(7)af
-
Upload
teslianiptar -
Category
Documents
-
view
27 -
download
1
description
Transcript of (7)af
Reading: Definitions, Types, and Models of Processing
Chapter TwoReading: Definitions,
Types, and Models of Processing
2.1 IntroductionThough our work is primarily concerned with speed-reading, the
direction of the study makes it inevitable to specify a space for discussing
some basic issues that are thought to be more related to normal reading.
Discussing such facts is meant to arrive at a resolution of whether or not
speed-reading is similar to or different from normal reading.
The first section of this chapter will essentially provide an outline
of the attempts of some researchers to define reading which would form
the support we need to present our own definition of it. In the subsequent
section, we will tackle what taxonomies of reading types are there; as
well as what types of reading each taxonomy involves. Then, we proceed
to discuss the processes taking place during reading by looking at some of
the most prominent process models that take into consideration the
processing of low-level functions only. That is, we will describe some of
the work cognitive psychologists have done to understand how isolated
words are perceived, recognized and understood. Such models are to be
compared, then with the position of the componential models(1) that pay
attention to the high-level functions as well.
By so doing, we will arrive at the position we adopt, viz. skilled
reading involves a number of component processes which can be studied
and tested. That is not to say we believe that reading is merely identifying
individual words and stringing the meaning of words together; the
13
Reading: Definitions, Types, and Models of Processing
process of comprehending text is much more complex than that (See 3.8
and 3.9).
2.2 Reading Definitions Our efforts in this section are geared towards discussing
researchers’ divergent viewpoints of reading and to present what they
mean and assume by this term. This shall pave the way for our own
definition of the term.
Like Urquhart and Weir (1998), we start approaching our
conception of the term reading by dictionary definitions. Oxford
Advanced Learner’s Dictionary (2000: 1053) gives sixteen entries for
the word read, of which fourteen refer to the verb. Below, some of them
are given with examples:
1- to show a particular weight, pressure, etc., e.g. what does the
thermometer read?;
2- to guess what somebody else is thinking, e.g. I can read your
mind;
3- to look at and understand the meaning of written or printed words or
symbols, e.g. Some children can read and write
before they go to school;
4- to go through written or printed words, etc. in silence or aloud to
others, e.g. I am going to go to bed and read; and
5- to study a subject, especially at a university, e.g. She is reading
for a law degree.
We should say right away that the first two definitions of reading
are out of our objective here. Furthermore, reading dreams, faces, lips,
road signs, music and the like are also not included in our attempt to
define reading(2). Although they have an obvious legitimate use of the
verb read, but we are more concerned with using the term in relation to
14
Reading: Definitions, Types, and Models of Processing
written texts. Therefore, the last three definitions, mentioned above, are
more promising for our own purpose.
Urquhart and Weir (1998: 22) define reading as “… the process of
receiving and interpreting information encoded in language form via the
medium of print”. In this sense, reading is closely connected with
language texts(3) which requires primarily as an initial step the ability to
decode which is not the only sub-skill. Accordingly, the reader comes to
the reading task with existing language and comprehension sub-skills.
Hoover and Gough’s (1990: 128) central claim of their view of reading is
that it consists of two components, decoding and linguistic
comprehension. These two parts are of equal importance and reading is
not reduced to decoding only, but involves the full set of linguistic skills,
such as parsing, bridging inferences, and discourse building (See also
Ferraro, 1987: 294). It follows that another claim of the simple view (4) is
that “both decoding and linguistic comprehension are necessary for
reading success, neither being sufficient by itself” (Hoover and Gough,
1990: 128).
One more support of such a view of reading comes from
Haberlandt (1994: 2) who distinguishes between different component
processes of reading at the word, sentence, and text level. These
components are insinuated in the generation of meaning representations
of the text. For example, word-level processes comprise both encoding
and lexical access; sentence-level processes include syntactic parsing
processes, and text-level processes establish links between a sentence and
the previous text. Any particular text has meaning and Nuttall (1982: 18)
views reading as essentially concerned with meaning specifically “with
the transfer of meaning from mind to mind: the transfer of a message
from writer to reader”. She has excluded, from her work, any
interpretation of the word reading in which meaning is not central(5). Such
15
Reading: Definitions, Types, and Models of Processing
view of reading is strongly supported by Ajideh (2003: 2) who points out
that reading is regarded not only as a reaction to a text, but also as an
interaction between writer and reader mediated through the text (6). It
follows from this that reading efficiency cannot be measured against the
amount of information contained in a text. Rather, it is a matter of how
effective a discourse the reader can create from the text depending on
either his rapport with the writer or his purpose of reading.
Alderson (1984: 3) views reading both as a product and a process.
Research has concentrated on the product rather than the process, as the
former is, in his view, inadequate for two reasons: first, because of the
unpredictable and normal variation in product. Second, because knowing
the product does not tell us what actually happen when a reader interacts
with a text. Or, in Alderson’s opinion, a product view relate only to what
the reader has got out of the text, while a process view investigates how
the reader may arrive at a particular interpretation.
For the purpose of finalizing our own definition, we shall consider
reading as being composed of both process and product. The process
form the basis of the product which are both different regarding the
cognitive demands that each requires. When one reads, s/he constructs
mental models of what the writer wants to communicate; and to succeed
in communication, i.e. understand meaning, the reader draws both on
existing linguistic and schematic knowledge and the input provided by
the printed or the written text. The two phases of process and product will
vary from reader to reader, purpose to purpose, one type of text to
another, on level of difficulty and familiarity to another, and from one
type of reading to another. Once we include other kinds of reading, than
what is known as normal or careful reading, we may then be tempted to
take the view that different tasks may require different types of reading
and different models of the processes involved.
16
Reading: Definitions, Types, and Models of Processing
2.3 Different Taxonomies of
Types of ReadingTalking about reading assumes distinguishing between different
kinds. This process of differentiating enables us to match a specific
reading type with the needs of the reading material. In fact, many of the
reading models available in the literature tackled careful reading. Rayner
and Pollatsek (1989: 439), for instance, remark that most of their work
discusses careful reading of a textbook through which one should pay
great attention to the written material. In spite of this concern with
reading carefully, they point out that one can also read coping with other
types of reading under certain conditions. No two persons disagree about
the fact that reading a novel, that is entertaining, is much faster than
reading a book on a biological or philosophical matter.
Urquhart and Weir (1998: 101) remark that reading researchers, in
Britain, have to some extent ignored expeditious reading behaviours.
