Beyond the Sentence: Finding a

Balance Between Bottom-Up and Top-

Down Reading Approaches
Date:
January 1998
Issue:
The Language Teacher - Issue 23.1; January 1999
Writer(s):
Padraic Frehan, The British Council Tokyo
This paper argues that drawing students' attention to the advantages of activating their background
knowledge and helping them to activate and start developing their prediction skills, both at the pre-
reading stage and during reading of a text, can help them develop into more effective readers.
The students who were the subjects of this study were already capable of utilizing bottom-up
processing skills acquired in past learning environments but were deficient in, or unaware of, the
benefits of utilizing top-down skills. Since they consistently used a word-for-word text approach, I
decided to focus on the top-down skills of predicting and activating background knowledge in order
to provide them with a wider range of skills when reading a text. I will demonstrate that, when
provided with alternative approaches, students are capable of processing a text more efficiently.
Evidence will be based on written feedback from the participants and observations of what occurred
during the study, which took place during one three-hour lesson at The British Council English
School Tokyo. The retrospective written work was submitted by the students the week following the
study.

The Reading Process

A brief synopsis
The last four decades have seen the emergence of three reading models. In the bottom-up (data-
driven) reading process, the reader decodes, letter-by-letter, word-by-word the written symbols in the
text and then reassembles the pieces to form meaning. However, this process creates problems
such as fragmentation and memory overload because the reader attempts to store too many
separate pieces of information without any higher-order relationship between them (Carrell, 1988b).
Despite these deficiencies, the bottom-up approach has remained popular in the teaching of reading
in Japan (Kitao & Kitao, 1995).
The second model, the top-down (concept-driven) reading process, arose out of psycholinguistic
research by such scholars as Goodman (1971) and Smith (1971). In this model, the efficient reader
does not need to use all of the textual cues (Carrell, 1988a). Goodman (1971) described reading as
a "psycholinguistic guessing game" (p. 135) and later wrote that the better the reader is able to make
correct predictions, the less confirming via the text is necessary (Goodman, 1973, in Carrell, 1988a).
The third model is the interactive processing approach, developed in response to the deficiencies of
both bottom-up and top-down approaches. In this model interaction refers to the constant interaction
between bottom-up and top-down processing skills (Eskey, 1988).
Stanovich's (1980) interactive-compensatory model deals with the shortcomings of both approaches.
The bottom-up model assumes background knowledge cannot be activated before lower level
decoding while the top-down model does not allow lower level processes to influence or direct higher
level ones. The basic premise of Stanovich's model is that reading involves an "array of processes"
(Grabe, 1988, p. 61). Therefore, a reader who is weak in one particular skill area will compensate by
bringing into effect other reading processes. Grabe's (1991) interpretation of an interactive approach
is one that takes into account the critical contribution of both lower level processing skills
(identification) and higher level comprehension and reasoning skills (interpretation) The process
then, is reciprocal.
Advantages of an interactive approach
An overemphasis of either a bottom-up or a top-down approach will not realize a reader's potential
for comprehension of a text. Developing readers must work at perfecting both their bottom-up
recognition skills and their top-down interpretation skills. Thus, the reading process can be viewed
as a combination of interactive bottom-up and top-down procedures (Clapham, 1996). Readers often
decontextualize and just think about the words so that by the time they reach the end of a page they
have forgotten what the top was about (Eskey & Grabe, 1988). This was a pronounced problem with
the students in this study: Their word-by-word text approach resulted in a very slow comprehension
rate. Such a slow reading process can cause tunnel vision because the brain is overloaded with
visual information when the reader is reluctant to use non-visual information and/or when the reader
is unwilling to predict what may follow in the text (Smith, 1985).
Kitao and Kitao (1995) found that most Japanese students read by replacing all English words with
Japanese words one by one. Students ascribe equal importance to each word and use only their
syntactic knowledge to understand the sentence. They work very slowly through the text and
struggle to comprehend its overall message (Kitao, Kitao, Nozawa, & Yamamoto, 1985). The
reading experience my students bring to class reflects the above. Further, they tend to read a
passage through from beginning to end without surveying the reading beforehand or making
predictions based on the title or illustrations (Kitao, 1994). The present study focuses on this last
point.

The Study
The subjects
The subjects for this study were eight Japanese students in an Academic Study Skills class, whose
average age was 26. Their English level was lower advanced and their primary reason for attending
was to prepare for postgraduate study in the U.K. Since the focus of the class was on developing
academic reading, writing, listening, and speaking skills, only a limited time could be allocated to
reading.
Materials
The subjects were provided with a selection of reading materials drawn from Japanese (L1) and
English (L2) magazines and a number of texts from our coursebook, Campus English (Foreman,
Donoghue, Abbey, Cruden, & Kidd, 1990). I also transferred two of the texts from the coursebook to
overhead transparencies (OHTs) in order to display them separately on the whiteboard and thus
facilitate the elicitation of prediction ideas from the class as a whole.
Procedure
I began by asking students to choose an article to read from a selection of L1 magazines (Executive,
Newton, and Online Today Japan). Next, we repeated the procedure with the L2 magazines (The
Economist, New Scientist, and New Statesman). Students discussed their choices together, first the
L1 articles, then the L2 articles, focusing on what they chose and why, and how they made their
choices. I then summarized the results on the whiteboard under the headings L1 magazines and L2
magazines. The class discussion which followed revealed that the students applied similar strategies
when choosing both the L1 and L2 articles. The students thus realized that the selection strategies
which were applied in L1 had similar and useful applications in L2. The points noted were:
  they knew something about the topic already and wanted to know more
 they became curious about the article after reading the title
 they had a personal interest in the topic
 they already had some idea as to the focus and contents of the article and wanted to clarify
this and pursue it further.

