Published as "Tanulási stilus, ellenörzés, kurzustervezés [Learning Styles, Locus of control and Courseware Design]", in Kecskés, István & Papp, Tünde, eds., A felsooktatás fejlesztését szolgáló kutatások, Budapest, Felsöoktatási Koordinációs Iroda, 1994, pp. 83-94.

Abstract

It's been known for a long time that people have different ways of
collecting, processing and organizing information, different ways of
learning.  Some people seem very structured and organized, others are
all over the place.  Some people like theory, others concrete examples;
some gobble up details, others prefer a broader picture.  Over the last
two decades, an enormous amount of work has been done to study these
differences with the aim of taking them into account when devising
instructional strategies.  Much of this research uses the concept of
"learning styles".

Learning styles research makes three assumptions.  The first is that
different people learn in different ways.  Few would dispute this. The
second is that there is a useful theoretical construct which we can
call a learning style, which is each person's unique way of interacting
with the environment cognitively, affectively and phsysiologically.
We all learn differently, so the argument goes, because we have
different combinations of cognitive, affective and physiological
reactions.  The third assumption is that it is possible to measure the
individual's learning style by using an appropriate psychometric
instrument.  Once the learning style is measured, it is hoped, one can
use this knowledge to develop more appropriate instructional
strategies.  As will probably be clear, the concept of a learning style
is not unrelated to the concept of a cognitive style or of a
psychological type (in the Jungian sense); learning styles research has
also found support in, and indeed is sometimes explicitly tied to the
theory of multiple intelligences developed by Howard Gardner (1).

Learning styles research is extraordinarily popular in North America,
with some 9000 items in the ERIC database for the last ten years alone.
This popularity has probably a great deal to do with the fact that the
dominant educational ideology in North America is anti-elitist, aiming
to provide equal levels of education for all citizens.  Learning styles
theory allows one to be positive and optimistic about the possibilities
of improving our educational systems, offering the hope that
underachievers can be helped through the development and use of new
teaching strategies which take their individual learning styles into
account.

As a result of this popularity, a great number of instruments --
learning style inventories or delineators -- have been developed to
measure our individual styles.  Some of the most commonly used are the
Myers-Briggs Type Indicator, the Gregorc Delineator, the Kolb Learning
Style Inventory, the Canfield learning styles inventory and the Dunn
instrument (2).

Although I will be using one of these delineators and the concept of
learning styles, I wouldn't like to give the impression that the whole
area is unproblematical.  Quite the contrary.  Although it seems
obvious that we all learn differently, both the theoretical construct
of a learning style and the contention that it can be measured in a few
minutes using one of the instruments which are presently available are
much more controversial, though this fact is rarely mentioned in the
learning styles literature, little of which is theoretical.
Psychologists in particular have expressed serious doubts about the
psychometric validity and reliability of most of these instruments
(Sewall 1986), and the whole concept of learning styles is far from
being universally accepted.

A more practical problem with learning styles research is that no two
instruments or conceptualizations use the same categories, so
generalizations about research conclusions are difficult to make.  The
Myers-Briggs which is no doubt the best known instrument (though it
does not purport to measure learning styles as such) uses 16 categories
based on four polar scales (Extraversion-Introversion, Sensation-
Intuition, Thinking-Feeling and Judgement-Perception); the Kolb has
four learning modes (Concrete Experience, Reflective Observation,
Abstract Conceptualization and Active Experimentation) and four
learning style types (Assimilator, Accommodator, Converger, Diverger);
The Gregorc, which is the one I used, has four categories based on two
polar scales (Concrete vs Abstract and Sequential vs Random).

Although there are great divergencies among the different
classifications, there are two learner stereotypes which most of them
oppose (along the lines of analytic/relational, field independent/field
sensitive, reflective/impulsive, serialist/holist, sequential/random).
The first is the "academic type" who learns best quietly and without
any extrinsic motivation when information is presented in a structured,
sequential manner.  The second type is often less successful in our
educational systems: he may appear to some of us to be erratic,
emotional, and troublesome, and he often learns best when personally
motivated, in stimulus-rich (noisy, busy) environments and when
information is available to him in discovery oriented, non pre-
structured ways.  Of course, these thumbnail sketches are
simplifications and represent extreme cases.  Most people possess a
combination of styles, one or two of which are usually dominant.
However, it has been found that our educational systems and teaching
strategies tend to favor the "academic" student and are much less
successful with students whose dominant learning mode is random or
holistic, even though these latter are by no means necessarily less
intelligent or less successful in life than their academic
counterparts.  A 1980 study found that three-quarters of the population
do not learn by their preferred methods in the formal schooling process
(Davidson 1992: 349).

