis destrezas relacionadas con el aprendizaje de un
lenguaje.

The main conditions under which children central hypothesis arising from this strategy is
develop competencies in first and second that repeated exposure to high-interest illus-
languages are the subject of much researchtrated story books in the target language will
and debate. How do children, given a mass of produce rapid L2 learning. To test this
haphazard linguistic input, master the basichypothesis, an empirical study was conducted
structures and vocabulary of their home in 12 primary schools in the Fiji Islands.
language before they go to school? Is The following factors appear critical in
exposure sufficient? How important is repeti-differentiating between LI and L2 learning in
tion? Must motivation be intrinsic? Why is the
contexts where the language of the school is
second language learned so laboriously and not the language of the home, as is the case in
all South Pacific countries.
ineffectually in the classroom, yet so readily
when the learner is part of a community toat Strength of motivation. As a growing
speaks the second language? organism, a young child has an urgent need to
Indeed, recent case studies of second communicate (Halliday, 1973) and to com-
language (L2) learners (e.g., Fillmore, 1976; prehend the communications of others. By
Milon, 1974; Ravem, 1974) suggest that L2 contrast, the L2 learner already has a
acquisition in naturalistic contexts has much perfectly good language to communicate his
or her needs; hence there is a greater
in common with first language (L 1) acquisition,
thus raising the possibility that carefully dependence on extrinsic (or instrumental)
structured L2 educational programs of the motivation to learn the language-to pass an
audio-lingual variety may be misconceived. examination, to please a teacher. Many
Dulay and Burt (1974) have supported this bilingual scholars (Dodson, 1978; Macnamara,
view with studies of error behavior in L2 1973) have deplored the consequences of this
learners, and found similar developmentaldifference and spelled out its undesirable
effects in L2 classroom practice and child
sequences in L I and L2 acquisition, regardless
behavior.
of age, learning method, and nature of the first
language. Emphasis on meaning vs. form. Young
Numerous empirical studies have demon-
L I learners make use of their language to
convey and receive new meanings. Typical
strated that L2 learning can be very effective
using strategies derived from our understand-
parents are more concerned with the meaning
than in
ing of LI learning, such as total immersion with the form of their children's
L2 from school entry (Bruck, Lambert, language
& (Brown & Hanlon, 1970). Widdow
Tucker, 1974), partial immersion (Barik son &(1978) makes a telling point when he
claims that "by focusing on usage, the
Swain, 1975), and prolonged passive exposure
before oral expression (Tucker & d'Anglejan,
language teacher directs the attention of the
1973). learner to those features of performance
With a view to furthering our under- which normal use of language requires him to
standing of how to bring L2 learning more ignore" (p. 17). However, in typical L2
into line with L1 acquisition, the present classrooms, the focus is clearly on form,
article enumerates the critical differences practice, and repetition of structure. Rarely is
between typical L1 and L2 learning and the purpose to convey new meaning.
proposes a practical instructional strategy for Amount of exposure to language. Ll
reducing the effect of these differences. The learners are continuously surrounded by their

54 READING RESEARCH QUARTERLY * Fall 1983 XIX/ 1

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
vernacular language. In the typical South they are engaging in an activity which reduces
Pacific family, children will be exposed to the effect of the five listed differences. It thus
over 40,000 hours of their home language by makes L2 acquisition considerably more like
the end of six years of schooling. By contrast, L1 acquisition, and consequently facilitates
at most, there will be only 3,000 hours of the acquisition process.
English instruction. Other things being equal, Such a proposition is alien to ponventional
any strategy which can reduce this gap would L2 teaching methodologies, in all coyntries of
also reduce the obvious disparities which exist the South Pacific region, and in many
in competence levels. education systems where audio-lingual ap-
Type of exposure to language. Exposure proaches are favored. In all the South Pacific
to the second language is normally planned, territories, English is taught in schools
restricted, gradual, and largely artificial, quite through an Oral English Syllabus (Tate,
unlike the typical interactions which young 1971), in which selected structures and words
children have with their mother tongue. are identified, taught orally in a particular
While it may be a common sense strategy to sequence, and practiced until mastered. Only
control the L2 learner's exposure to some then are the children exposed to the same
extent, it may easily become counterproduc- language items in the printed form. Reading
tive. Children learn much from the redundancy follows along behind the oral learning process,
in their language environment. Linguists have as a means of consolidating the pupils'
shown that gestures, facial expressions, into- learning, not of providing for the learning of
nations, and events of the moment correspond new forms. Thus, all reading material is
with and support the language children see carefully controlled, and exposure to English
and hear. As Donaldson (1978) points out, "it is very limited throughout the child's primary
is the child's ability to interpret situtations school years. As in most structural and audio-
which makes it possible for him...to arrive at lingual programs, errors are kept to an
a knowledge of language" (p. 38). Studies of absolute minimum. Children get little practice
vocabulary acquisition in L I learning also or encouragement for guessing the meanings
indicate that children learn many new words of unfamiliar words. In terms of the L1-L2
naturally from context, and less effectively differences listed earlier, it is perhaps not
from deliberate drilling. Thus, reducing the surprising that this program has not been
range of situations, and vocabulary loads, eminently successful (Elley & Mangubhai,
could well be ineffective. Furthermore, exper- 1979). The oral drills practiced by the pupils
iments aimed at increasing comprehension of have little justification as a form of genuine
prose by deliberately simplifying vocabulary communication; the repetitions are contrived
and structural complexity have often proved and often monotonous; the exposure is
disappointing (Peltz, 1973). minimal and carefully controlled; the teach-
The quality, of models. Practically all L2 ers, as models, are themselves teaching in
instruction in the South Pacific is undertaken their second language and are prone to error;
by non-native speakers of the language. This and the learners are frequently only half-
frequently means that pupils are exposed to hearted in their efforts to induce the rules of
faulty models of language. By contrast, L 1 the language for themselves.
teachers are usually more fluent, can clarify The Oral English Syllabus is not unique.
more readily, and can provide greater variety Rivers' (1968) description of typical audio-
of form and meaning. lingual programs in L2 makes it clear that
The task facing the L2 teacher is to "the student must never be allowed to read
reduce the differences listed above and to alone a script which he has not learned orally
increase the efficiency of L2 learning. This
first, or else heard a number of times" (p. 220).
task can be attempted profitably through the Similarly, Ching (1976) maintains that "...be-
fore bilingual children can learn to read
use of an abundance of high-interest, illustrated
story books, printed in the target language. English, they must be able to understand and
When children read high-interest story books,speak it effectively" (p. 4). The idea that