The students were thus activating schemata (Rumelhart, 1980) relevant to the topics chosen.
Furthermore, the activity demonstrated that they were activating their prediction
skills unconsciously in L1 and in L2 in order to make a choice which in turn led to conscious thinking
about the articles they had chosen.
After establishing that certain skills were identical or similar in L1 and L2 prior to reading the text
itself, the next set of exercises focused on prediction from titles of texts.
Prediction from titles
I shall regard prediction in its general sense as outlined by Tadros (1994), which is guessing or
anticipating what will come next in the text based on the reader's common-sense knowledge of the
world.
I wrote the title of an article from The Economist magazine on the board, "Webbed Flight" (The
Economist, 1997), and asked the class to predict the contents of the article. Key words from their
predictions were written under the title. I then placed the article on the overhead projector (OHT) and
asked the students to scan the article for the key words to check if their predictions were correct.
Most were, but not all. Rather than predicting correctly all the time, it is more important for readers to
be actively involved in the processing of the possible contents and meaning of the text (Nuttal,
1996).
I repeated the activity twice, each time using new material, once as a group exercise, then
individually. Following this, students came up with the following advantages of prediction: (1) it
recalls what you already know about the topic and so can help prepare you better for the reading; (2)
you are thinking about the topic before you begin to read so this can help you associate what you
already know with the contents of the text; (3) it makes it easier to understand new information if you
already know something on the text; (4) it may help improve reading speed; and (5) the reader can
feel less anxious approaching a text due to the familiarity already established between reader and
text.
Prediction from within the sentence and paragraph
Next, I used an exercise, "Prediction within the sentence" (Nuttall, 1996, p. 14) which consisted of
group discussions followed by class consolidation. Students applied their syntactic (e.g., third person
singulars) and semantic knowledge (e.g., collocation) to predict what words would follow
consecutively in the sentence, thus speeding up their progression through the sentences. This was
an important awareness-raising exercise: Since the word order in English sentences is different from
Japanese, students usually replace English words with Japanese ones to make (Japanese)
sentences before trying to comprehend meaning (Kitao & Kitao, 1995).
Using an OHT projection of Nuttal's, "Predicting our way through a text" (1982, p. 13) I conducted a
class discussion, monitoring comprehension using the questions provided in the text, drawing the
students' attention to both the syntactic and semantic relationships within the text, and reiterating
that predictions were not always correct. The purpose of the above exercise was to make the
students more aware of an interrogation factor which Swales (1990) refers to as "a reciprocity of
semantic effort" (p. 62), so as to instill in the students the necessity to continually question the
direction of the author and thus place themselves in a better position to comprehend the text as it
unfolds.
Finally, I gave the students a complete text with title from their coursebook and asked them to apply
the strategies they had practiced. After this, a class discussion took place reviewing all the
exercises. To conclude, I invited feedback from the students on their impressions and thoughts of
the strategies covered. This mainly revolved around the points written on the board during the
previous exercises.
As a follow-up task , students wrote their opinions and comments on these exercises for homework,
three examples of which appear in Table 1.

Table 1: Students' retrospective comments on the exercises

(unedited)

Student #3
The method you've introduced us made me conscious my
subconscious. I tended to read a text word for word until then, being
afraid to misunderstand the contents. Now I'm trying to skip as many
words as possible even when I'm going to read about something not
familiar, and I am picking up some key words when I am going to
deal with the text I've already had quite a few knowledge.
Student #4
I have never noticed the importance of prediction in reading without
your lecture. So far, I have paid attention to the sentence structures
and the word meanings rather than the whole meaning of a story.
Therefore, it takes a long time for me to read through a whole story. I
am afraid that the English classes which I have attended in Japan
made me read like this.
Student #5
There are many positive aspects of using "prediction skill". Firstly,
we immediately thinking about the topics helps us to understand
contents of articles. Secondly, we can improve our reading speed by
predicting the following contents. Thirdly, we can associate our
knowledge we already have concerning the topics and it can help to
make our learning much more easier.