The challenge for those of us who would like to provide for all members
of society equally is then to adapt the educational system to the
students in such a way that not only the 25% of us who are "academic
types", but the other learners, the majority of learners, are reached
by our instructional strategies, stimulated and helped by the system
rather than being put down by it.  And we would like to do this without
disadvantaging those who are served well by the present system.
Individualized instruction, in which not just one but a variety of
different learning styles are addressed seriatim or simultaneously, is
often seen as one way to achieve these aims.

 Computer Assisted Instruction

Now the capacity to provide individualized instruction is often touted
as one of the big advantages of CAI, along with benefits which are not
always convincingly proven, such as more effective and cheaper
instruction and some motivational improvement.  In addition to any more
general advantages it might offer the educational system, CAI appears
then to offer hope to non academic learners.  Whereas in a traditional
classroom it is difficult to offer different sorts of instruction for
different sorts of learners, the computer can in theory do so easily.
Many studies have found that the ability to individualize instruction
was the biggest advantage of using a microcomputer. (Chamberlin
1988:14)

However, individualization can mean many things.  We know it is
possible to set up CAI that will appeal to non-academic learners, and
even to people with various disabilities.   LOGO is one example of a
program which does this, and Papert's  Mindstorms  and Sylvia Weir's book
 Cultivating Minds  are fascinating casebooks of students with non-
academic learning styles who prosper in a LOGO environment.  Some work
has been done too on how it is that certain types of computer projects
-- not only LOGO type activities but also programming and games and some
"emancipatory" activities involving for example information retrieval
systems (O'Shea 1983: 113) -- self-motivate people and keep them
working and concentrating for long periods (Malone 1981) with the
possibility for effective learning during this time.   These sorts of
activities do appeal to non-academic learners.

Individualized learning here is clearly not a question of an instructor
adapting material and strategies to different students, according to
their learning style, for example.  It is almost entirely a question of
the student taking over the learning process; it is a question of
learner choice and learner control.  Weir says: "When LOGO is used by
teachers whose teaching practice is compatible with LOGO philosphy, the
individual student is encouraged to choose what to do and how to do it.
Under these circumstances, a student is able spontaneously to adopt her
preferred problem-solving style more readily than is the case under
didactic teaching conditions." (Weir 1987: 115)  The student works on a
project, either chosen by him or proposed by the instructor, but in any
case adopted by him as a challenge; and he develops his own strategies
to achieve his goal.  There are many testimonies to the success of this
particular sort of individualized instruction.

Now, most CALL I know about in secondary and post-secondary
institutions is not of this discovery or inquiry oriented type.  It is
more frequently of the tutorial or drill-and-practice variety, using
either programs provided by publishers to accompany their second
language textbooks or programs produced on authoring systems such as
CALIS or CALLGEN, or Hypercard or Toolbook.  When we talk about
individualization in programs such as these, we also mean learner
choice and learner control, but not in the same sense at all as in the
case of LOGO or of games or programming.  Learner control here does not
involve a choice of strategies or being driven onwards by step by step
progress towards a clearly identifiable goal.  In tutorial and drill-
and-practice CALL you can choose when to stop, skip and start again,
you can go fast or slow, you can choose to ask for the right answer and
sometimes for extra help, but whatever you choose, you still have the
same ground to cover, you have a very limited choice (often none) of
preprogrammed strategies, and instead of feeling that you are
controlling your own progress towards a clearly defined goal (make a
square, or get out of the maze), you often get the feeling that the
computer is almost arbitrarily asking you a whole series of questions
in no particular order (many programs randomize).  This particular type
of individualization, which might be called psq (pick, skip, quit), is
often unjustifiably assimilated to the sort of individualization that
programs like LOGO offer because in both cases, locus of control
appears to be with the learner.  However, while the particular sort of
individualization LOGO can provide has been found almost uniformly to
be advantageous, if difficult to assess, studies of individualization
in tutorial and drill-and-practice programs have with almost the same
degree of uniformity found it not to be advantageous.  It seems that
students left alone with these programs do not do as well as one might
expect.