There is “very little on how readers process texts quickly and selectively,
i.e. expeditiously, to extract important information in line with intended
purpose(s)”. In this section, therefore, we shall review some taxonomies
of reading that include other types of reading in addition to the careful
one. Moreover, we are going to clarify what each type means according
to the taxonomy that subsumes it.
2.3.1 Freeman’s Taxonomy (1988)
Freeman (1988: 25) starts his taxonomy of reading types by
describing the skilled reader as “one who varies his reading speed and
method to suit both the material he is reading and his reason for reading
it”. He gives an example of a student who is looking through a list of
17
Reading: Definitions, Types, and Models of Processing
exam results to find his own result. Of course, as a first step, he shall scan
the list very quickly for his name. Finding his name, he will read his
result carefully and might read it more than one time to make things sure.
Freeman claims that this supports his viewpoint that one can vary his type
of reading according to his purpose. Freeman’s taxonomy includes the
types that shall follow.
2.3.1.1 Scanning
Scanning is a very rapid search for some important point which
might be a page number, a title or a key word. Scanning involves
a deliberate ignorance of everything but the item(s) for which the text is
being scanned. Freeman adds that the student needs this type of reading
to scan books or notes for a point for an essay. Moreover, the student may
have to scan periodicals and indexes for items that are important in study.
The most important requirement for scanning is concentration and paying
great attention to the original purpose behind it, otherwise readers would
be bad scanners. Freeman defines a bad scanner as “someone who allows
his attention to be caught by matters which are irrelevant to the purpose
in hand” (ibid.: 26). To get rid of bad scanning, Freeman (1988: 10) sets
out an advice of five steps to apply so that readers would be good
scanners. These steps are:
1- removing irrelevant and unwanted stimuli, e.g. noise, hunger, cold,
2- putting aside other pressing matters by listing or time tabling them,
3- suppressing unwanted thoughts by quickly switching to the topic
under study,
4- understanding what the reader is studying, and
5- taking a lively interest in the subject outside one’s study hours.
18
Reading: Definitions, Types, and Models of Processing
2.3.1.2 Skimming
In skimming, we are simply looking at a text to see what is there. It
is very much like scanning with the exception that in skimming we are
not looking for anything in particular. Freeman gives the following
example to illustrate this difference: suppose one picks up a book on
Democracy to look at the main ideas, chapter headings, and so on to see
whether or not the book is of interest or usefulness to him, the reader here
is skimming. If he is looking to see any paragraph or sentence on, for
instance, the birth of democracy, then the reader is scanning. Thus, when
we are searching for something specific, we are scanning, while when we
are formulating an overall picture of the text, we are skimming.
2.3.1.3 Reading to Study
As the title of this section suggests, this type of reading is
associated with study. Because its aim is to control what is being read, it
is slow and repetitive. The reader ends the reading task, with the hope
that he has comprehended the entire major facts, ideas, and arguments in
the text. Freeman states that reading to study might be broken down into
five stages: Survey, Question, Read, Recall, Review, i.e. SQ3R formula.
He advises the student to use this method seriously and regularly for the
sake of understanding what is being studied (See also Crumpler et al.,
2002, who advise students to employ this formula to survive and succeed
at college). Freeman (1988: 40) summarizes the application of this
formula as follows:
1- Surveying it in two minutes, jotting down what you observe.
2- Noting down the questions you hope to be able to answer by reading
it.
3- Reading.
4- Closing the book. Jotting down the main points of the chapter.
19
Reading: Definitions, Types, and Models of Processing
5- Reviewing what you jotted down in (4) against your questions in (2)
and the chapter itself.
2.3.1.4 Light Reading
Most people adopt light reading most of the time especially in
reading novels. The aim related to this type of reading is escapism, or in
Freeman’s words “…to fly for an hour or two into another world, away
from the problems and distress of the world of today” (ibid.: 26). The
reader, in such reading, does not attempt to digest the material, nor does
he assess the text critically. The value of light reading lies in that much
happiness and relief is achieved or gained.
2.3.1.5 Word-by-Word Reading
There are two most important occasions that demand
word-by-word reading. These are foreign language materials and
mathematical or scientific formulae. Freeman (1988: 27) gives the
difference in reading the following two lines of print:
(a) a stitch in time saves nine; and
(b) methoxyhaemoglobinaemia.
The reader needs only one glance at (a) to understand it, while he would
look at (b) for longer time to know what it means. This difference in the
ability to read (a) and (b) depends on what Freeman calls familiarity with
the material. Common words and phrases are automatically recognized
without the need to focus every word and letter as the case with (a). With
(b), the reader’s eye has to stop and spent more time to recognize the
word, which is only one letter more than (a), because it is an unfamiliar
word.
The other occasion that is much similar to foreign language in
using word-by-word reading is mathematical and scientific formulae. The
author gives an example of the formula H2O that contains the statement
20
Reading: Definitions, Types, and Models of Processing
‘that which is formed when two atoms of hydrogen are combined with
one atom of oxygen’. Being unfamiliar with formulae requires slowing
down one’s reading wherever he comes across a formula(7).
2.3.2 Urquhart and Weir’s Taxonomy (1998)
This taxonomy is a reaction against the restricted attention given to
careful reading at the global level, i.e. comprehension of the main ideas
in a text. The emphasis is often given to the local level, e.g. word
recognition or syntactic parsing.
This is why Urquhart and Weir feel a need to discuss further kinds
of reading in addition to careful reading at the local level. Their main
illustration includes the following: search reading, skimming, scanning,
careful reading (at the global level), and browsing. The first four types
appear frequently in their work and are categorized along two axes of
(a) local vs. global(8), and (b) careful vs. expeditious. Added to that, they
have also examined each type in terms of: purpose, operationalisations,
comprehension focus, text coverage, rate of reading, direction of
processing, and relationship with underlying process. In what follows, we
are going to summarize the exposition put forward by Urquhart and Weir
for these types of reading.
2.3.2.1 Search Reading
Search reading is concerned with locating information on
predetermined topics. This means that the reader is looking for
information to answer question set prior to reading. This type differs from
skimming in that it is guided by predetermined topics so that there is no
need to establish a macropropositional structure for the whole of the text.
