Review of Early Reading Instruction
|Date: Mon, 23 Aug 2004 13:13:16 -0700
From: Stacia Levy
Subject: Early Reading Instruction
AUTHOR: McGuinness, Diane
TITLE: Early Reading Instruction
SUBTITLE: What Science Really Tells Us about How to Teach Reading
PUBLISHER: MIT Press
Stacia Levy, University of the Pacific
In this book, the author provides for teachers and
scholars a comprehensive review of the research on
effective reading methods, starting with addressing
early writing systems and how they developed and
continuing to the contemporary times. She critiques
the work of National Reading Panel (2000) in screening
and reviewing thousands of research studies on
reading. Thoughout the book, the author addresses
the ''reading wars,'' that is, the conflict over which
method works best, ''phonics'' or ''whole word,'' coming
down strongly on the side of phonics, showing that all
languages are based on sounds, and all alphabetic
languages are based on sound-letter correspondences
which must be explicitly taught. The English sound-
letter system is particularly complex or ''opaque,'' as
the author calls it, leading to a high failure of
reading in English speaking countries. She also
critiques research methodology and develops a
prototype for an ideal phonics program. She ends by
suggesting that the research needs to move beyond the
''reading war'' of phonics or whole word, as we have
known for years that phonics is the only effective
method of early reading instruction, and now we must
move forward to determine which elements of phonic
programs are most effective.
The author addresses world writing systems and the
high failure rate in literacy in English-speaking
countries. She introduces the work of the National
Reading Panel (NRP) and its attempt to sort through
the database of reading research of 1,075 studies over
the past thirty years, only 38 of which were found
methodologically sound. The author also addresses the
notion of ''phoneme,'' or individual speech sound, an
important concept in reading instruction because in
alphabetic writing systems, letters represent
phonemes: the spelling ''code.''
The author provides an overview of the book's
purposes: to review the research on reading teaching
methods, provide insights into our spelling code and
how to teach it, provide a prototype for an effective
reading program based on the research, and critique
the methodology of reading research. The author also
touches on how to teach writing systems, which are
based on phonemes or other units below the level of
the word, such as syllables or consonant-vowel units.
Because there are so many words in a language, ''no one
can learn to read by memorizing whole words by sight''
(p. xiv), and instruction should be based on whatever
unit the writing system was originally designed of.
European languages are written in alphabets, which are
based on the phoneme. The author then touches on the
''opaqueness'' of the English alphabet system, with its
many exceptions and multiple spellings for the same
phoneme. This leads into chapter one.
This chapter reviews statistics of literacy in
different countries. English-speaking countries have a
high illiteracy rate compared to European nations:
e.g. 43% among American nine-year-olds (Mullis,
Campbell, Farstrup, 1993), compared to Austria, where
even ''poor readers'' in fourth grade scored close to
100% correct on reading accuracy and spelling tests
(Wimmer 1993). This is due to the ''opaqueness'' of the
English spelling code, according to the author, with
its many exceptions, compared to many European
nations' codes with their nearly ''one-to-one''
correspondence of phonemes and letters. In addition,
reading instruction is systematic in countries like
Austria, where children are taught the letters and the
sounds they represent, and reading and spelling are
integrated. In English-speaking countries, with an
opaque writing system and whole-word methods, neither
of these is true.
The author also states that the notion of ''dyslexia,''
or reading failure, presumed to be the property of
individual children and of biological basis, is
untenable, as the disorder afflicts only English-
Chapter 2 On the Nature of Writing Systems
The author takes a historic view on the development of
writing systems and their properties. The author deals
with writing systems as ''codes'' which can be broken
and which are reversible: that is, what can be put
into code or encoded can also be decoded. So words
that are put into print can also be read. She again
addresses the phoneme and letter as the basic units in
an alphabetic system. Writing systems are invented by
scholars; they do not ''evolve.'' As human memory has a
limit for the symbols it can contain, some writing
systems transitioned from whole-word to phonemic
systems but did not evolve. In addition, people do not
develop the ability to hear phonemes, an ability
present at birth in normal-hearing infants. Because
writing systems are inventions and not a part of our
evolutionary heritage, as is speech, people have to be
specifically instructed in their use. And because our
memory has a limit for the amount of symbols it can
contain, whole-word methods doom the majority of its
learners to failure.
