Using the computer to teach

reading comprehension skills

Roland R. Nyns

This article claims that despite the shortcomings of current software for
Computer Assisted Language Learning (CALL), the computer is an
adequate medium of teaching (some) reading comprehension skills. The
author stresses using the computer as a tool at the teacher’s disposal, as
opposed to using the computer as a surrogate teacher. Three examples of
exercises which are part of a semi-structured CALL system are given,
following which there is a short description of how users have access to a
database system which is designed to correspond to a student population’s
specific vocabulary requirements, while keeping the teacher informed of
these.

Introduction I shall begin by assessing the limitations of Computer Assisted Language

Learning and then move on to consider some generally accepted strategies
in teaching reading skills. I hope to convince the reader that the computer is
an appropriate medium to teach these skills, and will show three examples
of exercises which are part of a semi-structured CALL course I am
currently developing for use in self-access mode.

The limitations of In a recent book with the promising title The Cognitive Computer: On Lan-
CALL guage, Learning and Artificial Intelligence,
Roger Schank, America’s leading
figure in the field of artificial intelligence (AI),1 repeats once again the
hopes that have been voiced by part of the teaching community as a
consequence of the computer revolution:
The computer can enable us to carry out a revolution in education. . . .
How?
1 Computers are fun.
2 Computers can be programmed to teach far more thoroughly and
interactively than textbooks.
3 They can be individual - a child can have his [sic] own computer
teacher who keeps track of progress.
4 Computers can be used by almost anyone, no matter how hyperactive
or lazy.
5 Computers don’t get bored or frustrated with students or with teach-
ing. They won’t punish the student or single him out for contempt.
Computers can serve as excellent individualized teachers. (Schank
1984:204)
Now, setting aside the fact that the new computer revolution which AI
represents is still marching on rather experimental legs and has a lot of
ground to cover before robot teachers walk into the classroom, let us

Computers and reading comprehension skills 253

 consider the points made by Schank in the light of existing (unintelligent)
software.
Games were not only among the first programs to appear on the micro-
computer software market, they were also among the first productions of AI
researchers interested in problem solving. That was because the rules and
elements of games constitute finite sets and thus proved relatively easy to
programme. Games are fun, whether played on computers or otherwise.
Because they are the most popular type of software, many of us identify
games with microcomputers: but this does not guarantee that every kind of
interaction with computers is fun.
In fact, much of the current teaching software is a source of frustration for
the student-user in that communication with the machine is often awkward
(Nievergelt et al. 1986). Jenny Thomas points out that, ‘In my experience, it
is much more likely to be the semi-(computer)illiterate teachers who are
enthralled by CALL software than the average blasé teenager!’ (Thomas
1986:117).
Computers can be programmed to teach more thoroughly and interac-
tively than textbooks, but this statement implies that, whereas books are
regarded as one medium at the teacher’s disposal along with some others,
the computer is viewed as a surrogate teacher. Indeed, computers could do
things no other media can, such as keeping track of students’ progress,
choosing grades of difficulty and providing immediate feedback to the
students’ responses to exercises. To this extent, it could take over a part of
the teaching process which has so far been reserved for human teachers.
But as it stands, Schank’s (1984) statement simply does not reflect the
prevailing state of affairs in CALL.
Most CALL still consists of rather tedious drill activities, cloze type
exercises, or multiple-choice questionnaires. These activities stand in
strong contrast to the generally accepted concept of communicative teach-
ing, which sheds light on the limits and perils of repetitive drills in language
learning and teaching.
Those programs which do allow for creativity and initiative on the
student’s behalf all have a very restricted domain of discourse, meaning
that they model part of a complex system which has been arbitrarily
reduced to a finite set of objects, as for example in Higgins’ (1985) ‘Gram-
marland’ program.
Computers do not get annoyed, frustrated, or depressed. They don’t
punish, insult, or judge students. Consequently, timid students who are
often afraid of making a fool of themselves in the classroom or of being
judged by their teacher are expected to overcome their self-consciousness
when interacting with the computer.
Self-conscious students may feel more at ease in front of a computer
monitor than in the classroom, but many people do not enjoy interaction
with computers. They lack the typing skills necessary to work at a reason-
able pace. They are apprehensive about what the machine does with their
input (keeping track of their mistakes!). They miss the warmth of human
interaction. In fact, though this may be only transitory, many students may
prefer human teachers who get bored, red-herringed, and frustrated to a
machine which never makes a mistake and which sticks adamantly to the
curriculum.
Finally, the computer’s greatest drawback for FLT applications is that it
does not have a voice (or has an awful one), and that it cannot cope with
oral input; much less with correct pronunciation.