The text is processed rapidly and selectively. When the information
searched is located, the reader relies on careful reading. The type of
21
Reading: Definitions, Types, and Models of Processing
processing involved is both bottom-up and top-down (See 2.4.1.1 and
2.4.1.2 for explanation of such processing). The periods of closer
attention to the text tend to be more frequent and longer in relation to
those observed in scanning. Such periods involve more than mere
matching of words. Urquhart and Weir (1998: 214) set out the following
operationalisations that might operate in search reading:
1- Keeping alert for words in the same or related semantic field.
2- Using formal knowledge of text structure for locating information.
3- Using titles and subtitles.
4- Reading abstracts where appropriate.
5- Glancing at words and phrases.
2.3.2.2 Skimming
This type of reading is used to answer the question: what is
a certain text, as a whole, about? So, it is reading for gist that involves
processing a text selectively to get the main idea(s)(9). Both expeditious
and careful reading are involved, here, in addition to bottom-up and
top-down processing. The rate of reading is rapid, but with some care, i.e.
quick processing of the text to locate important information that is to be
read more carefully after that. The purposes of using this type of reading
are:
1- to establish a general sense of the text;
2- to make an outline summary with macropropositional structure; and
3- to decide on the relation between the relevance of texts to the reader’s
needs.
Wherever it is possible, the reader might use one or more of the
operationalisations designated for skimming. These are:
1- identifying the source,
2- reading titles and subtitles,
22
Reading: Definitions, Types, and Models of Processing
3- reading the abstract carefully,
4- reading the introductory and concluding paragraphs carefully,
5- reading the first and last sentence of each carefully,
6- identifying discourse markers,
7- noting repeated key words,
8- identifying markers of importance,
9- skipping clusters of detail, and
10- glancing at any non-verbal information.
2.3.2.3 Scanning
Scanning involves locating a specific symbol or group of symbols
quickly through a text, e.g. finding a number in a directory, names,
figures, dates of particular events…etc. the focus here is on local
comprehension that leads to ignore most of the text. Reading rate is rapid
and the level of processing is surface rather than deep and it is mainly
reader-driven processing(10). The operationalisations used may include
looking for or matching, for example:
1- specific words/phrase,
2- figures/percentages,
3- dates of particular events, and
4- specific items in an index or directory.
2.3.2.4 Careful Reading
Careful reading is the kind favoured by many educationalists and
psychologists to the exclusion of all other types often associated with
reading to learn, i.e. reading of textbooks. Urquhart and Weir (ibid.: 103)
mention the following defining features of careful reading:
1- The reader attempts to handle the majority of information in the text.
This means that the process involved here is not selective.
23
Reading: Definitions, Types, and Models of Processing
2- The reader has a submissive role in that he accepts the writer’s
organization and what the writer appears to consider the important
parts.
3- The reader attempts to build up a macrostructure on the basis of the
majority of the information in the text.
2.3.2.5 Browsing
Urquhart and Weir have included this type of reading in their
taxonomy because not all people are likely to engage in careful reading
for a large part of the time. Consequently, they have added browsing to
describe a sort of reading in which the reader’s goals are not well defined.
Here, readers may skip parts of the text fairly randomly. There is little
attempt to integrate information into a macrostructure only for a topic like
‘this text seems to be about….’. In relation to the resulting
macrostructure, the outcome of the comprehension process is
indeterminate. Urquhart and Weir added that “ it is not a requirement for
browsing that the text be ‘loosely structured’; we can ‘browse’ through
virtually any text, given only that it consists of more than a few words”
(ibid.: 104).
Having established these types of reading, Urquhart and Weir
mention that there is a correlation between one type of reading and
a particular genre of text. Although people are more likely to apply their
careful reading processes to a study text, however such kinds of texts
might be read for amusement. Moreover, readers do not maintain one
type of reading throughout the length of a text. On the contrary, they
jump from one type to another over a small number of pages.
Comparing and contrasting the five types mentioned above
terminates this exposition of reading types. Urquhart and Weir have
24
Reading: Definitions, Types, and Models of Processing
examined the factors that differentiate them and came out with the
following:
1- Skimming, search reading, scanning, and careful reading differ from
browsing. The former types have what is called a clearly defined goal,
which is absent in the latter type.
2- In relation to selective processing, search reading, scanning,
skimming, and possibly browsing involve the factor of selectivity,
while careful reading lacks the presence of this factor. In the first three
types, little attention is given to some parts of texts, while in careful
reading all the text is to be examined.
3- In both careful reading and skimming, the reader’s aim is to construct
a macrostructure, i.e. the gist of the text. In the former type, this is
achieved by reference to the whole text, while in the latter, it is
a matter related only to parts of the text. This kind of construction is
not available in scanning, and in browsing only a vague notion of the
topic might be constructed without mentioned attempt to retain it.
Turning to search reading, it is manifested in searching key ideas in
the macrostructure.
2.3.3 Harvard-Westlake School Taxonomy (2000)
Harvard-Westlake School considers reading the most valuable skill
that one can master. With no mastery, there is no assurance that one can
succeed. In order to talk about reading, this school distinguishes between
different types of reading so that one can fit his reading style to suit the
needs of the material. The types that have been put forward are:
skimming, exploratory, close, review and audible reading(11). For
whatever reason one reads, the primary goal of all of his reading is
comprehension (Harvard-Westlake School, 2000: 1). In the sections to
follow, we are going to elaborate these types a little.
25
Reading: Definitions, Types, and Models of Processing
2.3.3.1 Skimming Reading
According to Harvard-Westlake School, skimming is the most
rudimentary sort of reading. Its aim is to familiarize the reader as quickly
as possible with the material to be read. This is done by leafing through
the material to look at titles, subheadings, illustrations, maps, and, charts.
Skimming is also used to search out passages that have been best lost. In
such a situation, the eyes should race over the pages looking for key
words that help to locate the exact position of those lost materials. In
skimming, the reader feels himself lost, but by practicing this act over and
over he shall refresh his memory and develop it. Harvard-Westlake
School claims that skimming is considered a valuable step in all other
types of reading to the extent that one may skim the material before
starting any type of reading even pleasure reading (ibid.).
2.3.3.2 Exploratory Reading
Harvard-Westlake School treats exploratory reading as half-way
point between skimming and close reading in addition to being similar to
pleasure reading. The aim of using this type is to acquaint the reader with
the subject matter of the text, but with no complete understanding and
retention of it (ibid.: 2). Exploratory reading is used with supplementary
material, and to gain general knowledge from a text. To say that a reader
is using exploratory reading, the following requirements must be met:
1- Reading as quickly as possible.
2- Keeping the mind on the material.
3- Pausing to rest the eyes upon finishing each section of the material.