A few findings:

     The TICCIT project was one of two (the other was called PLATO)
     funded by the National Science Foundation in the seventies to test
     the effectiveness of CAI.  One of the features of TICCIT was a
     special keyboard which apart from the standard keys had a dozen so
     called Learner Control Buttons.  At any time, students could ask
     for a  map  of the lesson (for those who like to know where they're
     going), for  objectives  (for those who like to know why), for rules
     (for those who like theory), for  examples  (for those who work
     inductively), for  practice  (for those who learn by doing), etc.
     The project designers hypothesized quite reasonably that if you
     gave students control not only over their rate of progression but
     also over the sorts of information they received, they would learn
     better and be more motivated.  However, the final evaluation of
     the project noted that students had not taken advantage of these
     buttons, except for the practice button.  In addition, the
     completion rate dropped from 50% in regular classes to 17% in the
     CAL courses.  (The experiment took place in two community
     colleges.)  The results of the project show very clearly that this
     particular sort of learner control does not increase motivation or
     lead to better learning.  (O'Shea 1983: 92)

Most of the other studies I have looked at were concerned with self-
pacing and with feedback, which are two features of tutorial and drill-
and-practice programs most often left to learner control.

     Belland, Taylor, Canelos, Dwyer and Baker (1985) found "that
     learners [college freshmen] who selected elaborative feedback
     (which provided additional instruction) were the ones who best
     understood the concepts presented in the lesson.  In spite of
     their apparent need for additional instruction, the students who
     were performing poorly on the lesson did not make effective use of
     the option for elaborative feedback." (Sales 1988: 98) In fact,
     low achievers chose to see feedback only 29.72% of the time, as
     against 86.56% for high achievers.
 
     Humphrey, in his 1983 thesis on "A Comparison of How Paced and
     Unpaced Problems Affect Learning During CAI Math Drills" found
     that self-paced programs were less effective than those in which
     some guidance of time was used. (Vargas 1986: 740)
 
     Carrier, Davidson, Williams and Kalweit found that students
     selected less feedback later in a lesson, and that some never
     selected any.  (Chamberlin 1988: 26)
 
     In a 1989 study in CALICO Journal, Chapelle and Mizuno found that
     students of ESL used available help (3 options, meaning, grammar
     or facts) only every 8.6 minutes in sessions lasting about twice
     that time.

     In their 1990 study, Hasselerharm & Leemkuil found that "LC
     [=learner control] was not an effective strategy for low-achievers
     with regard to the transfer of learning" (78).  Summarizing
     several studies which they had consulted (different from the ones
     quoted above), they say "An overall conclusion is that many
     students, especially low achievers, lack the knowledge to make
     appropriate decisions." (69)

What these studies found then, was essentially that it is not a good
thing to leave students in control of their own learning at the
computer in tutorial and drill-and-practice type programs.  The only
students who sometimes benefit from such an approach are the high-
achievers.  Low-achievers, and these often coincide with non-academic
learning style profiles, need guidance and support.  However, despite
the consistency of these findings, students continue to demand learner
control and publishers and to some extent CAI course designers still
continue to tout it.  In Hasselerharm & Leemkuil's study quoted above,
one group of students had considerable control over their progress on
the computer while another group had very little.  The learner control
students, especially the low achievers, did not learn as well as the
others, but they were much more positive about the control strategy
than their more controlled colleagues.  And a recent book by J. Steven
Soulier on courseware design has a section on "Individualized Computer
Based Instruction" in which learner control is considered automatically
to be best: "Keep the learner in control" (94), "The concept of self-
pacing should be applied" (91).

Why, despite the research, do we still vaunt learner control here?
Probably because learner control means individual choice and individual
choice is ideologically unassailable, just like freedom of expression,
free enterprise and saving the planet.  Our culture has an
individualist ethos, in which making your own decisions is highly
valued -- I know what I want, says the voter.  In addition, we have a
love-hate relationship with technology -- see Hal in Kubrick's  2001  --
in which the human has to be in control over the machine.

So, if students want to be in control, but most of them really learn
better when they are not, what does this mean for program design?  Is
it possible and advisable to design programs which guide and control
learning while leaving enough choices to the learner to give him a
comfortable feeling of being in the driver's seat?

This turned out to be the main focus of the study I carried out.
After being convinced by previous research that learner control was
not very successful, especially for non-academic learners, I wanted to
find out how much real resistance to guidance and outside control there
was on the part of the students.  Despite their ideological preference
to be in control, did the students mind if the program often took
control away from them and forced them to do things?  In addition, did
their preferences for control correlate with their learning styles?