Chapter 3 The Structure of the English Alphabet Code
The author again addresses the nature of the English
writing code and ways to teach it: teach the sound
units first, then the symbols that represent them,
from simple to complex. Students must see that this is
a reversible code: reading and spelling are connected
at every level through seeing, hearing, and writing.
The author also addresses the reason for the
opaqueness of the English spelling code: the
influences of a number of languages on English and
attempts historically to reform our spelling system.
Chapter 4 How to Teach Reading: Lessons from the Past
Here the author revisits a lot of ground already
covered: the opaqueness of our writing system, the
NRP's review of the literature and how it has provided
''incontrovertible support'' that whole word methods
''lead to consistently lower reading test scores than
methods that emphasize phoneme-grapheme
correspondences'' (p. 74) or correspondences between
sounds and letters. The author also covers past
research on effective reading instruction and problems
in doing research on teaching methods. An example is
the 1963 Cooperative Research Program, a massive
project which set out to study first graders and
reading methods to ''settle'' the phonics/basal reader
debate. Basal readers tended to emphasize whole-word
methods; phonics programs teach sound-symbol
correspondences. A number of classrooms were observed
and pre-and-post test scores reported. However,
because of problems in the final statistical analysis,
such as looking at mean test scores from each
classroom rather than individual student scores, the
research results were ambiguous. This was the last
study of its kind, possibly due to later budget cut
Chapter 5 How to Teach Reading: Modern Research
The author addresses the ''whole language'' movement,
popular throughout the seventies and eighties: the
''natural language'' approach of listening to stories,
reading along, creative writing, and invented
spelling. It was fun, according to the author, and
''and it took the English speaking world by storm ... with
catastrophic consequences'' (p. 108).
Research in the 1960s and 1970s had found that
activities such as sound-to-letter analysis, writing
letters and words, teaching the basic code of 40
phonemes (sound units and their letters), identifying
the sequence of sounds in words, integration of
reading and spelling, segmenting and blending, lots of
writing activities, led to increases in reading
scores; memorizing words, vocabulary lessons, using
invented spelling, and listening to stories all led to
lower scores. In a review of the literature (Stahl and
Miller, 1989), whole language's advantages were mostly
in ''nonreading'' effects, such as print readiness and
attitudes toward reading.
>From the research, the author provides a ''prototype of
an effective reading/spelling program'': no sight
words, no letter names, sound-to-print orientation,
teach phonemes only, begin with the basic code, teach
children to write each letter, and link writing to
reading. '' Linguistic phonics,'' a phonics program
designed by linguists, is closest to this prototype.
The author also reviews a number of reading programs,
such as Reading Recovery and Open Court.
Chapter 6: Phoneme-Awareness Training
Here the author addresses programs in ''phonemic
awareness,'' based on the theory that phonemic
awareness is developmental and proceeds from words to
phonemes. This idea is not supported in the
literature. Phonemic awareness training has a low
effect size on standardized reading tests (Bus and Van
Ijzendoorn 1999) but does have an effect on phonemic
In addition, it appears that awareness of phonemes is
present at birth or shortly thereafter as this is how
children learn language: by perceiving the individual
speech sounds in their environment and repeating them.
Chapter 7 Reading Fluency
The author addresses other reading instruction issues,
such as improving speed. Slow readers often are
inaccurate readers as well. There are effective
methods to improve their reading rates, such as
rereading the same passages and setting target rates
over a period of several weeks.
Chapter 8 Vocabulary and Comprehension Instruction
There is solid evidence on effective vocabulary
instruction, such as repeated exposure to the same
words over a short period of time. Exposing children
to new words though stories is effective but only if
the students are guided in understanding the words;
allowing them to try to arrive at meaning through
''context'' is not effective. Apparently this is too
demanding a cognitive task for most children.
Vocabulary can be learned as part of an overall
reading comprehension program.
Chapter 9 How Does Anyone Learn to Spell?