4- Summarizing what you have just read.
26
Reading: Definitions, Types, and Models of Processing
2.3.3.3 Close Reading
Close reading is the essence of the academic study in that it aims at
full mastery of the material in addition to full retention of details. It is
composed of five separate steps that are vital and end as a whole. These
steps are summarized as follows:
1- Skimming: This step involves skimming the material as described
above.
2- Pre-reading: This step includes the following sub-steps:
One- Reading carefully the first paragraph or the introductory
section.
Two- After that, reading quickly through the body of the material.
Three- Avoiding stopping for unfamiliar words or ideas.
Four- Putting discrete marks in the margin alongside words or
passages that trouble you.
Five- Returning to careful reading when you have reached the final
paragraph or concluding section (For comprehensive details on
pre-reading plans and their benefits, see also Langer (1981) and
Ajideh (2003).
The reader at this point should have a sense of the argument and
development. If this is not the case, the reader should look for the
author’s thesis, which is usually located in the opening paragraph to help
him make sense of the rest of the material.
3- Clarification: Here, the reader clarifies his understanding of the
material by looking up difficult words.
4- Careful reading: During this step of careful reading, the following
requirements need to be met.
One- The reader has to be conscious of forcing himself to read
quickly, but with understanding.
27
Reading: Definitions, Types, and Models of Processing
Two- The use of textual marks to help in retaining the information
later.
Three- Reading with a pencil in hand to underline only those items
to be memorized verbatim.
Four- The use of a combination of marginal checks and notes to
identify important short passages and for a more complete
understanding.
5- Synthesis: It is the last step of close reading through which the reader
tries to synthesize the entire passage. He has to be certain that there is
a sense of continuity and that he understands the whole as well as the
parts.
2.3.3.4 Review Reading
Review reading is the form of reading that involves the rapid
re-reading of the text to refresh the memory. Both the marginal and
textual notes are quite helpful in this concern. The reader starts by
re-reading the introduction and conclusion, then he skims the passages, in
the body of the material, that he understands clearly. As for those parts
that are more confusing, he reads them more closely. Questions or areas
of doubt are put forward to this reading to clear them up by the end of the
reading session (Harvard-Westlake School, 2000: 2).
In addition to those taxonomies presented so far, Carver (1990: 70)
distinguishes between five types of reading, viz. scanning, skimming,
rauding, learning, and memorizing. This distinction is based upon the
different reading processes that take place(12). Carver characterizes, for
example, the types of learning and memorizing as representing more
memory intensive reading processes. The reader’s goal, in the learning
type, is to know the material to enable him to answer typical multiple-
28
Reading: Definitions, Types, and Models of Processing
choice questions about its content. In memorizing, on the other hand, the
reader tries to remember all of the material in as much detail as possible.
Viponed and Hunt (1984-1989) also posit rather different more
general types of reading. These are, viz. information-driven, story-driven,
and point-driven reading. Employing a story-driven type, the reader
focuses, when reading a fictional text, on the plot, events, and characters
to engage with the story. Regarding point-driven story, the reader
distances himself from the text and views it as a product written not for
enjoyment or certain exposition, but he attempts “to construct a version of
what the narrator might be getting at” (Vipond and Hunt, 1989: 157). It is
quite clear that this way of reading underlies academic literary studies.
It follows from what has been said above that our focus is not on
teaching methodologies of these reading types. Rather, it is on the
application of a description of reading types. Although we agree with the
view that readers should be flexible, but Baker and Brown
(1984: 30) have the right to suggest that “there are students who still fail
to set their own purposes, reading everything at the same rate”. However,
we are more concerned with finding out how many of these types are
available to our readers of EFL. That’s why further consideration might
be given to questions like: Are the different types of reading, mentioned
in this section, really reading? If so, do our English learners have access
to all of them? If not, what are the types that learners are restricted to?
Are some types more accessible than others?
2.4 Models of Reading Cognitive Science approach views the mind as an information
processing system. Within this framework, information input to the
system is visualized as flowing between various processes in the brain.
29
Reading: Definitions, Types, and Models of Processing
Therefore, it is cognitive scientists’ job to determine what these processes
are, how the information flows between them, and what each one does
with the information it receives.
The purpose of this section is to explore how the above mentioned
approach has been applied to the study of reading, and to familiarize both
the researcher and the reader with the state of the field. In an attempt to
explain what occurs during reading, different cognitive models and
theories(13) have been posited. Urquhart and Weir (1998: 39) distinguish
two types of models, process models in contrast to componential models
(Cf. de Beaugrande, 1981; Rayner and Pollatsek, 1989; Groome, 1999;
and Zakaluk, 2003). The former type includes descriptions of how words
are memorized, how long they are kept in working memory(14), and so on.
The latter merely describes what components are thought to be involved
in the reading process, with little or no attempt to say how they interact,
or how the reading process actually develops in time (Urquhart and Weir,
1989: 39). Below, we shall cover these models with examples of some of
the more influential ones.
2.4.1 Process Models
Process models are either sequential or non-sequential. Some
process models are sequential in that the reading process involves a series
of stages. Every single stage is complete before the next stage begins.
Others are non-sequential, because there is simultaneous information
available from different sources to synthesize (See Chang, 1983: 217ff).
Process models development goes roughly as follows. There is, first of
all, what is called bottom-up models which was replaced by the top-down
models that are in turn replaced by interactive models.
30
Reading: Definitions, Types, and Models of Processing
2.4.1.1 Bottom-up Models
In bottom-up models, the processing of information is very fast and
flows through the processing system in a series of stages, or in a serial
fashion. These models have little influence from general world
knowledge, contextual information, or higher order processing strategies.
Downing and Leong (1982: 209) remark that these models are also
referred to as data-driven which work their way from sub-skills to the
integration of skills, or from sensory representation to the
syntactic-semantic level. In other words, they are initiated by stimulation
at the bottom end of the nervous system, i.e. the sense organs, which then
proceeds up towards the higher cortical areas. For such kind of
stimulation, Groome (1999: 14) calls these models as stimulus-driven in
which the type of processing carried out is determined by the nature of
the incoming stimulus.
A representation of the bottom-up models is depicted in the
following figure.