I should at this point briefly describe the experiment.  The students
involved were enrolled in a third year French course at the University
of Calgary.  A component of the course which accounts for about 10% of
the final grade is grammatical analysis, or parsing, and this was
taught exclusively by computer, using two different programs with
almost identical content, ANALYSE and CALLGEN.  (I was responsible for
the content of both programs.)  ANALYSE allows maximum learner control:
once a topic has been chosen from an initial menu, the student can
browse through explanations, choose one of two different sorts of
exercices, go back to the explanation from the exercice and return
without losing his place, etc.  CALLGEN can be used in different ways,
but in this case, it was used to develop sequential lessons on each of
the various topics.  Once the students chose to study a particular
subject from an intial menu, the program forced them to follow a
predetermined path: they were given an explanation, then an exercice,
then more explanations and more exercices; sometimes the program would
slip in a quick comprehension question, ask them to repeat something
they'd just been told, a little bit like a classroom lecture
interspersed with questions and exercices.  Students were shown the two
programs at the initial session and told they could use either or both
in subsequent sessions.  They were also told that we were interested in
seeing which one they liked best.  However, they were not told at any
time that what we were interested in was the amount of control they had
over their progression.

The students had 6 one hour sessions, at each one of which they were
asked to study a particular set of items, using either program, and
they had a mini test each week on the computer on what they'd learned
the week before.  One week after completion, their mid-term tested them
on what they had learned on the computer.  After receiving their scores
from the mid-term, they filled out a questionnaire which elicited
information about attitudes to computers and CALL, which program they
had used and why, and asked to rate on a scale of 1 to 4 (so I could
dichotomize the results later) a whole series of features of both
programs mixed together among which were hidden questions that could
later be interpreted in terms of program and learner control.  Finally,
all students completed the Gregorc Style Delineator to provide some
information on their learning styles.

I have to admit at this point that because of a high initial drop-out
rate in the course -- before the CAI was begun -- we ended up with only
12 students, so that the experiment has no statistical validity.  The
results are nevertheless striking.

My primary objective was to determine whether there was any resistance
to program control.  I had two measures of this: first, if the students
preferred ANALYSE, which gave them much more control over their
progression, that would be an indication that they did not like program
control; second, if features of CALLGEN which took control away from
them (spot questions, or forcing them to do an exercice after an
explanation) were rated low, that would also indicate resistance to
program control.  I was also interested to know if high achievers, who
in theory knew how to use feedback and other features, would be more
resistant to program control than other learners.  Some studies have
shown in fact that high achievers did less well when forced to see
feedback they didn't need.

The results of the questionnaire were quite clear.  Students preferred
CALLGEN, the sequential program, using it on average about 80% of the
time.  Casual observation by myself and an assistant -- we were in the
lab with them -- would have put the figure even higher.  As far as
program control goes, it was rated quite useful or very useful 88% of
the time (48% very useful).  It should be remembered that the students
were not asked questions such as "do you like the program to take
control?" but were asked to rate on a scale of 1 to 4 the usefulness of
features such as "Every time a point is explained, the program makes
you do an exercice to check if you have understood."  Twelve out of
twelve students found this to be a useful or very useful feature.

I would conclude from this that there was very little resistance to
program control among these students.  It must be said that the
students were generally very favorable to CALL, and this was reflected
in the overall ratings of the various features.  Of all the features
rated, 82% of the responses were of the quite or very useful sort.
Nevertheless, and surprisingly, the program control features were rated
a little higher than the others.

As far as learning styles were concerned, there was only one
correlation that was striking, but it was an interesting one.  When
students were split into two groups according to whether they were
sequentials or randoms, their preferences for locus of control were
diametrically opposed.  Sequentials rated program control features
highly 95% of the time and student control features only 71% of the
time.  Randoms did almost the opposite.  They rated program control
features highly only 75% of the time and student control features
highly 95% of the time.  It should be noted that these were independent
variables: it was possible to rate both program and student control
features highly 100% of the time.

Taken together, though with caveats about small sample size, these
results are consistent with the hypothesis that randoms prefer to
control their own progress while sequentials are not as interested in
doing so.  This is what the definitions of the characteristics of
the learning styles would lead you to expect.   Sequentials are the
structured, field-independent, "academic" types, randoms the more
holistic, field-sensitive, non-academic types.