254 Roland Nyns

 In sum, from the moment computers became financially accessible to the
ordinary mortal, a whole range of enthusiastic teachers, computer
scientists, and marketing managers have suggested that CALL might solve
many of the current problems of the educational system. However, mainly
because of software limitations resulting from an incorrect appreciation of
the computer’s power and role in the complex process of teaching and
learning, the hopes pinned on CALL have turned out to be unjustified.
If we regard the computer not as a magical box, but as a medium as
useful and as limited as other media at our disposal, a different approach to
its use in teaching emerges. Our starting point will not be the question
‘What can I do with my computer?’ (Answer: ‘Cloze exercises, drills,
simple multiple-choice questions’), but ‘Which medium should I use to
teach such and such a skill?’ The answer to this latter question might be the
blackboard, the video, printed matter or the tape recorder or whatever. It is
rarely ‘the computer’. Thomas (1986) rightly observes that:
The only justification for the wholesale introduction of computers into
the language-teaching classroom is if they can be shown to do something
which research into second-language acquisition suggests is ped-
agogically well-motivated and that they can do something which either is
not being done at present, or can do it more cheaply or efficiently than is
being done at present. (Thomas 1986:129)
I claim that it is justifiable to use the computer as a teaching aid within
the framework of teaching reading comprehension, and I hope that the
material and software that I have developed fulfil such conditions.

Teaching reading At my university my colleagues and I teach French-speaking students in

comprehension skills Psychology and Social Sciences to read English for professional purposes.
These students are not taught to write, speak, or comprehend oral dis-
course. In the following section, I shall briefly outline some of the sub-skills
which we attempt to develop with these students, bearing in mind that ‘a
learner will not become a proficient reader simply by attending a reading
course or working through a reading textbook’ (Williams 1986:44).

Speed and flexibility Reading speed is a major problem in this type of teaching. Among a given
class of students, reading speed may vary enormously, resulting in boredom
for faster readers waiting for classroom activity to continue, and frustration
for the slowest, who are never allowed to read a text to the end.
Furthermore, we believe that one should attempt to increase the reading
speed of all our students. This requires the apparently contradictory action
of controlling reading time and doing so for each student individually. It is
clear, that for this purpose at least, the computer is the only medium which
offers such control.
In the context of reading, flexibility expresses the need for different
modes of reading, adapted to various types of text and goals in reading
them: for example, one reads an abstract in an academic journal one way,
and an article in Reader’s Digest in another.
This is to some extent related to reading speed control, as skim-reading of
certain texts can be ‘forced’. Highlighting a text and focussing on logical
connectors will also develop this sub-skill. This can be achieved using
printed matter and, more vividly, overhead projection, but the computer
allows the combination of visual display with time control, and can do this
individually for each student.

Computers and reading comprehension skills 255

 Vocabulary and coping Although unknown words in texts may cause no hesitation to the Ll reader,
with unknown words they can often completely block comprehension and hold up the reading
process for the reader of a second or foreign language. Thus, it is necessary
to practise some simple strategies which are probably used unconsciously
in the mother tongue, such as inferring the meaning of new words, deciding
on the importance of certain words for the global comprehension of a text,
and dealing with compound and affixed words. The conscious practice of
these strategies requires a step-by-step presentation of the material, which
only the computer can provide for individual study.
No reading strategy or inference technique can do without a basic
knowledge of vocabulary. In teaching to read for professional purposes, it is
generally assumed that students master the 2,000 words (more or less) that
make up about 80 per cent of an average text in the subject field, plus a few
hundred more. Since specialist vocabulary is international to a large extent,
the vocabulary which such students must acquire consists of sub-technical
expressions common to most academic texts.
The huge data-storage capacity of computers can be quite useful here in
creating a dynamic dictionary which can inform students of the importance
of the words that they look up, while keeping teachers informed of the
vocabulary needs of a given student population. It also affords the possi-
bility of restraining too frequent consultation or access to certain categories
of words (affixed words, for example).