Fig. (1) Bottom-up Models (Based on Zakaluk, 2003: 3)
31
MEANING (deep structure)
WORDS (lexical level)
LETTER/SOUNDS (character level)
Level I
Level II
Level III
Reading: Definitions, Types, and Models of Processing
The models of reading often cited as bottom-up are Massaro
(1975), Laberge and Samuels (1974), and Mackworth(1972). However,
the most comprehensive and influential model was proposed by Gough
(1972). In his one second of reading, Gough has posited a number of
representations and processes which information must go through to
attain its first final form. Each letter is first thought to be captured by the
visual system via the initial fixation as a set of lines, curves, and angles
which are stored as an iconic representation. An icon may contain
materials corresponding to 15 or 20 letters. This icon will persist until it is
replaced by the icon arising from the second fixation some 250 msec
later(15). These representations are recognized then by a scanner. The
recognized patterns are stored after that as letters in what Gough calls the
character register. In collaboration between a decoder and some
code book, the string of letters in the character register is converted into
a string of systematic phonemes. By a process known as the librarian,
and with the help of the lexicon, the string of phonemes is recognized as
a word. All the words in a sentence are processed in the same manner.
After that, all the words proceed to an area which is known as the
merlin in which the syntactic and semantic rules operate to assign
meaning and attain understanding of the sentence. The sentences proceed
then to Gough’s TPWSGWTAU area (The Place Where Sentences Go
When They Are Understood). At this point, the text has been read and
understood. To vocalize this text, sentences from TPWSGWTAU are
an input to an editor, which scans them for phonological correctness, and
then passes them to an overall contextual script.
This chain of events taking place during one second of reading may
be visualized as in figure (2).
32
Reading: Definitions, Types, and Models of Processing
Fig.(2) Gough’s Model of Reading (1972)
(Adapted from Katz, 2003:3)
33
VISUAL
SYSTEM
VOCAL
SYSTEM
SUPPOSE THE EYE… “SUPPOSE …”
ICON SCRIPT
SCANNER
CHARACTER
REGISTER
DECODER
PHONEMIC TYPE
EDITOR
TPWSGWTAU
MERLIN
PRIMARY MEMORY
PATTERN RECOGNITION
ROUTINES
CODE BOOK
LIBRARIAN
LEXICON
PHONOLOGICAL RULES
SYNTACTIC & SEMANTIC
RULES
Reading: Definitions, Types, and Models of Processing
In addition to the weaknesses of Gough’s (1972) model mentioned
by Rayner and Pollatsek (1989: 465ff), and Urquhart and Weir (1998:
41), we believe that it has other limitations. According to this model,
information flows in a bottom-up fashion. This amount to say that
information sequentially proceed from lower to higher levels in a way
that each level can directly affect only the level immediately above it.
The model does not take into consideration the effect of higher level on
lower level processes. It goes without saying that context or schema can
affect the meaning ascribed to a word and overall comprehension. This is
not allowed in Gough’s model, which does not provide for schema effects
until after meaning has been assigned. Furthermore, Groome (1999: 14f)
points out that perception of complex stimuli is difficult to be explained
by these models. This difficulty is due to the fact that these models
“assume that the stimulus strikes an entirely nervous system”.
For these reasons, theorists assume that there is a second type of
processing in reading based on the generation of schemas acquired from
past experience. These schemas are sent down the nervous system to be
compared with the incoming stimulus. This type of processing is known
as top-down, schema-driven, or concept-driven processing.
2.4.1.2 Top-down Models
The major motivation for top-down models to be invested is the
need to overcome various bottlenecks in the processing system. The
reader achieves this goal by using world knowledge and contextual
information from the passages being read in making hypotheses about
what will come next during reading. The reader’s job is seen as the
formulation of hypotheses and their confirmation by sampling the visual
information on the printed page (Rayner and Pollatsek, 1989: 462 and
de Beaugrande, 1981: 263).
34
Reading: Definitions, Types, and Models of Processing
In this sense, it is evident that the flow of information proceeds, in
these models, from the top downward. That is why the process of word
identification is dependent upon meaning first. Thus, both higher level
processes of past experience and the reader’s knowledge of the language
pattern interact with the flow of information and direct it. The direction of
processing is quite clear in figure (3).
Fig. (3) Top-down Models (Based on Zakaluk, 2003: 4)
The best-known top-don models of the reading process are those of
Goodman (1970) and Smith (1971). Goodman is often cited as the
representative of these models. He considers reading a process of
hypothesis verification whereby the reader uses selected data from the
text to confirm his guesses. The reader in this model starts with an eye
fixation on a point on the line in order to select graphic cues from the
field of vision. This process of selection helps to formulate a perceptual
35
MEANING (deep structure)
SYNTAX (language pattern)
Grapheme/Phoneme Correspondence
(letter/sound relationship)
Level I
Level II
Level III
Reading: Definitions, Types, and Models of Processing
image of part of the text; and it is guided by a number of factors including
reader’s strategies, cognitive styles, and prior knowledge. The formulated
perceptual image is formed partly of what the reader sees and partly of
what he expects to see.
The next step is to enrich this perceptual image. This is achieved
by the reader searching his memory for related syntactic, semantic, and
phonological cues(16). After that, he tries to make a tentative choice
according to the graphic cues. If the reader is successful in guessing the
word, the resulting choice is held in STM, but if he is not successful, he
looks back to the earlier text. When the choice is made, it is tested against
the prior context for grammatical and syntactic acceptability. If it fits in
with the earlier material, its meaning is assimilated with prior meaning
from the text and the results are stored in LTM. Here, a new hypothesis
about the forthcoming text is made and the cycle is repeated. Goodman’s
model is schematically outlined as in figure (4) below.
36
Reading: Definitions, Types, and Models of Processing
Fig. (4) Goodman’s Model of Reading (1970) (Adapted from Rayner & Pollatsek, 1989: 463)
37
Reading: Definitions, Types, and Models of Processing
The top-down models of the reading process, Goodman’s model is
one among them, are also challenged. An opposing view is that of
Stanovich (1980: 42). Stanovich draws on more than one study to provide
a contrary view to that of the top-down theorists. He states that (1) good
readers are more mindful of graphic information than poor readers; (2)
skilled readers did not prove to be more reliant on contextual information
than unskilled readers; and (3) good readers are predominantly text-
driven. This reinforce that good readers do rely on graphic information
which may be more efficient than predicting words based only upon
context and language structure. Moreover, good readers use context when
orthographic and phonemic cues are minimal (See also Urquhart and
Weir, 1998: 44).
A further critique of these models is presented by Wildman and
King (1979: 137) who suggest that the implementation of the hypothesis
test strategy as an approach to word identification impedes rather than
enhances reading speed. The amount of time required to generate
a prediction by the top-down procedure simply does not account for the
high speed word responses of fluent readers.