 
 Conclusions

Previous studies found quite consistently that in tutorial and drill-
and-practice programs students did not benefit by being in control of
certain areas of the learning process, in particular pacing and
feedback.  The present study tends to confirm the hypothesis that a
substantial amount of program control, in the form of automatic
feedback, spot questions, and obligatory exercices at set points in a
lesson, is not resisted by students.  In this study, students
overwhelmingly liked the program that took control of these areas, they
liked the specific features in which it took control, and they reported
high levels of satisfaction with CAL in general.  A further hypothesis
which now needs testing is that learning is improved when the program
takes control in this way.

On the question of learning styles, the results are more difficult to
interpret.  What are we to do if randoms (or holists, or field-
sensitives) prefer to be in control yet do not learn better either when
they are or when they're not?  The conclusion seems inescapable that
tutorial and drill-and-practice programs are not well suited to these
types of learners, which is tough, considering their ubiquitousness,
but not altogether surprising.  One interesting study (N = 155), which
found a high degree of correlation between a random learning style
(actually an extravert, intuitive and perceiving personality type) and
failure with tutorial CAI reported that several steps had been taken to
correct the problem.  These included mini lectures to personalize the
instruction, more interaction among students with two or three at a
terminal, and question-discussion sessions.  The study reported that
when these measures were introduced, the correlation disappeared
(Hoffman 1982).  This is in some ways going outside of CAI to correct the
problem but it does seem very promising.  However, I believe the real
solution for the holistic/ random/field-sensitive learner is to make
available programs other than tutorial and drill-and-practice.  Both AI
techniques, involving models of the student's knowledge, and so-called
"emancipatory" or discovery modes of learning, are under-utilized in
our educational system and an effort could be made to develop materials
for them.  Failing some sorts of corrective measures, the much vaunted
individualization of CAI may turn out, like so many other things, to
benefit those least in need and leave the others out in the cold.
 
 

                                      Brian Gill, University of Calgary
 Notes

1.   Gardner (1983) distinguishes seven sorts of intelligence:
     linguistic, musical, logical-mathematical, spatial, bodily-
     kinesthetic, intrapersonal and interpersonal.  He stresses and
     opposes the prevailing undervaluing of some of these intelligences
     at the expense of others, primarily the more "rational",
     "intellectual" ones.

2.   For more information about these inventories, see the entries for
     Canfield, Dunn, Gregorc, Kolb and Myers-Briggs below.

References

Belland, John C., et al. 1985. "Is the self-paced instructional program, via
microcomputer based instruction, the most effective method of addressing
individual learning differences?"  Educational Computers and Technology
Fall:185-198.

Canfield, A. A. 1980.  Learning Styles Inventory: Guide . Birmingham,
Michigan: Humanics.

Carrier, Carol. 1984. "Do Learners Make Good Choices?"  Instructional
 Innovator  29(2), Feb:15-17,48.
 
Carrier, Carol A., et al. 1986. "Instructional Options and
encouragement effects in a microcomputer delivered concept lesson."
 Journal of Educational Research  79(4):222-229.

Chamberlin, Joseph A. 1988. "The computer in education: An investigation of
research dealing with maximising achievement with CAI." Eric Document
ED301181.

Chapelle, Carol & Mizuno, Suesue. 1989. "Students' Strategies with Learner-
Controlled CALL."  Calico Journal  7(2), December:25-47.

Cordell, Barbara J. 1991. "A Study of Learning Styles and Computer-Assisted
Instruction."  Computers and Education  16(2):175-83. [not seen]

Davidson, Gayle V. et al. 1992. "How Do Learning Styles Relate to
Performance in a Computer Applications Course?"  Journal of Research on
 Computing in Education  24(3), Spring:348-58.

de Klerk, Len F. W. 1987. "The Role of Prior Experience in Learning
from Instruction."  Paper presented at the Annual Meeting of the
American Educational Research Association (Washington, DC, April 20-24,
1987) 8 p.

Dunn, R., Dunn, K. & Price, G. E. 1975.  The Learning Style Inventory .
Lawrence, KS: Price Systems.

Gardner, Howard. 1983.  Frames of Mind. The Theory of Multiple
 Intelligences . New York: Basic Books.

Geisert, Gene, Dunn, Rita. 1990.  Learning Styles and Computers . 14 p.
New-York.