Evaluation
The students' written feedback suggests that they became more aware of the positive roles that
activating background knowledge and prediction can play in the L2 reading process: "I learned ...
that predicting is one of the most important aspects in reading" (Student #1); "I have understood and
reconfirmed that the 'prediction,' or the 'active reading' is very important and useful skill (sic)"
(Student #2).
In the "prediction within the sentence," exercise, students focused on how words can be anticipated
and chunked together rather than on the individual meaning of each word. It showed students how to
read fewer words and hypothesize more.
The extended "prediction within the paragraph" exercise helped students become more aware of
how a writer's ideas can progress in the paragraph. One student commented, "So far, I have paid
attention to the sentence structures and the word meanings rather than the whole meaning of a
story."
It is difficult to judge whether the students really began to utilize the skills just introduced during the
final reading exercise. However, because they were more willing to discuss the possible overall
meaning of the text than on previous occasions, because the exercise took less time than previous
exercises of this kind, and because the students refrained from using dictionaries during the reading
of the text, it is reasonable to assume that they were beginning to utilise these new skills. The above
points, supported by the students' retrospective comments on the exercises, illustrated their
increased confidence in approaching and reading a text with a more balanced reading approach.

Conclusion
Students have different reading abilities, possess different background knowledge, and have
different linguistic competence. The focal point of the study was highlighting the positive benefits in
the activation and use of the two skills regardless of the different elements present in each student's
reading and linguistic background.
No substantial changes can be expected after one or two lessons, but it is important to set such
reading strategies in motion and to give students ample exposure to them. Reading skills develop
gradually and the reader does not become fluent suddenly. Instead, fluent reading is the product of
long term effort and gradual improvement (Grabe, 1991).
Teachers need to continually adapt their teaching methodology to their teaching environment,
regardless of what is currently fashionable in ELT. By taking into account the learners' background
learning experiences we can adapt our teaching to allow for the maximum benefits to our students.

References
Carrell, P. L. (1988a). Introduction: Interactive approaches to second language reading. In P. L.
Carrel, J. Devine, & E. Eskey (Eds.), Interactive approaches to second language reading (pp.1-7).
New York: Cambridge University Press.
Carrell, P. L. (1988b). Some causes of text-boundedness and schema interference in ESL reading.
In P. L. Carrel, J. Devine, & E. Eskey (Eds.), Interactive approaches to second language
reading (pp. 103-113). New York: Cambridge University Press.
Clapham, C. (1996). The development of IELTS: A study of the effect of background knowledge on
reading comprehension. Cambridge: Cambridge University Press.
Eskey, D. E. (1988). Holding in the bottom: An interactive approach to the language problems of
second language readers. In P. L. Carrel, J. Devine, & E. Eskey (Eds.), Interactive approaches to
second language reading (pp. 93-100). New York: Cambridge University Press.
Eskey, D. E., & Grabe, W. (1988). Interactive models for second language reading: perspectives on
instruction. Interactive approaches to second language reading (pp. 223-238). New York: Cambridge
University Press.
Foreman, D., Donoghue, F., Abbey, S., Cruden, B. & Kidd, I. (1990). Campus English. London:
Macmillan.
Grabe, W. (1988). Reassessing the term "interactive". In P. L. Carrel, J. Devine, & E. Eskey
(Eds.), Interactive approaches to second language reading (pp. 56-70). New York: Cambridge
University Press.
Grabe, W. (1991).Current developments in second language reading research. TESOL Quarterly,
25(3), 375-406.
Goodman, K. S. (1971).Psycholinguistic universals in the reading process. In P. Pimsleur & T. Quinn
(Eds.), The psychology of second language reading (pp. 135-142). Cambridge: Cambridge
University Press.
Kitao, S. K. (1994). Reading strategies in the teaching of second foreign language
reading. Doshisha Women's College Annual Reports of Studies, 45, pp. 118-142.
Kitao, S. K., Kitao, K., Nozawa, K., & Yamamoto, M. (1985). Teaching English in Japan. In K. Kitao,
K. Nozawa, Y. Oda, T. Robb, M. Sugimori, & M. Yamamoto, TEFL in Japan-JALT 10th anniversary
collected papers (pp. 127-138). Japan Association of Language Teachers: Kyoto English Centre.
Kitao, K. & Kitao, S. K. (1995). Difficulties Japanese have in reading English. In K. Kitao & S. K.
Kitao (Eds.), English teaching: Theory, research and practice (pp. 147-167). Tokyo: Eichosha.
Nuttall, C. (1982).Teaching reading skills in a foreign language. London: Heinemann Educational
Books.
Nuttall, C. (1996).Teaching reading skills in a foreign language. Oxford: Heinemann English
Language Teaching.
Rumelhart, D. E. (1980). Schemata: The building blocks of cognition. In R. J. Spiro , B. C. Bruce, &
W. F. Brewer (Eds.), Theoretical issues in reading comprehension. (pp. 33-58). Hillsdale, NJ:
Erlbaum.
Smith, F. (1971).Understanding reading: a psycholinguistic analysis of reading and learning to read.
New York: Holt, Rinehart and Winston
Smith, F. (1985).Reading (2nd ed). Cambridge: Cambridge University Press.
Stanovich K. E. (1980).Toward an interactive-compensatory model of individual differences in the
development of reading fluency. Reading