The English spelling system is ''deeply opaque,'' with
multiple spellings for phonemes, only eight of which
are reliable, but even half of those have double
spellings. Spelling requires recall memory, while
reading only requires recognition, so spelling is more
difficult. The author addresses issues of spelling
ability predictors: IQ, sex, and reading scores. She
provides data on poor instructional methods: learning
letter names rather than sound-symbol correspondences,
reliance on sight-word memory and random word lists,
and divorcing reading and spelling instruction. The
author then addresses ''stage models,'' which are based
on the assumption that children teach themselves to
spell and is grounded in Freudian and Piagetan
psychology: stages follow a fixed sequence, do not
progress backward; old learning is integrated into a
new stage and is generalized to new tasks. Stages
cannot be taught. This is in contrast to a learning
model, in which new skills are gained through training
and not a biological program, as is implicit in the
stage model. And, according to the author, reading and
spelling are the result of learning, not of biological
stages. She critically evaluates the literature on
''spelling stages,'' which is weak. Research on poor
spellers has found they score below good spellers on
reading tests although they did not differ on tests of
verbal IQ. They are likely to be poor spellers because
they are poor readers and do not read enough to see
different spelling patterns. The author also includes
instructional tips for teaching spelling, such as
writing out words.
Chapter 10: The Many-Word Problem
The English spelling code is made up of hundreds of
patterns worth teaching: this is the ''many word
problem.'' It is impossible to teach all of the
exceptions. This problem doesn't exist in languages
with ''transparent'' writing systems. For the brain to
solve the many-word problem, it must see a lot of
print. Stanovich and West (1989) explored this: the
best predictor of spelling ability was the author
recognition test, designed to determine reading
Aaron et al. (1998) studied the spelling of profoundly
deaf children compared to hearing children to tease
out the effects of phonological awareness on spelling.
Aaron proposed that deaf readers rely on ''statistical
probabilities in spelling sequences,'' a ''frequency
model'' (p. 284). In addition, while hearing children
rely more on phoneme-grapheme patterns, deaf children
rely on visual processing. However, on further tests,
it was found this visual memory alone is not enough to
spell common English words that were flashed on a
screen for a short period of time; deaf children
performed worse on this test than hearing children.
Here the author moves into a discussion of computer
models of reading and the question of which model fit
the way humans read. Computer models of reading are
statistical and process structural redundancy in the
input, using feedback from the environment about
success. Input on a word is processed in a parallel
rather than disconnected pathway, taking into
consideration the contextual dependencies of the
entire word. All information about a word, such as
phonology and syntax, is accessed at once. The author
found methodological problems with the research on
computer models: there were too many factors involved
in reading for the computer programs to account for.
Seidenburg and McClelland (1989) and Plaut et al.
(1996) found that ''attractor networks'' in computers
can decode new words as well as humans, through
''recurrent connections,'' feedback within the system,
in deciding on the most consistent interpretation of
input (p. 311). Whether or not the human brain does
this, the author points out, is a problem for the
future (p. 312).
Chapter 11 New Directions for the Twentieth Century
This chapter summarizes the book and makes suggestions
for future. Whole word methods don't work; teach the
code: that is, the sounds of the language and the
written symbols that represent them. We have a
prototype for effective phonics instruction. We need
to know now which specific features of phonics
methodology works best: e.g. are kinesthetic/actions
and other ''extras'' incorporated into phonics lessons
really helpful to learning to read? Is phonemic
awareness training beyond linguistic phonics needed in
beginning readers? In addition, some research should
be done on the effects of rereading and its effect on
speed, accuracy, comprehension, and transfer effects.
More work also needs to be done researching and
designing programs that teach the advanced spelling
code: that is, the 136 common spelling alternatives:
work to date suggests that teaching this not only
improves spelling but also reading and reading
comprehension (Smith 1999).
Computer models of reading imply that computer
programs simulate how the human brain reads. The
author has found problems in the research design of
these studies, such as their word lists used to
''train'' the computer program on the input.
In addition, the research is based on logical
fallacies: e.g., that speed of reading words aloud
indicates brain processing time, that fluent readers
recognize words by sight because it seems that way.