Gregorc, A.F. 1984.  Gregorc Style Delineator: Development, Technical
 and administrative Manual . Maynard, Mass.: Gabrial Systems, Inc.

Hannafin, Michael J. 1984. "Guidelines for using locus of instructional
control in the design of Computer Assisted Instruction."  Journal of
 Instructional Development , 7(3):6-10.

Hasselerharm, E. & Leemkuil, H. 1990. "The relation between instructional
control strategies and performance and attitudes in computer based
instruction." In: Pieters, J.M., et al. 1990.  Research on Computer-Based
 Instruction . Amsterdam: Swetz & Zeitlinger, 67-80.

Hoffman, Jeffrey K. & Waters, Keith. 1982. "Some Effects of Student
Personality on Success with Computer Assisted Instruction."  Educational
 Technology  22(1), March:20-21.

Humphrey, John. 1983. "A comparison of how paced and unpaced problems affect
learning during CAI math drills." PhD West Virginia University.

Isaacs, Geoff. 1990. "Course and Tutorial CAL Lesson Design: Helping Students
Take Control of Their Learning."  Educational and Training Technology
 International  27(1), Feb:85-91.

Kolb, D. A. 1976.  The Learning Style Inventory: Technical Manual . Boston:
McBer.

Lombardo, Linda. 1990. "Autonomy in Foreign Language Learning: A
Report."  Rassegna Italiana di Linguistica Applicata  22(3), Sep-Dec:73-
82.

Malone, T. W. 1981. "Toward a Theory of intrinsically motivating
instruction."  Cognitive Science  4:333-369.

McCarthy, Bernice. 1980.  The 4-MAT system . Oak Brook, IL: Excel, Inc.

Myers, I.B. 1962.  The Myers-Briggs Type Indicator . Palo Alto: Consulting
Psychologists Press.

O'Shea, Tim & Self, John. 1983.  Learning and Teaching with Computers.
 Artificial Intelligence in Education . Brighton (U.K.): The Harvester Press,
307p.

Papert, S. 1980.  Mindstorms: Children, Computers, and Powerful Ideas . New
York: Basic Books.

Rowland, Paul & Stuessy, Carol L. 1988. "Matching Mode of CAI to Cognitive
Style: An Exploratory Study."  Journal of Computers in Mathematics and
 Science Teaching  7(4), Sum:36-40,55.

Sales, Gregory C. & Williams, Michael D. 1988. "The Effect of Adaptive
Control of Feedback in Computer-Based Instruction."  Journal of Research
 in Computing in Education . 21(1), Fall:97-111.

Sewall, Timothy J. 1986.  The Measurement of Learning Style: A Critique of
 Four Assessment Tools . Wisconsin Univ., Green Bay. Assessment Center. 70 p.

Soulier, J. Steven. 1988.  The Design and Development of Computer-Based
 Instruction . Boston: Allyn and Bacon, Inc.

Vargas, Julie S. 1986. "Instructional Design Flaws in Computer-Assisted
Instruction."  Phi Delta Kappan  67(10):738-744.

Viteli, Jarmo. 1989. "Learning Styles and Individual Differences in Learning
English Idioms via Computer Assisted Language Learning in English as a
Second Language."  Paper presented at the Annual Meeting of the
European Association of Research on Learning and Instruction (3rd, Madrid,
Spain, September 4-7, 1989) 25 p.

Weir, Sylvia. 1987.  Cultivating Minds. A Logo Casebook . New York: Harper &
Row, 258p.
 

         Learning Styles, Locus of Control and Courseware Design

Abstract

The concept of learning styles, despite some theoretical problems,
helps to make us aware of the difficulties faced by many learners who
tend to be classed as non-academic and which our educational systems
may turn into so-called underachievers.  Although it seems that
computer-assisted instruction should help these learners because it
offers choices to the learner and is individualized, studies have
almost uniformly shown that it does not.  When they are allowed control
over what speed to work at, what help to seek, what feedback to ask
for, such learners do not make good choices, although they, like most
of us, do like the idea of being in control.  It was hypothesized that
despite their expressed preference for being in control, a preference
which is ideologically comforting, learners might in fact be quite
satisfied with programs which largely took control away from them.  To
test the hypothesis, a study was carried out in which a university
level French language class was given access to two CALL programs with
almost identical content, one largely learner controlled, the other
largely program controlled.  Small final class size made the study
statistically unreliable, but there was very substantial support both
for the program, and for features within both programs, which
effectively took control away from the learner.