Most of our reading material is in fact neither sufficiently
stimulus-driven processing nor adequately predictive for the
hypothesis-test processing. Neither of these processes operates effectively
in isolation as the unique means of word identification. The third
alternative, then, would sustain a balanced view of the reading process.
The interactive model described as follows as such a theory.
2.4.1.3 Interactive Models
Since neither the bottom-up nor the top-down model of the reading
process totally accounts for what occurs during reading, cognitive
psychologists have proposed the interactive models. In these models,
38
Reading: Definitions, Types, and Models of Processing
readers draw upon both the bottom-up and the top-down information
before they arrive at an eventual interpretation of the text. Grabe (1991:
383) interprets the term interactive approaches as that referring to “the
interaction of many component skills potentially in simultaneous
operation; the interaction of these cognitive skills leads to fluent reading
comprehension”.
The most frequently cited example of the interactive models of
reading is that of Rumelhart (1977) whose model aims at providing
a framework for the development of models that are alternatives to serial
processing mechanisms (Rayner and Pollatsek, 1989: 468). Furthermore,
the model is based on the assumption that our perception of words
depends not only on lower level functioning, but also on higher level
contexts which include the syntactic and semantic ones and the
interaction between them. Reading here is seen as a synthesis of patterns,
calling for the application or integration of different knowledge sources,
viz. the orthographic, lexical, syntactic, and semantic knowledge
(Katz, 2003: 8). The basic design of this model is depicted in figure (5).
39
Graphemic Input
Feature Extraction
Device
PatternSynthesizerVIS
Most Probable Interpretation
Syntactical knowledge
Semantic knowledge
Orthographic knowledge
Lexical knowledge
Reading: Definitions, Types, and Models of Processing
Fig. (5) Rumelhart’s Interactive Model of Reading (1977: 730)
The pattern synthesizer receives input from the different forms of
knowledge mentioned above in order to determine the meaning of a text.
This pattern synthesizer is visualized, by Rumelhart, as a message centre
which is constantly being scanned by each source of knowledge for the
appearance of hypotheses relevant to its own sphere of knowledge. The
centre, then, evaluates that hypothesis which is either confirmed or
disconfirmed. Reading, according to Rumelhart, is thus neither
a bottom-up nor a top-down process, but a synthesis of the two(17)
(See also Downing and Leong, 1982 and Urquhart and Weir, 1998).
Another proponent of interactive models is Stanovich (1980) who
calls his model an interactive-compensatory one. The term compensatory
refers to the idea that any weakness in any one of the knowledge sources,
mentioned in Rumelhart’s model, can be compensated for by strength in
another source (Fecteau, 1999: 476).
By combining the bottom-up and top-down approaches and
positing multiple knowledge sources working in parallel, interactive
models avoid the problem of unidirectionality. However, these models
suffer from the opposite extreme of componential complexity that is
40
Reading: Definitions, Types, and Models of Processing
evidenced in the so many interactions taking place and leading to
a complex function; which in turn requires the message centre to manage
and control the constantly changing and rejected hypotheses (Katz, 2003:
10). What is more, these models are weak from the experimental point of
view. They can account for all sorts of data, and are very good at
explaining the results, but at the same time, they are poor at predicting
such results in advance.
41
Reading: Definitions, Types, and Models of Processing
2.4.2 Componential Models
All the models of reading looked at so far have attempted to
describe the actual process of reading. In this section, we shall describe
alternative theories that try to model not only the reading process, but the
reading ability as well. Hoover and Tunmer (1993: 4) state that the use of
componential models is “to understand reading as a set of theoretically
distinct and empirically isolable constituents”.
More than one example of such models are available in the
literature at our disposal. Hoover and Tunmer (1993) is the simplest
model which consists of only two components, viz. word recognition(18)
and linguistic comprehension. It was originally inspired by the simple
view of reading proposed by Gough and Tunmer (1986) and Hoover and
Gough (1990). Other models include three components. Examples are
Coady (1979), Bernhardt (1991b), mentioned in Urquhart and Weir
(1998: 49), and Breznitz and deMarco (2002). The components of
Coady’s model are conceptual abilities, process strategies, and
background knowledge. Bernhardt’s model involves language, literacy,
and world knowledge. Breznitz and deMarco’s model has decoding,
comprehension, and reading rate as its components.
2.5 Some Components in
DetailsIn componential models, a component is defined as
“an independent elementary information processing system that operates
upon internal representations of objects and symbols” (Joshi and Aaron,
2000: 87). Building on this, reading is composed of more than one
independent component. While reading, one of these components may
42
Reading: Definitions, Types, and Models of Processing
function normally where as other components may fail to function in
an optimal level.
In what follows, we shall discuss low-level functioning
components, viz. word recognition and lexical access in details.
Although other high-level functioning components mentioned earlier are
of equal importance, we opt, however, to delay discussing them, with
other ones, to the next chapter (See 3.9); to examine whether or not they
have any mentioned effects in relation to reading speed, our main
concern.
2.5.1 Word Recognition
The meaning of the term word recognition is rather disputed. It has
come across more than one elaboration until an ultimate definition of it
has been formed. Hoover and Tunmer (1993: 8), for instance, mention
three interpretations of word recognition. The first one has “the term to
mean recognizing an English word in print, be able to pronounce it, and
give its meaning”. It seems that Hoover and Tunmer forgot all about the
use of pseudo-words that are involved in many experiments on word
recognition. So to include such words, they extend the term, in their
second interpretation, to mean “recognition of pronounceable strings of
letters, which are not actual words in English”. However, there are also
unpronounceable pseudo-words. Thus, their third interpretation has
involved “the recognition of any letter string with space boundaries on
either side”.
Despite this, the other pertinent fact is that the process of word
recognition is still extremely confusing and not well understood. For this
reason, we will content ourselves with what is generally agreed upon.
From the standpoint of the simple view of reading of Hoover and Gough
(1990: 130), efficient word recognition is simply skilled decoding. They
43
Reading: Definitions, Types, and Models of Processing
define it as “ the ability to rapidly derive a representation from printed
input that allows access to the appropriate entry in the mental lexicon,
and thus, the retrieval of semantic information at the word level”. This
means that there must be efficient, i.e. fast and accurate, access of the
mental lexicon for proper, arbitrary orthographic representations.