Impact of reading ELLEY & MANGUBHAI 55

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
children might actually learn new language affected by such community influences as
from contextual cues in print is given no peer groups, television, church, or even
credence in such programs. Nor is there any parental example. Thus, any differences in the
recognition of the widely demonstrated find- potential impact of language programs should
ing that a multi-sensory approach, in which reveal themselves in bold relief, unlike the
children see and hear new language items situation in most English-speaking countries.
simultaneously, is superior to the single- What evidence is available from the
sensory method followed in audio-lingual South Pacific on the effects of reading on
programs (see Dodson, 1976; Lado, 1977). acquisition of English as a second language?
Learning language through reading is actively Numerous visiting teachers have investigated
discouraged. the issue informally, with promising results.
By contrast, the reading of story books Thus, McKeating (Note 1), a former teacher
could play a dominant role in determining and English adviser with the Fiji Ministry of
what language children learn, when they learn Education, used book-based programs in
it, and most important, how they learn it. separate experiments in the Solomon Islands
High-interest story books provide a basis and Fiji secondary schools during the 1970's,
for language learning which goes a long way and reported dramatic improvements in
to bridge the gap between L1 and L2 learning pupils' external examination pass rates. At
contexts. Good story books provide strong primary school level, a more carefully con-
intrinsic motivation for children and an trolled evaluation of the effects of a book-
emphasis on meaning rather than form.based
Whenprogram on the island of Niue also
read often, these books increase exposure to marked improvements, particularly
produced
the target language. They become the in basis
reading comprehension, word recognition,
for discussion about the pictures andand oral sentence repetition. Class 3 and 4
story.
Through expressive activities, the stories children (8-10 year olds), whose teachers were
assist children to learn naturally, from trained to teach by the Shared Book Experi-
context, and provide excellent models of ence method, using a set of locally produced,
written English. high-interest readers, made rapid progress
There is, of course, empirical data from when compared with children who used the
Ll learning which supports the value of traditional Tate Syllabus (see De'Ath, 1980;
regular reading by children (e.g., Chomsky, Elley, 1980).
1979; Clark, 1976; Durkin, 1966; Milner, Further indirect support comes from a
1951; Ritchie, 1978). However, little empirical national survey of reading in Fiji at Class 6
work has been done within an L2 context to level (12-13 year olds). In this study, there was
explore the proposition that L2 learningclear evidence that access to books in school
was an important factor in distinguishing
would be more effective when based on story
reading and related activities. between good and poor readers. Even after
In most of the islands' education systems
home background factors were partialled out,
of the South Pacific, English is learned as schools
a with large libraries produced good
second language, usually after literacy has readers; schools without libraries did not
first been acquired in the vernacular. Thus, (Elley & Mangubhai, 1979). While such a
after Class 3 (8-year-olds), English is the relationship has been found in other develop-
ing countries, it is not clearly a causal link.
language of the school, although it is seldom
the language of the community. Under these The presence of a large library in a school may
circumstances, there is a golden opportunity merely indicate an enthusiastic staff, who
to investigate the proposition that increasedbelieve in the importance of books. However,
L2 exposure through extensive reading of it is an important relationship to explore, as a
story books provides a beneficial approach toplentiful supply of good books is a factor that
language learning. Practically all the child's
can be manipulated, far easier to produce and
control than general improvements in the
exposure to English takes place within the
school. Whatever English program the teacher
quality of instruction or teacher control over
follows is therefore critical. It is seldom English. Therefore, the present authors re-

56 READING RESEARCHI QUARTERLY * Fall 1983 XIX/ 1I

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
solved to conduct a "Book Flood" in a set of receptive language use (reading and listening
rural primary schools, where English language
comprehension) than on measures of produc-
standards were low and resources meagre.tive language use, since both methods empha-
size the former. However, some transfer was
expected to these latter measures.
Method The Shared Book Experience group
required some teacher training, as it was a
new approach for the eight teachers concerned.
Design A specialist adviser in this method was
brought from Auckland, New Zealand, to
The design involved the random assign-
ment of pupils in Classes 4 and 5 (9-11 year
conduct a three-day workshop early in March
olds) to one of three treatments-the Shared
1980, before the first books were placed in
Book Experience, Sustained Silent Readingschools. He also visited the schools to give a
brief demonstration lesson in each one.
of books, or the control group, employing the
traditional Tate Oral English Syllabus. Using
Teachers in the Silent Reading Group were
analysis of variance on residual gain scores,
not given a special course, but a set of notes
pretest reading comprehension scores were outlining the method was prepared by the
used as a covariate, and two a priori contrasts
experimenters and discussed with the teachers
were employed to evaluate treatment effects. during an early visit to the schools. The
Contrast 1 compared the two groups (Shared Control Group teachers were given a one-day
Book and Sustained Silent Reading) that refresher course by Ministry of Education
employed large numbers of books with the staff in the principles of the Tate Method, in
control groups. The second contrast compared order to reduce any effect which might be
the Shared Book method with the Sustained caused by the additional stimulus of the
Silent Reading method. The hypothesis special course provided for the Shared Book
behind contrast I can be called a "Book Flood teachers.
hypothesis": exposure to large numbers of All teachers were aware of the fact that
story books will have an effect on general they were participating in an experiment, but
language competence. The hypothesis behindnone knew the nature of the tests that their
the second contrast was that the exposure pupils
to would be exposed to in November. All
new language in the Shared Book method is
teachers adhered to the normal timetable, but
more persistent and concentrated, and that Book Flood schools replaced 20-30 minutes
pupils become more actively involved in of theTate Reading activities each day with
learning experience than in the Silent Reading
Shared Book or Silent Reading activities. No
Method. extra time was devoted to English in any
Finally, these contrasts were expectedgroup.
to Figure I shows the project design for
show up more dramatically on measurestheof first year.