The role of syntax and Nominalizations, relative clauses, and complex noun groups are some of
text structure the grammatical constructions which may be a source of comprehension
difficulty at this level. The solution to such problems is often simply to
analyse and rewrite the sentence in which they occurred (Nuttall 1982).
Once more, visual presentation, highlighting, and restructuring make the
overhead projector the ideal mode of presentation in the classroom, but for
self-study and additional practice the computer display may appropriately
replace that medium.
But of course the mere comprehension of vocabulary items and isolated
sentences does not ensure the comprehension of a text. It is also necessary
to grasp the complex relations between sentences which make up a text.
Hasan and Halliday (1976) have shown how the correct interpretation of
pronouns, articles, and conjunctions is paramount to the understanding of
discourse. Although this is rarely a source of difficulty for Ll readers, it
appears that L2 readers often fail to interpret reference items or logical
connectives correctly. Hence the need for specific practice.
The conscious use of world knowledge, including knowledge about the
structure of the type of text that is being read, can considerably enhance
comprehension if it is used consciously in reading predictively. However, in
order to develop predictive reading, some control and guided practice has
proved necessary. Such controlled practice can advantageously be supplied
by the computer (Higgins 1984).
In other words, therefore, in the teaching of reading for professional
purposes, many of the limitations of CALL are irrelevant, whereas its
strengths (in being individualized and interactive) remain. This makes the
computer an ideal medium for complementary practice in learning some of
the sub-skills required in reading comprehension. For these reasons, my
colleagues and I have begun developing a semi-structured reading course
for self-access use as a complement to classroom activities and extensive
personal reading. The course is semi-structured, in the sense that it consists

256 Roland Nyns

 partly of a flexible tutorial and partly of the use of the computer as a tool, i.e.
as a language database for the student. In the present article we shall put
the emphasis on the structured part, and present three types of exercises
which the program can handle.

Examples At present, the program deals adequately with single correct answer
responses, with correct answers requiring more than one item (or character
string), with alternative possible correct answers, and with expected key-
words or phrases in open-ended answers. It also allows for changes of
answer types from one question to the next within exercises.
Since technicalities will not be considered in the present article, the
examples below have been chosen to illustrate how various sub-skills can be
effectively practised, using exercise formats possible with the software I
have developed. The first example is an exercise meant to demonstrate and
develop inferencing skills necessary for dealing with unknown lexical items.
The second concerns correct recognition of anaphoric reference. The last
encourages predictive reading.

Example 1 The following inference-training exercise sequence accepts only one correct
answer at first, but four alternative items later on. The text is displayed in
segments after an introduction explaining the pedagogical goal of the
exercise. The item which students are asked to infer is ‘squid’.
The Giant Squid
This deep-sea cephalopod is so seldom seen, dead or alive, that it
seems mythical. Numerous specimens have, however, been studied, and
they have begun to reveal the animal’s anatomy and its ecology. (limit
second display)
In Moby Dick Herman Melville describes a sea-creature of vast pulpy
mass, furlongs in length . . ., long arms radiating from its center and
curling and twisting like a nest of anacondas. He apparently had in mind
the giant squid; his description reflects the meager information that was
available in his time. (limit third display)
Even now the giant squid (weighing up to 1,000 pounds and having an
overall length of some 18 meters if the tentacles are extended) remains
largely mysterious.
The first display contains only the title of the article and the information
that it is an extract from Scientific American of April 1982. Asked the meaning
of ‘squid’, the student has little to go on at this stage, so that a correct
answer could only be the result of actually knowing the word and not of
inferring its meaning. Consequently, the exact French equivalent of ‘squid’
is expected as a correct answer. If it is given, the rest of the exercise is not
very useful to the student. In that case, the student is given the whole
extract to read and then he or she goes on to a new exercise.
Those students who do not answer correctly are shown the title and the
short summary of the article (down to ‘ecology’). At this stage, they might
make a correct inference, but not distinguish between squids and
octopuses. Since either is a correct inference (i.e. it makes perfect sense), I
see no reason to refuse the French equivalent for octopus (the students are
not Biology students). Thus, the matching procedure must allow for four
correct answers at this stage, whereas it did not for the previous question
since within that context only direct translation could provide the correct
answer.