Generally, the adequate measure of decoding, for Hoover and
Gough, must tap the ability to access the mental lexicon for arbitrary
printed words. In relation to beginning readers, an adequate measurement
must assess skill of deriving appropriate phonologically based
representations of novel letter strings. It is quite clear that Hoover and
Gough’s definition of decoding is inconsistent with their measurement of
it. Although the words efficient and rapidly are used in the definition,
however, there is no mentioning of either rate or efficiency in the
measuring criterion used (See also Carver, 1993: 441).
Given that the purpose of word recognition is to access the lexicon,
and derive meaningful representations, Goodman (1982b: 53) equalizes
these representations to meaning. He considers decoding as going from
language to meaning which contrasts with encoding, going from meaning
to language. So there is no decoding without a message. Anything short
of meaning, and does not go from code to meaning is not decoding. If the
reader shifts from print to sound with no meaning resulting, then we have
what Goodman calls recoding, a transforming of code to code. This
recoding process is one of the steps that are involved in word recognition
according to Hirai’s (1999: 368) view. The latter states that word
recognition may involve (a) a pattern analysis of the visual stimulus, (b)
some phonological recoding, and (c) lexical search on the basis of either
the visual or phonological representations, or both. This issue has been
originally raised by Rubenstein, Lewis, and Rubenstein (1971: 655) who
sharpened the phonological recoding hypothesis which holds that word
44
Reading: Definitions, Types, and Models of Processing
recognition is done in two stages. The first one is the stage of converting
a string of letters into a string of phonemes by the grapheme-phoneme
correspondence rules. For example, a string of letters, like thin, is to be
converted into a string of phonemes, like /In/. The second stage includes
searching the mental lexicon to find an entry that matches the string of
phonemes /In/. When this entry is found, the word is recognized.(19)
Another important word aspect, which is of great significance for
our concern is sight reading skill. This skill emerges as a result of
treating decoding as a basic requirement for word recognition. Thus, it is
not treated as an independent component, in the componential models,
but as a skill built on decoding and, therefor, not independent of it
(Ehri, 1998: 98 and Aaron et al., 1999: 95). Sight-word reading is
considered a speeded up decoding process, i.e. we have the following
equation: decoding + speed = sight-word reading. Rapid and automatic
recognition of words and the rate of processing are important factors to be
reckoned with in reading. When the process of decoding becomes
automatic, by training to increase its speed, reading comprehension
would be improved. In this way, the reader’s attention would be free to
process meaning (Grant and Standing, 1989: 519). In this concern, Hirai
(1999: 369) remarks that “the speed of word recognition appears to be
a crucial factor underlying fluent reading”. Nevertheless, the disputed
issue of considering reading speed, as an independent component is not
resolved yet (Joshi and Aaron, 2000: 87).
Turning to the factors that affect the speed and accuracy, with
which a word is identified, the most clearly established factor is the
frequency with which a certain word is encountered in the language.
Groome (1999: 223) points out that frequency can be measured by taking
a sample of books, magazines, or any other printed material and counting
the number of times that a particular word appears. These occurring times
45
Reading: Definitions, Types, and Models of Processing
amount to some twenty million words in the case of the Thorndike Lorge
Word Frequency Count (1944) or to one million words in the case of the
Kucera and Francis Count (1982). Rayner and Pollatsek (1989: 68) give
examples about the frequency of words such as IRK, JADE, COVE,
VANE, and PROD that have counts of 2 to 5 per million; and words like
CAT, COAT, GREET, and SQUARE that have frequency more than 30
per million. A high frequency word such as COAT is different from a low
frequency one as COVE in the lexical decision time which is about 100
msec. They also differ in their naming times which is 30 mesc.
Rayner and Duffy (1986) investigated the effect of word frequency
on reading. They measured eye fixations during reading that enabled
them to remark that low frequency words were fixated for about 80 msec
longer than high frequency words. Rayner and Pollatsek (1989) consider
the difference in fixation time, as the better guess of the effect of
frequency on the time needed to identify a word. In this sense, it is quite
clear that high frequency words are recognized faster and more
accurately, i.e. more efficiently than low frequency words.
Another important factor of word recognition efficiency is the
context in which a word occurs. This issue has been the major concern in
cognitive psychology that always questions whether words are recognized
less quickly and accurately in isolation than when they follow a preceding
word or is part of a sentence. More specifically, there is evidence for such
context effects on word recognition; but it is “unclear whether all of these
effects can be ascribed to intra-lexical processes, … or whether
an interaction between higher-order processes and lexical access needs to
be invoked” (ibid. : 221).
One way of investigating the effect of context is via eye-fixations
experiment. It has been demonstrated that in a particular context, words
are fixated for shorter periods and skipped more often than in
46
Reading: Definitions, Types, and Models of Processing
an unpredictable context. Zola (1984: 280) gives examples on such types
of effects on the recognition of a certain word. He also gives the
following two sentences:
(1) Movie theatres must have buttered popcorn to
serve their patrons.
(2) Movie theatres must have adequate popcorn to
serve their patrons.
The word popcorn in the second sentence is fixated longer than in the
first. Zola justifies this by the limitation of the range of possible
succeeding nouns imposed by the word buttered in (1) than does the word
adequate in (2). Hence, popcorn is recognized faster in (1) than in (2).
The other way in this concern, i.e. context effects, studies the effect
of what is known as semantic priming. This relates to whether the
meaning of a preceding word or sentence can affect the recognition of
a succeeding word. The task used in these studies is the lexical decision
task where two words in sequence, the prime and then the target, are
presented on a screen. The experimenter, then, measures how quickly the
second word, i.e. the target, is processed. Subjects are faster in processing
when the words are related (e.g. cat-dog) than when they are unrelated
(e.g. cat-pen). This demonstrates that the speed of processing the target is
dependent on the meaning of the prime (See Meyer and Schvaneveldt,
1971: 229 and Balota, 1992: 8).
Additional studies provide an abundant body of evidence regarding
factors affecting word decoding. These studies suggest that the major
determinant of word decoding development is phonological abilities
(Goswami and Bryant, 1990: 10; Share, 1995: 182; and Elbro, 1996: 8).
dejong and van der Leij (2002: 52) point out that such abilities “pertain to
the ability to detect, to store, and to retrieve the basic sound elements of
one’s oral language”. What is more is that Abu-Rabia and his colleagues
47
Reading: Definitions, Types, and Models of Processing
(2003: 427) treat these abilities not only as a determinant of word
decoding, but more importantly as one of the basic cognitive processes
taking place in word recognition. They define it as “the knowledge that
spoken words are composed of phonemes and syllables”. Wagner and
Torgesen (1987: 196) distinguish three types of phonological abilities,
viz. phonological awareness, verbal working memory, and rapid naming.