Figure 1
Design of Book Flood Project - 1980
February March April-October November

Shared Book Pretests 3 day 250 books Posttests

Experience Workshop supplied to
Group Classes 4 & 5

Silent Pretests No 250 books Posttests

Reading Workshop supplied to
Group Classes 4 & 5

Control Pretests I day Usual program Posttests

Group Workshop No extra books

Impact of reading ELLEY & MANGUBHAI 57

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
Sample but required later adjustment for Class 4 (see
Classes 4 and 5 were chosen initially Table 1).
because Class 6 was an examination year for
Fijian pupils, and it was felt that school Teaching Procedures
principals would be reluctant to reduce the The eight "Book Flood" schools received
time of formal English lessons for such an their books in five lots of approximately 50
unusual approach to language learning. Class for each class, at intervals of 4 to 5 weeks. The
3 children were considered less suitable experimenters visited the schools at that time
because their English was so limited that it on the procedures being used, to
to check
observe lessons, and to inspect materials
would be difficult to identify suitable reading
produced.
materials. It is again stressed that all pupils in The main features of the three
groups are outlined below.
Fiji spend the first three years of schooling
Shared Book Experience Method. Hold-
acquiring literacy skills in their vernacular
language, and switch to English as the away (1979) developed this method in New
medium of instruction during Class 4. Zealand primary schools where it is now
Rural schools were used in this investiga-extensively used by European and Polynesian
tion because their pupils are less exposed tochildren. The teacher chooses a high-interest
the neutralizing effects of English in the story with appropriate language and illustra-
community than are city children. The tions and introduces it to the pupils in a
particular schools chosen were typical of "sharing experience," similar to that of a bed-
many in Fiji, in that they were controlled by a time story. Discussion is encouraged about
local committee, had between 15 and 40 the pictures, the likely contents, and a few new
pupils at each class level, had pupils and words. The teacher then reads all, or some of
teachers of predominantly one ethnic group, it, to the class. To ensure that all of the group
and had very few resources--books, audio- can see the text and illustrations, the book is
visual aids. All schools followed the same frequently "blown-up" or rewritten in the
form of a giant book, with suitably-sized
timetables, which are standardized throughout
Fiji. All teachers had been locally trained illustrations.
in
the traditional two-year training college During the second or third readings, on
course, and all followed the Tate Oral subsequent days, children are encouraged to
English Syllabus which is universal in Fiji.join in and read easier sections with the
Six Fijian and six Indian schools wereteacher, who continues to encourage discussion
about the contents of the book. Emphasis is
chosen at random for possible inclusion in the
project. All headmasters who were approachedplaced on prediction and confirmation of
agreed to participate, and pretesting took events in the story, so that children are
place in late February 1980. After scoring andconstantly striving for meaning. If children
analysis, schools within each ethnic group enjoy the experience, they will want to read it
were placed in two achievement levels, asoften, in the class group, in small groups, in
indicated by mean score. Within each level, pairs, or as individuals. The intention is that
the schools were assigned randomly to one of they master the language of the book, with a
three groups: Shared Book, Silent Reading, minimum of pressure and strain.
or Control. One Fijian school had to be Follow-up activities include role playing,
transferred from Shared Book to Control word study, art work, writing activities. The
Group because the Class 4 teacher was origin on of these activities is always determined
leave and unable to attend the training session
by the story, not by any pre-ordained system
when required. Each of the three groups had of the proper sequence of structures and
two Fijian and two Indian schools, and vocabulary
each to follow. Theoretically, the essence
had a cross-section of teachers of both sexes of the method is that new learning takes place
with various levels of experience. Matching at the point of interest, rather than in accord-
on pretest scores was very close for Class ance
5, with a carefully graded linguistic pattern.