''The idea that you can infer something about
perception, cognition, and brain processing from a
single measure of simple response time is extremely
naïveÿÿNo single measure can reliably predict a
reader's efficiency in decoding text'' (p. 342-343). In
addition, the author states, ''Our sense that we read
whole words instantly by sight via some direct
pipeline from the eye to meaning is an illusion. No
matter how much something 'seems like' it happens
instantly, it does not. Conscious awareness and brain
processing run on different clocks'' (p. 344). However,
more recent research has shown that the expectations
readers bring to the text based on syntax and
semantics govern how the text is read'' (p. 344-345).
More research should be done in this area.
1. How Nations Cheat on International Literacy Studies
2. Misuse of Statistics
3. Analysis of Word Lists from Treiman et al.
6. Author Index
7. Subject Index
Overall this is a very thorough review of the current
and past literature on reading instruction. The
author's main thesis is that this research shows that
phonics is superior to whole word methods in early
reading instruction and that we have a prototype of
good phonics instruction; it is now time to move on
and stop investigating the differences between whole
word and phonics methods. She repeats this point ad
nauseam. Her evaluation of the state of reading
research is scathing, from the poorly designed studies
that nevertheless make it into refereed journals, poor
dissemination of results due to publication policies
that allow poorly designed studies into the top
journals while well-designed studies go begging, to
repeated investigation of the same research questions
we already the answer to. Throughout the author calls
for more and better research.
The book is densely packed with ideas, information,
and statistics throughout and can be hard to follow.
It is at times repetitive, making the same points over
and over again: e.g. we now know the best method to
teach reading and the prototype of a good phonics
program. The last chapter basically just reiterates
In addition, some of the chapters are jumbled. For
example, chapter 10 covers too many topics from
spelling instruction to computer models of reading,
making the chapter difficult for the reader to
A last quibble is that the author seems to spend a lot
of time promoting her own phonics program.
Her bias for phonics instruction over whole word
methods is apparent, but after reading the evidence,
it would appear with due cause. Apart from the
concerns mentioned, this is a thorough and thoughtful
review of the literature on early reading instruction.
Aaron, P. G., Keetay, V., Boyd, M., Palmatier, S, and
Wacks, J. (1998). Spelling without phonology: A study
of deaf children. Reading and Writing: An
Interdisciplinary Journal, 10, 1-22.
Bus, A. G., and van Ijzendoorn, M. H. (1999).
Phonological awareness and early reading: A meta-
analysis of experimental training studies. Journal of
Educational Psychology, 91, 403-414.
Mullis, I. V. S., Campbell, J. R., and Farstrup, A. E.
(1993). National assessment of educational progress
1992: Reading report card for the nation and states.
Washington, D.C.: Office of Educational Research and
Improvement, U.S. Department of Education.
National Reading Panel (2000). Report. Washington,
D.C.: National Institute of Child Health and Human
Plaut, D. C., McClelland, J. L., Seidenberg, M. S.,
and Patterson, K. (1996). Understanding normal and
impaired word reading: Computational principles in
quasi-regular domains. Psychology Review, 103, 56-
Samuels, S. J. (1972). The effect of letter-name
knowledge on learning to read. American Educational
Research Journal, 9, 65-74.
Seidenburg, M. S. and McClelland, J. L. (1989)
A distributed, developmental model of word recognition
and naming. Psychological Review, 96, 523-568.
Smith, A. A. (1999). The simple logic of sound-to-
letter mapping: A reversible code. Unpublished
master's thesis, Massey University, Albany, New
Stahl, S. A. and Miller, P. D. (1989). Whole language
and language experience approaches for beginning
reading: A quantitative research synthesis. Review of
Educational Research, 59, 87-116.
Stanovich, K. E. and West, R. (1989) Exposure to print
and orthographic processing. Reading Research Quarterly,
Treiman, R. (1994).Use of consonant letter names in
beginning spelling. Developmental Psychology, 30,
Varnhagen, C. K., McCallum, M., and Burstown, M. (1997).
Is children's spelling naturally stage-like? Reading
and Writing: An Interdisciplinary Journal, 9, 451-
Wimmer, H. (1993). Characteristics of developmental
dyslexia in a regular writing system. Applied
Psycholinguistics, 14, 1-33.
| ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Stacia Levy is an English and education professor
in California. She recently completed her
dissertation, which examined the vocabulary patterns
found in college student and professional writing.
Her areas of research interest include academic
writing instruction, adolescent literacy, and