The relation between phonological awareness and rapid naming on the
one hand and word decoding on the other is causal in nature. This nature
is supported by longitudinal studies, and studies of children with dyslexia.
Such support stems also from studies that have demonstrated that the
training of phonological awareness enhances the acquisition of word
decoding. The other relation is between verbal working memory and
word decoding which is mainly correlational. However, the effects of
verbal working memory on word decoding were small and could be
completely accounted for by phonological awareness (Wagner et al.,
1994: 85 and 1997: 475).
2.5.2 Lexical Access
Closely related to word recognition is the concept of mental
lexicon. The latter was proposed by Treisman (1960, 1961) and further
developed by Oldfield (1966) both cited in Downing and Leong
(1982: 158). The lexicon is defined as:
An abstract associated network of information. In it,
each event or concept has a unique internal
representation. These internal representations have
different degrees of association with one another,
depending on how frequently they have been
continguously activated
(ibid.)
48
Reading: Definitions, Types, and Models of Processing
Given that the purpose of word recognition is to access the lexicon,
it is generally agreed that there are three different conditions under which
the lexicon has to be accessed, viz. in reading, listening, and talking. In
reading, the lexical items are arranged in a system similar to pages of
words of similar orthographic properties. What is of significance for our
concern, here, is questioning the way the reader proceeds from the
information that he obtains from a stimulus word to the internal lexical
entry related. It is a process of transition from perception to conception;
from a series of black-on-white characters to the corresponding concept.
More directly, how do readers access the appropriate word so efficiently?
Do they search right a way through the lexicon? Or, are there other routes
that they have to follow?
Both Rayner and Pollatsek (1989: 84) and Urquhart and Weir
(1998: 52) recognize two routes for accessing the lexicon. The first is the
direct route that goes directly from the visual input to meaning. The
second one is indirect and known as the phonological route, since it goes
from visual input to meaning via sound. Downing and Leong (1982)
mention another route known as the dual or parallel route. It is termed as
such, because it mixes both visual and phonological input to access the
lexicon. When we turn to Groome (1999: 224ff), we find him employing a
rather different typology of the models of accessing the lexicon. He
differentiates between simultaneous access models, and serial search
models. Simultaneous models assume that there are feature detectors, in
the lexicon, for each lexical item that are simultaneously activated by
many lexical entries. This activation process continues until one becomes
pre-eminent. Within the serial search models, the lexicon contains
a master file of words linked to a series of access file which are of three
types, orthographic, phonological, and syntactic-semantic access files.
These files are searched to match the word perceived by directing the
49
Reading: Definitions, Types, and Models of Processing
master file which holds all the information of the word including meaning
(For more details on these models, see Forster, 1976).
To sum up, Rayner and Pollatsek (1989: 109) come up with the
conclusion that the common ground for all positions is that direct visual
access is important and that sound encoding plays some part. The direct
route alone cannot explain the subject’s ability to handle pseudo words
like mand, or brin which are not likely to be present in the reader’s
lexicon. Moreover, the phonological route is necessary to explain the
phonological influence on word recognition. For example, the recognition
of words like touch slows down, when they are preceded by words like
couch (Urquhart and Weir, 1998: 52).
50
Reading: Definitions, Types, and Models of Processing
Notes to Chapter Two1- Some of the high-level functioning components of these models are to
be discussed later on in chapter three to examine their relation with
speed-reading in particular.
2- Our main focus is on how to read language messages rather than
reading maps, for example. In other words, we are concerned with the
non-technical, every day use of the word.
3- This kind of connection is viewed differently by different scholars. For
such views, see Downing and Leong (1982).
4- See Carver (1993) who merges this simple view with what he calls
rauding theory.
5- Cf. Zakaluk (2003: 11) who states that “...reading for children is
a matter of word calling and correct pronunciation than a search for
meaning”.
6- Rusenblatt (1994), cited in Falk-Ross (2002: 2), characterizes the
reading process as a transaction between the reader and the text,
strengthening the importance of the reader’s prior knowledge and
goals.
7- The only respect that foreign languages and mathematical and
scientific formulae differ is that the latter are highly condensed
methods of conveying information.
8- According to Kintsch and van Dijk (1978), local comprehension refers
to the decoding of micropropositions and the relations between them,
while global comprehension is related to the level beyond that of
micropropositions, i.e. from macroproposition to discourse topic.
9- Hoover and Tunmer (1993) distinguish between skimming and reading
for main ideas, but for Urquhart and Weir (1998), they appear to be the
same.
10- This type of processing is the opposite of that of text-driven. In such
kind of processing, i.e. reader-driven, the reader has a previously
51
Reading: Definitions, Types, and Models of Processing
formed plan, and might delete chunks of the text that are irrelevant to
his purpose (Urquhart and Weir, 1998: 42).
11- We opt not to discuss this type since it is related to reading aloud.
12- In other attempts, Carver (1992, 1993) points out that in rauding
theory, the most important kind of reading is the one associated with
the reading process called rauding, i.e. typical or normal reading.
13- Downing and Leong (1982: 202) state that “the terms, model and
theory are often used interchangeably. The difference is relative”.
14- Earlier models of human memory storage distinguished between two
memory storages, viz. short-term memory (STM) and long-term
memory (LTM). The former is a part of entry into the latter. With the
accumulation of evidence that these storages are separate memory
stores, Atkinson and Shiffrin (1968) developed the concept of the STM
as a conscious working memory.
15- From the time the icon is formed, the letters are identified and read out
at the rate of 10 to 20 msec per letter. With an icon persisting for some
250 msec and with at least 3 fixations per second, Gough estimates that
the rate of reading can be in excess of 300 words a minute.
16- Rayner and Pollatesk (1989: 462f) point out that the way of relating
these cues to a perceptual image is not clear unless the image is
identified as a sequence of letters or a word.
17- Hirai (1999: 368) remarks that most versions of interactive approaches
to reading have taken a story bottom-up orientation to the processing
of lower level linguistic structure.
18- Hoover and Gough (1990: 130) use the term decoding to refer to
efficient word recognition (Cf. Hoover and Tunmer, 1993).
19- It is important to note that these two stages work properly on both
spoken and written input.
52