58 READING RESEARCH QUARTERLY o Fall 1983 XIX/I

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
 Silent Reading Method. The eight teach- Assessment Procedures
ers who used the Silent Reading Method were In order to assess the impact of the books
given no special workshop. They were advised and to compare the two approaches used by
by the experiments to display the books the teachers to introduce pupils to the books,
attractively, to read them aloud regularly, and all pupils in Classes 4 and 5 in the twelve
to spend 20-30 minutes each day in sustained schools were tested in reading comprehension
silent reading, with books of the children's under standardized conditions in February.
own choice. The principles of this method This test served as a matching variable and as
were developed by McCracken (1971). Briefly, a baseline for estimating the growth made by
the rationale is that children best learn to read
the various groups. Early in November,
by reading, as often as possible. A definitealmost eight and a half months later, the same
period is set aside every day for reading pupils were given an extensive range of tests
during which time the teachers must set aof reading and other English language
good example by reading also. No book abilities. All tests used in the first year are
reports are required, and no written exercises outlined below. (A twelve month follow-up
performed. The children are reading for enjoy-assessment, using other tests, is described
ment and practice. later.)
Control Group. The eight teachers in the For Class 5, the pretest for reading
Control Group were advised to follow theircomprehension was a sentence-completion
normal curriculum in English, which was the test of 35 multiple-choice items, developed by
SPC/Tate audio-lingual program referred tothe senior author and used for research
above. Children have two 15-minute oral purposes in several Pacific Islands. The
English lessons each day in which new reliability
struc- of the test is high (Split-half r =
tures are systematically introduced, in 0.90),
appro- and correlations with other reading
priate classroom situations, with repeatedtests on single grade groups range from 0.78
drills, variations, substitutions tables, etc.
to 0.83.
Reading is taught through carefully graded The posttests used were as follows:
STA F Reading Comprehension (Form Y)
SPC readers (in Class 4) or the Fiji Ministry's
text "Stories for Us" (in Class 5). Both series six passages and 32 multiple-choice
contained
items developed as a standardized test of
of books are graded and activities are intended
primarily to provide practice in order to
achievement for Fiji (STAF) but not available
consolidate the structures and vocabulary in schools in November 1980. Split-half
taught in the oral lessons. reliability is reported as 0.92, and correlations
In general, most teachers followed with
direc-
other reading tests and relevant external
tions. However, one Shared Book teacher examinations are consistently high (See Elley
used the method rarely as he felt it was& Achal, 1980). The STAF Listening Com-
unsuitable for him and his pupils; one Control prehension Test (Form Y) consisted of 36
Group teacher was found to read regularly to multiple-choice questions based on seven
her class, a practice very rare in Fiji schools.short passages read aloud by the administra-
Several teachers were replaced while on leavetor. Split-half reliability was estimated as
for maternity or sick leave, for periods up to 0.88, and correlations with other English tests
10 weeks, but these disturbances, which arein the series range from 0.59 to 0.72 (Elley &
common in Fiji primary schools, appeared Achal, to 1980). English Structures Test had 20
be equally distributed across all three groups.open-ended items. It was specifically designed
Their net effect was probably to make it more for the project and trial-tested on comparable
difficult to demonstrate real differences be-pupils. Pupils completed, in writing, short
tween the various approaches. A school-by- sentences in their own words using structures
taught in Class 5 in the SPC/Tate Oral
school analysis suggests that such disturbances
produced measureable effects in their chil- Syllabus. Assessment focused on the appro-
dren's progress (see Table 5, p. 64). priateness of the structure used. The papers

Impact of reading ELLEY & MANGUBHAI 59

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
were marked by the authors "blindly," Results
without knowledge of which group the pupils
belonged to, and checked-marked by an
Pretest Scores
Adviser in English. Correlations between
scores of the three markers across three The pretest reading comprehension results
for each group, along with the percentage of
schools were all 0.98 or above. The English
Composition Test required students Fijian
to com-(as contrasted with Indian) pupils are
given
plete a short story about an old lady who in Table 1.
lived
The groups were closely matched except
alone with her cat and her hen. Equal weight
for the Class 4 Control Group. However, it
in marking was given for content, sentence
was impracticable to rearrange them without
sense, and mechanics (2 marks each). Marking
changing
was done blind and check-marked by an the Class 5 pattern, so it was
English Adviser. Correlations between resolved mark- to make adjustments for all pretest
ers' assessments were all above 0.90. differences in the posttest analyses.
The pretest used for Class 4 was the same As the same pretest was given at both
as for Class 5. The posttests for Class 4 wereclass
as levels, it was possible to estimate the
follows: Reading Comprehension Test wasnormal amount of growth in 12 months by
the same as for the pretest. The English
comparing the mean score for the total group
Structures Test consisted of 35 multiple-
at each level. The difference between the
choice items assessing pupils' ability to choose
means was 3.97 points or 11.35%. This figure
the correct structures in a series of sentences.
was used later as a guide to assess the amount
All structures were taken from the Class 4 of progress made by pupils in the project.
Tate Syllabus. The test was trial-tested and
revised before use. Split-half reliability was
Posttest Results
0.83 in three schools, and the correlation with
Although an attempt was made to match
the Reading Comprehension test was 0.73.
the three groups on pretest scores at both class
For the Word Recognition Test, each Class 4
levels, it was found that some discrepancies
pupil was interviewed individually, and every
still existed at Class 4 level. To match
second pupil was given a Word Recognition
precisely, it would have been necessary
Test during this interview. Pupils had to
drop nearly 40 cases from the Class 4 cont
recognize and correctly pronounce a series of
50 graded words. They were stopped after
missing four consecutive words, the total
Table I Pretest reading comprehension r
score being the number of words correctly scores
pronounced up to that point. The test was Class 4
identical to that used in a previous study
(Elley, 1980) in which it proved to be reliable
Group N M SD % Fijian
and sensitive to changes in reading levels. The
Oral Sentence Repetition Test was given to
Shared Book 81 15.07 5.72 51%
each child who did not take the Word Silent Reading 98 14.67 4.78 49%
Control 121 12.14 4.12 40%
Recognition Test. It is an adaptation of a test
Totals
used by Clay, Gill, Glynn, McNaughton, and 300 13.76 5.21 46%
Solomon (1976). Pupils repeat orally, after
the tester, a series of 28 English sentences Class 5
graded for complexity of structure. Children
who are unfamiliar with a given structure are
Group N M SD % Fijian
presumed to have more difficulty in repeating
Shared Group 105 17.99 5.90 47%
it correctly. There is evidence that the test is
Silent Reading 96 17.70 5.59 52%
more a matter of language mastery Control
than of 113 17.50 5.36 43%
memory (see Hanayan, Markman, Pelletier,
Totals 314 17.73 5.88 48%
& Tucker, 1978).