Computers and reading comprehension skills 257

 Students who make wrong guesses are first given some clues as to the
correct answer and then given another chance before the next part of the
text is displayed. At stage one, they would have been told that inferencing
without context is nearly impossible, but that the origin of the text and the
word ‘giant’ might be some help. After the second display, their attention is
focused on the words ‘animal’, ‘mythical’, and ‘deep-sea’. Next, the text is
displayed down to ‘his time’, mentioning a ‘pulpy mass’ and ‘arms radiat-
ing from its centre’, which hint very obviously at squids and octopuses. No
information rules out the latter, so that either is an acceptable answer.
Finally, the whole extract is displayed, confirming any correct hypotheses
the student may have constructed.

Example 2 One frequent source of misunderstanding for my students has been wrong
interpretations of anaphorical reference. I try to deal with this by choosing
texts such as the one below which contain some difficulties in this respect. I
present them in segments, and ask for the reference item corresponding to
the anaphor. In normal classes of twenty students this procedure is incon-
venient, in that some students don’t have time to think things out, and some
don’t ask for explanations. Also, in such classes, this type of exercise cannot
be practised as often as some students would need.
But the implementation of this type of exercise on computer is quite
simple. A question is asked after each segment of the text. After a first
incorrect answer, a hint at the correct solution is provided, and an explana-
tion is supplied after a second incorrect attempt. Furthermore, throughout
the exercise students have access to a dictionary.

Blame the medium, not the message

There are two ways to fight a mosquito-borne disease. One is to
develop a vaccine or drug; the other is to try to break the life cycle of the
disease by dealing with the mosquito. Most of the money spent on
research into malaria, yellow fever and the like goes on the medical
approach. Yet the other way seems more promising.
(Question: What exactly is this ‘other way’?)
It was, after all, the
draining of mosquito-infested swamps, not drugs, that rid Europe of
malaria.
The protozoan parasites that cause malaria are adept at evolving
resistance to each drug that is thrown at them; most vaccines would be
just as easily evaded.
(Question: ‘Just as easily’ as what?)

Example 3 The next kind of exercise focuses on the student’s ability to predict and
anticipate the content of a text. The student is first informed of the aim of
the activity and of the difficulty of processing open-ended answers. Then
some text is displayed with a corresponding question. If the question is
answered correctly, the student is given more text to read and asked to
check whether his or her predictions matched it. The student may then
proceed to read the next part of the extract. If the answer is not rated as
correct, i.e., if it does not hold one of the expected keywords, the student is
given a clue to the correct answer and allowed to try again. On a second
incorrect attempt, an explanation is provided before the complete extract is
displayed.

258
 The Structure and Function of Fish Schools
How do they do it? The question occurs naturally to anyone watching
a school of fish moving slowly over a reef in clear tropical water. Hun-
dreds of small silver fish glide in unison, more like a single organism than
a collection of individuals. The school idles along on a straight course,
then wheels suddenly; not a single fish is lost from the group.
Although schooling of fish is one of the most familiar forms of animal
social behaviour,
(First question: what do you expect the content/function of the next clause to be?)
until recently it was little understood, partly because of
the difficulty of observing minute changes of position and velocity in a
school under natural conditions.
It had been thought that each fish maintains its position in the school
chiefly by means of vision.
(Second question: What will the informational content of the next sentence be?)
Our work has shown that the lateral line, an
organ sensitive to transitory changes in water displacement, is as impor-
tant as vision.
The most appropriate format for this type of exercise seemed to be open
questions. I have chosen to encourage answers in the mother tongue, as this
allows more creativity on the students’ behalf and reduces spelling and
syntax errors which will make answer processing even more hazardous
than it already is. Open-ended answers are evaluated on the basis of
keywords taken to indicate a correct answer. Thus, in the answer to the
second question, the program must acknowledge as correct any sentences
containing the words inconnu, peu connu, ma1 compris.
This poses some problems, as there may well appear to be alternative
correct answers which had not been anticipated, or incorrect answers
which contain the expected keywords. There are three solutions to this
limitation of the software, the first of them being simply not to formulate
correction messages in too definitive a style. The second is to use the book-
keeping facilities of the program to update the program with unexpected
correct answers that students may produce. Thirdly, by imposing a certain
form on the answer (for example, asking students to formulate their
answers as questions), it is possible to control syntactical complexity so that
a simple parsing algorithm can analyse the answers in more detail.