60 READING RESEARCH QUARTERLY * Fall 1983 XIX/I

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
Table 2 Residual gain scores for Class 4, 1980

Shared Book Silent Reading Control

N M SD N M SD N M SD

Reading Compre- 75 0.59 4.06 84 1.21 4.78 106 -1.40 4.17

hension
English Structures 71 0.99 4.68 84 0.63 4.71 106 -0.95 3.48

Word Recognition 37 2.08 9.83 43 -0.35 12.73 54 -1.33 11.63

Oral Sentences 34 1.56 5.74 38 -0.68 5.16 49 -0.71 4.36

which is equivalent were

group. In case this procedure to approximately 6.5
to distort
the findings in any months
way, in grade
the norm units. The pupils
results reported
here are based on regression
exposed to a rich supplyestimates,
of books were in
which all pupils were retained
improving incomprehen-
their general reading all groups
Using the pretest as sion
a skill at over twice the normal
predictor, eachrate. pupil's
score was estimated, and subtracted
A similar pattern was found in from
the test of the
actual score, and analyses
English Structures. of variance
An analysis of variance tower
performed on these residual
test the gain contrasts
hypothesized treatment (or loss)
scores.' showed that the Book Flood groups performed
significantly better than the Controls, F
Class 4 results. Table 2 shows the residual
gain score means for Class 4. It should be (1,257) = 10.90, p <.001, but did not differ
noted that the correlations between the predic-between themselves, F (1,257) = 0.28.
tor test and each of the posttests was positive On the Word Recognition test, the Book
and high, the lowest being 0.48 for Oral Flood groups again showed higher means,
Sentences, and 0.69 for Word Recognition. but the difference was not significant, F
For Reading Comprehension, an analysis (2,131) = 1.06, due mainly to the larger
of variance showed that the differences were standard deviations (ranging from 9.8 to
significant, F (2, 258) = 9.30, p <.001. A 12.7), as well as the smaller N's, and the drop
partitioning of the variance for the treatment in precision caused by a lower correlation
effect was undertaken in order to test the two between predictor and posttest. Although the
planned a priori contrasts, that the Book Shared Book group appeared to have a
Flood groups surpassed the Controls, and superior result, the difference between the
that the Shared Book group surpassed the means of the two Book Flood groups
Silent Reading group. On the first contrast, represented only 0.2 of the pooled standard
the Book Flood groups were clearly superior, deviation.
F (1,258) = 17.70, p <.001. The second On the oral sentences, the differences
contrast showed no between groups again favored the Book
significant differences
between the Shared Flood groups, but the F test failed to reach
Book and Silent Reading
Groups, F (1,258) = 0.81. significance for similar reasons to those given
above for the Word Recognition Test, F
An inspection of the raw scores revealed
(2,118) = 2.60.
that, between late February and early Novem-
ber, the two book groups gained by a mean of In fact, a post hoc simple effects test
4.9 points, compared with an expected 3.97 showed that the Shared Book group produced
points over 12 months (estimated from the a significantly higher mean than the Control
pretests given to both Classes 4 and 5 in group, t = 2.59,p <.05, but the Silent Reading
February). In other words, the book groups Group was not different from either.
produced 15 months reading growth (on Class 5 results. Table 3 shows that at the
grade norm scores) in 8 months. By contrast, Class 5 level, all means again favored the two
the Control Group showed 2.1 points growth, Book Flood groups. In Reading Comprehen-

Impact of reading ELLEY & MANGUBHAI 61

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
sion, the hypothesized contrasts showed the the Class 4 test of structures did show a
two Book Flood groups to be significantly significant difference may have been due to
better than the Controls, F(l, 267)= 21.07, p the use of a longer and more reliable test at
<.001, and the Shared Book group signifi- that level. The trend on the written composition
cantly ahead of the Silent Reading group, measure favored the two Book Flood groups,
F(l, 267) = 9.87, p <.001. An examination of but the differences were slight and not
the raw score means was made to estimate the significant.
approximate size of the progress made. The
pretest was given a second time to one-thirdSummary
of of Findings: Classes 4 and 5
the pupils in each group, and, as the The general hypothesis investigated was
correlation between this test and that used as
that L2 pupils exposed to a rich variety of
the posttest was high (0.83), it was possible high-interest
to illustrated story books, would
assess the approximate amount of growth
show greater than normal gains in English
made by the total group by extrapolation,
language. Incidental hypotheses were that the
using regression estimates. Shared Book group would perform better
The Shared Book group was found to than the Silent Reading group, and that
differences would be greater in receptive than
have progressed by 4.86 points, or close to 15
months. The Silent Reading pupils showed expressive skills.
lesser gains (9 months), and the Control At Class 4 level, the Book Flood groups
were significantly ahead on the group tests,
group only 2.5 months growth in the same
period. The effect of the books was and
to marginally so on the two individual tests.
improve the children's reading growth by It should be noted that if the control group
nearly twice the normal amount in 8 months,teacher who regularly read stories to her
in the case of the Shared Book pupils. pupils had been dropped, these differences
A comparison of the Class 5 posttestwould all have been greater.
results in Listening Comprehension showed a At Class 5 level, the Book Flood groups
were again significantly ahead on two tests,
very similar pattern, favoring the Book Flood
groups again. The orthogonal contrasts both of which assessed comprehension. On
showed the experimental groups well ahead the other two, the differences were slight,
of the Controls, F(1,266) = 35.74, p <.001,although in the predicted direction. The
and the Shared Book group surpassed the comparisons between Shared Book and
Silent Reading group, F (1,266) = 5.19,Silent
p Reading methods showed significantly
<.01. greater benefits for the former in Class 5, but
not in
The differences on the open-ended test ofClass 4, where the measured impact of
the twoF approaches was very similar. As
English Structures were not significant,
predicted, the observed differences were
(2,267) = 0.83, although both experimental
groups showed higher means. The factgreater
thatin the case of receptive than of

Table 3 Residual gain scores for Class 5, 1980

Shared Book Silent Reading Control

N M SD N M SD N M SD

STAF Reading 91 2.08 5.23 88 -0.14 4.05 91 -1.82 5.11

Comprehension
STAF Listening 91 2.18 4.31 87 0.63 5.30 91 -2.13 4.25
Comprehension
English Structures 91 0.12 3.14 87 0.24 3.11 91 -0.33 3.17