The dynamic Throughout the session, the student is free to explore the sequence of
dictionary questions which make up the tutorial, using a set of five motion commands.
One of these gives access to what we call the dynamic dictionary.
At the moment, the dictionary routine (see Figure 1) handles four
possibilities:
1 The required word is in the database, and its meaning is communicated
to the student. In this case a ‘consultation index’ is increased by one point.
2 The word is not in the database as an entry, but it is an affixed word. In
this case the student is encouraged to decompose the word and look up the
meaning of the root only.

3 The word is not in the database because it is mis-spelt. The student is

asked to check this.

4 The word is spelt correctly according to the student, but has simply not

Computers and reading comprehension skills 259

 been included in the database. In this case the word is entered on a blank
record and left for the teacher to provide a definition later on.

Note that the decomposition and inserting of new items serves two
purposes. It keeps the database within manageable limits, in that rare
words, as well as generally known words, do not crowd it. And students are
discouraged from consulting the dictionary indiscriminately, as it will
eventually contain all those items necessary for the fulfilment of the reading
tasks set and no others, and as it will not allow students to look up affixed
words.
It is hoped that the dictionary will prove a powerful tool, both for
students and for teachers. With the dictionary teachers will be able to assess
the needs of a specific student population who are reading specific texts.
These needs are not as readily met by more conventional resources such as
specialized dictionaries and word-frequency lists.

I 1 .
I

Figure I: The dictionary routine

260 Roland Nyns

 Conclusion The program presented here is still under development. It has many
imperfections which observation and better programming may eventually
correct. The main reason for presenting it here is to seek reactions and
comment which will help to make it an effective aid in the teaching of
reading for professional purposes. However, I believe that even in its
present state, the program shows that microcomputers can be used as a
teaching medium without disregard to pedagogical principles or good
programming practice.
One of the interesting points in developing this software was that instead
of being superseded by programming imperatives, pedagogical questions
and detailed analysis of the material became fundamental. In fact, it was
while I was adapting existing classroom activities for use in the CALL
package that I discovered weaknesses in them. It is this extra care, neces-
sary in programming, which accounts for the reputed high cost in prep-
aration time of CALL lessons. In that respect, CALL is similar to editing
any other type of teaching material. Once developed and corrected, it may
be used over and over again.
I would also like to have shown that developing CALL courseware for
microcomputers can make a modest contribution to a better awareness of
the issues at hand. More importantly, such grassroots-level research will
provide teachers with some of the expertise they will require to retain
control over the more sophisticated CALL systems of the future. As Martin
Phillips writes, the alternative is ‘the inexorable advance of what is called
technological progress will remove the power of decision making from our
grasp’ (Phillips 1986: 10). •
Received August 1987

Note Phillips, M. 1986. ‘CALL in its educational context’

1 Artificial intelligence (AI) is the study of how to in Leech and Candlin (eds). 1986.
make computers perform tasks which are generally Schank, R. 1984. The Cognitive Computer: On Language,
considered proper to human intelligence, such as Learning and Artificial Intelligence. Reading, Mass:
problem-solving, or understanding and producing Addison-Wesley.
natural language utterances. Thomas, J. 1986. ‘Adapting dBase II: the use of
database management systems in English language
References teaching and research’ in Leech and Candlin (eds)
Hasan, R. and M. Halliday. 1976. Cohesion in English. 1986.
London: Longman. Williams, R. 1986. ‘ “Top ten” principles for teaching
Higgins, J. 1984. ‘Reading and risk-taking: a role for reading.’ ELT Journal 40/1:42-5.
the computer.’ ELT Journal 38/3: 192-8.
Higgins, J. 1985. ‘Grammarland: A non-directive use
of the computer in language learning.’ ELT Journal
39/3:167-73. The author
Leech, G. and C. Candlin (eds). 1986. Computers in Roland Nyns has an MA in Romance and Germanic
English Language Teaching and Research. London philology from the University of Freiburg (FRG) and
Longman. has been teaching EFL to adults since 1980. Since 1984
Nievergelt, J., A. Ventura and H. Hinterberger. he has been research assistant at the Free University of
1986. Interactive Computer Programming for Education. Brussels (ULB), where he has taught EFL and ESP,
Reading, Mass: Addison-Wesley. while working on a research project on reading com-
Nuttall, C. 1982. Teaching Reading Skills in a Foreign prehension. He is currently writing a PhD on text
Language. London: Heinemann. processing and production.

Computers and reading comprehension skills 261