Composition 91 0.05 1.51 87 0.16 1.40 91 -0.07 1.28

62 READING RESEARCH QUARTERLY * Fall 1983 XIX/I

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
expressive skills, although it is possible that vocabulary, an open-ended test of English
the greater reliability of measurement in the structures, and a written composition based
comprehension tests provided a greater op- on a sequence of four pictures depicting a
portunity for these differences to be detected. story. Procedures for piloting, administering,
and marking were similar to those of 1980.
One-Year Follow-Up Study, 1981 In addition to these tests, the authors
In order to examine the permanence of gained access to the results of the Fijian pupils
the effects noted above, a 12-month follow-up (i.e., those in 6 of the 12 schools) who took the
study was conducted in all 12 schools. The Fiji Intermediate Examination for Class 6
Book Flood groups were given an extra 100 pupils in July 1981. This national examination
books per classroom, as most of the 1980 provided results in English, Mathematics,
supply had been read by many of the pupils. General Studies, and Fijian language, each
New teachers were briefed on the procedures based on two hours examining time.
to be adopted, and all teachers were informed, The residual gain scores on all language
as before, that their pupils would be assessed tests given in November 1981 are presented in
with a variety of English tests. Table 5 for Classes 5 and 6 respectively. The
The tests used were similar, in style and pretest used for estimating 1981 scores was
skills measured, to those of the 1980 survey, the same pretest as was used for the 1980
although the individual tests were omitted. survey.
The younger pupils (Class 5 level), took group As in 1980, the two Book Flood groups
tests of cloze reading comprehension, listening demonstrated much greater progress in their
comprehension, and open-ended English struc- English language growth, this time on all
tures. The older pupils (Class 6) had multiple- tests. The hypothesized contrasts between
choice tests of reading comprehension and of Book Flood groups and Controls were highly

Table 4 Follow-up residual gain scores for 1981

Class 5

Shared Book Silent Reading Control

N M SD N M SD N M SD

Reading Compre- 66 2.13 5.83 70 2.67 6.62 91 -3.60 5.09

hension
Listening Compre- 66 1.10 3.36 70 0.81 3.72 91 -1.45 2.90
hension
English Structures 66 0.81 3.72 70 1.28 3.87 91 -1.55 3.41
Total 66 4.02 11.28 70 4.78 12.41 91 -6.59 9.33

Class 6

Shared Book Silent Reading Control

N M SD N M SD N M SD

Reading Compre- 81 1.27 5.87 64 1.40 5.20 87 -2.22 4.69

hension
Word Knowledge 81 0.92 4.54 64 1.62 4.42 87 -2.02 4.28

English Structures 81 1.65 6.53 64 1.22 5.84 87 -2.46 4.78

Written Composition 81 0.52 2.74 64 0.66 2.33 87 -0.99 1.89

Total 81 4.35 17.17 64 4.89 14.05 87 -7.64 12.28

Impact of reading ELLEY & MANGUBHAI 63

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
significant in all cases, F (1,225)= 58.14, analysis and comment (see Discussion). Similar
28.73, 27.49, and 55.21, p<.001, for Reading, variations were found in the November 1980
Listening, English Structures, and Total analysis.
Scores, respectively. In no case, however, did The last set of findings reported here is
the Shared Book group significantly surpass based on the Fijian Intermediate Examination
the Silent Reading group. in July 1981. Those Class 6 pupils in the six
In Table 4, the findings for Class 6 once Fijian schools, for whom the Examination is
again clearly favor the two book-based designed, were included in the results reported
programs. The planned contrasts showed in Table 6. The samples are smaller than in the
highly significant (p <.001) Fratios (1,230) ofearlier analyses, as the Examination is not
24.66, 30.17, 24.73, 25.00, and 37.96 for taken by pupils in the Fiji-Indian schools.
Reading, Word Knowledge, English Struc- Four more pupils were dropped from the
tures, Written Composition, and Total Scores
Book Flood groups in order to equate pretest
respectively, but no differences between scores before conducting an analysis of
Shared Book and Silent Reading groups.variance on the Examination scores.
A further analysis was undertaken to In all subjects, the Book Flood groups
estimate progress made, school by school, performed well above the typical performance
during 1981 based on 1980 posttest results. of rural schools (45-50%). The advantages
Total scores on all 1980 posttests were com-shown by the Book Flood pupils was greatest
puted, and predictions made for all pupil's in the case of English, F (1,100) = 14.82, p
total scores for the 1981 survey. Table<.001,5 and General Studies (Science and
reveals that, while all eight Control classes
Social Studies), F (1,100) = 17.62, p <.001,
were consistent in showing below average but there was also an unexpected spread-of-
residuals, there was considerable variation effect
in to Mathematics, F (1,100) = 13.01, p
the progress of the experimental groups. <.001, and a similar but less marked tendency
Eleven of the 16 Book Flood classes showed in the Fijian Language Examination, F
positive gains and 5 were negative. The (1,100)
net = 2.83, p <.10. The other contrasts,
between Shared Book and Silent Reading
effect was clearly greater gains for the Book
Flood treatment in 1981, as in 1980, but the
groups, proved significant only in the case of
English, F (1,100) = 6.51, p <.05.
class-by-class variation is worthy of further

Table 5 Mean residuals for total scores showing progress by schools in 1981
Class 5

Shared Book Silent Reading Control

School N M School N M School N M

V 15 -2.33 S 27 5.93 T 10 -1.70

N 15 7.93 C 14 -0.79 S 25 -5.72
SD 25 2.72 R 14 6.36 S 26 -3.46
K 15 0.47 B 20 -0.55 K 36 -3.11

Totals 70 2.23 75 3.03 85 -3.73

Class 6

Shared Book Silent Reading Control

School N M School N M School N AM

V 20 7.25 S 22 -1.95 T 16 -1.19
N 22 6.73 C 12 11.08 N 25 -1.00

64 READING RESEARCH QUARTERLY * Fall 1983 XIX/ 1

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
 Table 6 Mean percentage marks for Fiji intermediate examinations, Class 6, 1981

Shared Book Silent Reading Control

Subject N M N M N M

English 37 63.99 37 53.74 29 45.55

Maths 37 55.30 37 54.30 29 41.45
General Studies 37 57.51 37 57.97 29 46.03
Fijian 37 58.11 37 58.51 29 53.69

Discussion
The question arises as to what really
caused the differences between groups. Cynics
The general hypothesis investigatedmight
was point to a Hawthorne Effect, brought
that L2 pupils exposed to a rich variety ofnovelty of the supply of attractive
on by the
high-interest illustrated story books will show
books. In the authors' experience, the effect of
greater gains in English language than is program wears off in a matter of
any novel
normal for such children. The findings days
clearly
or weeks, and there was no exception in
support this hypothesis for the receptive
thisskills
case. If there was any greater motivation
during the first year of the project andtofor
read all
English, it was produced by the appeal
language areas sampled during the secondof the books themselves. Such an effect can be
year. The subsidiary hypothesis, that the by any teacher who sets out to
produced
receptive skills would benefit most,furnish
was pupils with a large supply of interesting
supported initially, but there was a transfer
reading materials. Nor can the stimulus of the
effect apparent in the second year initial
which workshop be credited with the cause of
encourages the view that marked improve-the improvement, as the Silent Reading group
ment in one aspect of L2 development hassuch advantage in either year of the
had no
substantial effects on related skills. In fact, the
project.
general improvement of the experimental Clearly the cause of the differences lies in
groups in the Mathematics, General Studies,the different classroom activities undertaken
and even Fijian Language Intermediate Ex-
by the pupils over the two-year period. While
aminations seems to imply an incidental
it wasand
impracticable in this project to
positive change in attitude to school.undertake
While detailed ethnographic studies of all
the improved General Studies results24might
classes with more than 35 teachers, some
be explained in terms of better communication
helpful inferences can be drawn about the
in English between teacher and child, the change agents. All experimental
critical
results for Fijian are harder to explain
groupsthis
were exposed to more than 300 high-
way, as English is not used as a mediuminterest
in this illustrated story books, and the
subject. control groups were not. Most experimental
The other a priori hypothesis, that the groups made substantial improvements;those
Shared Book method would show greater which did not had teachers who restricted the
gains than the Silent Reading method, use of books, and did not follow the
received support only in the first year analysis guidelines provided. Only one control group
at Class 5 level, on the two receptive skill tests, made better than average progress (in the first
and hardly at all in the second year tests. year) where the teacher had exposed her
While the Listening Comprehension Test did pupils to daily story book reading, from her
show greater benefits for the Shared Book private collection of books. Thus, repeated
group, as one might expect, the difference was exposure to print, pictures, and story lines
only significant in one case. was a clear and consistent differentiating

Impact of reading ELLEY & MANGUBHAI 65

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
factor between the treatment and the control teachers V (in Class 5) and SD (in Class 6) had
groups. been dropped, the Shared Book approach
In 1981, the majority of the teachers in would have been given greater support. Thus,
the Shared Book groups read aloud regularlythe Shared Book activities cannot be ade-
to their pupils. Inquiries and observation quately assessed without tighter controls over
revealed that four out of eight teachers in theteacher behavior.
Silent Reading group also did so, but no On the whole, however, it is difficult to
control group teacher did. These observedescape the conclusion that the critical factors
differences were followed up with the treat- which brought about the substantial improve-
ment groups with a posteriori tests, and veryments were related to greater and repeated
impressive differences emerged. Thus, in exposure to print in high-interest contexts, in
Class 5 (1981), Schools N, SD, S, and R (see conditions where pupils were striving for
Table 8) with regular reading aloud showedmeaning, and receiving sufficient support to
mean residuals of 15.38, compared with 9.22achieve it regularly. These features were
in the remainder of the Book Flood groups.common in the experimental groups and
The corresponding figures for Class 6 (withabsent in the controls. The more specific
Schools V, N, K, C, and R against the aspects of methodology in achieving these
remaining three schools) were 25.68 and 7.18.ends seemed of lesser importance as the
differences between the two rather different
Both differences are highly significant, z (143)
= 3.40 and z (146) = 7.68, favoring the regular
approaches designed to achieve them were
reading aloud treatment. This particularminimal.
strategy has confirmation in the literature The theoretical orientation of this article,
from Cohen (1968) with Ll children and, regarding the critical differences between L1
more recently, with L2 pupils in a small and L2 learning in schools, received consider-
follow-up study in Fiji primary schools able support from this study. There are also
(Ricketts, 1982). practical implications for teachers and parents
A few teachers in the Shared Book group of L2 learners. Time spent on reading in
prepared a number of "giant books" (with the school has been regarded largely as entertain-
help of their pupils and other teachers), and ing, as relaxing. However, it must also be seen
left them on display for repeated use. The now as educationally beneficial. Furthermore,
pupils in these classes made sufficiently large strict controls over the vocabulary and
gains to support the value of this activity. structures of L2 pupils' reading materials
However, differences noted could also reflect seem to be unnecessary and counterproductive.
a greater enthusiasm on the part of the teacher Children can learn new structures from
for the method as a whole. relatively uncontrolled materials, provided
Teachers in the Silent Reading method there is the support of cues from pictures,
provided no follow-up activities to support absorbing context, and teacher guidance.
the reading done, while those in the SharedWhether 250 books per classroom are
Book groups did. The lack of difference necessary to achieve double the normal rate of
between the two methods indicates that these reading growth is debatable. Many of the
activities are not as important as their books selected were used rarely, and there is
advocates claim, or the teachers who used much to be said for repeating certain stories if
them did so ineffectively. It should be language acquisition is seen as the primary
remembered that most primary school teach- aim. Where resources are scarce, a smaller
ers in Fiji left high school before their number, perhaps 150 well-chosen short story
university entrance year, have only limited books, capable of being read in one sitting,
teacher training, and are teaching in their has much to recommend it. Moreover, the
second language. Furthermore, it was obvious popular stories of the western tradition-The
from classroom visits, and confirmed by the Three Pigs, Cinderella, Red Riding Hood,
results of Table 5, that a few teachers did little and others of their kind, as presented in the
to follow the Shared Book methodology. If inexpensive "Ladybird" series-were consist-

66 READING RESEARCH QUARTERLY * Fall 1983 XIX/I

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms
ent favorites. While some research in the ELLEY, W.B., & MANGUBHAI, F. A research report on read-
ing in Fiji. Fiji English Teachers Journal, 1979, 15, 1-
South Pacific has supported the value of7.
books with local flavor, there is little doubt
FILLMORE, L.W. The second time around: Cognitive and
that imported books are still effective insocial strategies in second language acquisition. Un-
"hooking children on to the reading habit." published doctoral dissertation, Stanford University,
1976.
These trends were observed in both ethnic
HALLIDAY, M.A.K. Explorations in the functions of lan-
groups, and both sexes. Finally, it should be
guage. London: Edward Arnold, 1973.
added that the average cost of books purchased
HANAYAN, E., MARKMAN, B.R., PELLETIER, S., & TUCKER,
for this investigation was less than (US)$2.00.G.R. Differences in performance in elicited imitation
Thus, one formula for raising literacybetween French monolingual and English bilingual
standards in L2 situations such as those children. International Review of Applied Linguis-
tics, 1978, 14 (4), 330-339.
prevailing in the South Pacific is to provide
HOLDAWAY, D. The foundations of literacy. Sydney:
students with a range of suitable, well- Ashton-Scholastic, 1979.
illustrated, high-interest story books, and to LADO, R. Why not start reading earlier? In M. Burt, H.
set aside time in the school program to ensure Dulay, & M. Finocchiaro (Eds.), Viewpoints on
English as a second language. New York: Regents
that they are widely read. Pub. Co., 1977.
REFERENCES MCCRACKEN, R.A. Initiating 'Sustained Silent Reading'.
Journal of Reading, 1971, 14(8), 521-529.
BARIK, H.C, & SWAIN, M. Three-year evaluation of a large-
MACNAMARA, J. Cognitive strategies of language learning.
scale early grade French immersion program: The
In J.W. Oiler & J.C. Richards (Eds.), Focus on the
Ottawa study. Language Learning, 1975, 25, 1-30.
learner. Rowley, MA: Newbury House, 1973.
BROWN, R., & HANLON, C. Derivational complexity and
MILNER, E. A study of the relationship between reading in
order of acquisition in child speech. In J.R. Hayes
grade one and patterns of parent-child interaction.
(Ed.), Cognition and the development of language.
New York: Wiley, 1970. Child Development, 1951, 22, 95-112.
MILON, J.P. The development of negation in English by a
BRUCK, M., LAMBERT, W.E., & TUCKER, G.R. Bilingual
second language learner. TESOL Quarterli', 1974,8,
schooling through the elementary grades. Language
137-143.
Learning, 1974, 24, 183-204.
PELTZ, F.K. The effect upon comprehension of repattern-
CHING, D.C. Reading and the bilingual child. Newark,
ing, based on students' writing patterns. Reading
DE: International Reading Association, 1976.
Research Quarterl/, 1973, 9(4), 603-621.
CHOMSKY, C. Language and reading. In R.E. Shafer(Ed.),
RAVEM, R. The development of Wh- questions in first and
Applied linguistics and reading. Newark, DE: Inter-
national Reading Association, 1979. second language learners. In J.R. Richards4Ed.),
Error anal'sis. London: Longmans, 1974.
CLARK, M.M. Young fluent readers. London: Heinemann
Educational Books, 1976. RICKETTS, J. The effects of listening to stories on compre-
hension and reading achievement. Directions (Fiji),
CLAY, M.M., GILL, M., GLYNN, T., MCNAUGHTON, T., &
1982, 8, 29-36.
SOLOMON, K. Record of oral language. Wellington,
RITCHIE, J. Chance to be equal. Queen Charlotte Sound,
N.Z.: Educational Institute, 1976.
N.Z.: Cape Catley, 1978.
DE'ATH, P.R.T. The shared book experience and ESL.
RIVERS, W.M. Teaching foreign language skills. Chicago:
Directions (Fiji) 1980, 4, 13-22.
The University of Chicago Press, 1968.
DODSON, C.J. Language teaching and the bilingual meth-
od. London: Pitman, 1967. TATE, G. Oral English handbook. Wellington: Reed Edu-
cation, 1971.
DODSON, C.J. Bilingual education in Wales, 5-11. And-
TUCKER, G.R., & D'ANGLEJAN, A. Language learning pro-
over, U.K.: Evans/Methuen Educational, 1978.
cesses. In D.L. Lange (Ed.), Pluralism in foreign
DONALDSON, M. Children's minds. Glasgow: Collins, language education. Skokie, IL: National Textbook
1978.
Co., 1973.
DULAY, H.C., & BURT, M.K. Errors and strategies in child
WIDDOWSON, H.G. Teaching language as communication.
second language acquisition. TESOL Quarter/,. Oxford: Oxford University Press, 1978.
1974, 8, 129-138.
DURKIN, D. Children who readearly. New York: Teachers
College Press, 1966. Reference Note
ELLEY, W.B. A comparison of content-interest and struc-. 1. McKeating, D.F., personal communication, May
turalist reading programs in Niue primary schools. 1980.
New Zealand Journal of Educational Studies, 1980,
15(l), 39-53. Footnote
ELLEY, W.B., & ACHAL, S. The development of standard- 'An additional analysis based on raw scores, matched
ized tests of achievement for Fiji. Suva: Institute ofprecisely by omitting 40 cases from the Class 4 control
Education Report, University of the South Pacific,group, produced almost identical findings to those
1980. reported above.

Impact of reading ELLEY & MANGUBHAI 67

This content downloaded from

86.59.13.237 on Thu, 10 Jun 2021 15:40:40 UTC
All use subject to https://about.jstor.org/terms