Go to:  | Biblioteket | Digitale tekster | Vestfold College | Vestfold College Library |
Tønsberg: Høgskolen i Vestfold, 2001
Egil Børre Johnsen: Textbooks in the Kaleidoscope.
| Front pageContents | < Previous chapter | Next chapter >
 
 

Chapter IV
 

The Use of Textbooks
 
 
 

Introduction

An analyst may approach "books in use" in any number of
ways. Approaches will be determined by a number of factors
which may or may not be independent of one another, but
which will all be based on certain notions about objectives,
means and educational practice. Schematically speaking, most
of these notions will fit into one of three main categories,
depending on whether they deal with concepts based on
teaching, textbooks or knowledge.
  Teaching takes place at school, although it is to some extent
followed up at home. Today it is officially accepted - in
curricula and among educators - that the school's primary
objective is to provide a venue for cooperation, student-
centered and teacher-centered teaching and personal
development, although this view puts the role of the textbook
in a controversial light. It is therefore common to question the
extent to which teaching is textbook-centered. Some say that
not using textbooks is a prerequisite for independence as a
teacher, while others base their teaching on new textbooks
that endeavor to promote independence. As to the former
group, however, textbooks are often "avoided" by copying
parts of them into these "independent" teachers' own
programs, and it is not at all uncommon for the latter group
of teachers to exercise a great deal of discretion in relation to
the officially adopted text. Views expressed in connection
with debates and polls are one thing; in such contexts,
publishers often respond to the accusation that their
textbooks dominate the classroom situation by postulating
that teachers only choose books that fit their needs. What
actually takes place in the classroom is something else
entirely. The situation has been described as follows:

  Textbooks are hopeless no matter how you approach them.
  They're supposed to provide stability and continuity in
  studies, but can't be used during planning. They control
  the classroom situation to the extent that not using them is
  seen as a sign of independence. They also exert so much
  control that they have to be regulated by conditions laid
  down in a national control system. But they do not exert
  enough control for the author to be able to assume that
  teachers will follow them. (Johansson 1988, p. 35.)

There will probably always be a plethora of opinions, ranging
from vehement disagreement to concurrence, about the
 relationship between an explicit teaching ideal - as
formulated in curricula or in teachers' minds - and actual
practice. The magnitude of the variations will depend on the
correspondence between theory, as expressed in curricular
objectives, and reality, as experienced in teachers' training,
school working environments and teaching staffs. The time
factor plays an important part in this context. First, all but
the smallest schools are staffed with teachers of different ages
and educational backgrounds. Second, there is some
disagreement as to how long it takes for the comprehensive
new ideas introduced in new curricula to filter through
(Strømnes 1987). For example, the curriculum for primary
and lower secondary education in Norway mentions the word
"textbook" very few times in its 306 pages. In every case, the
word occurs simply as an item on a list of equally-ranked or
high-priority teaching aids. How can this be interpreted,
except as a clear signal? In light of that, it is surprising that
textbook sales have increased since the plan was introduced.
Although not based on empirical evidence, it therefore seems
logical to assume that textbooks still occupy a central position
in teaching. The most recent curriculum was introduced in
1987 however, so it has not yet had adequate time to filter
through into practice. Nonetheless, it should also be safe to
assume that the frequency and manner of using textbooks
varies according to teacher, subject, level, school, student and
home. As there are many different ways of using textbooks
and as teaching is a complex situation, it is particularly
difficult to draw any general conclusions on the basis of
studies that focus on the books in use.
  (As for the implications of the textbook concept, I refer to
the discussion on page 24.)
  The term knowledge has been substantially expanded in a
large number of Western countries during the past few
decades. Although pupils are supposed to learn facts in all
school subjects, they are also supposed to develop a special
understanding of each individual subject as well as certain
skills and general attitudes.
  At the same time, the sheer volume of knowledge and the
number of channels of information are proliferating so
explosively that scientists have seriously begun to examine the
need for what has been called unlearning (Lærum 1991).
  This picture is further complicated by the fact that
knowledge is used as a polarizing term in the political debate
about the school system. Apparently impassable schisms have
evolved between those who claim that pupils no longer
receive any core knowledge and those who take a broader
view of knowledge. Such a conflict does not necessarily
follow political dividing lines. It also exhibits some of the
traits seen in the debate for and against grades, where there
are divided views within several political parties. Insofar as
it is valid to say that it is possible to learn anything about
fact/non-fact oriented knowledge from curricula, there are
national divisions. At the upper secondary level, for example,
 the subject of literary history comprises a considerably larger
body of knowledge in France than in Norway, while Austria's
history curriculum and history books are much larger than
those used at the primary and lower secondary level in Great
Britain.
  If we add that subject distinctions continue to exist despite
attempts at integration among certain disciplines, it becomes
clear that if studies of "books in use" refer to measurements
of what is taught, one might ask whether the phenomenon is
quantifiable at all. What is there to measure? Compare this
with the situation when textbooks are written and developed.
What is to be taught? A look at new textbooks in the Nordic
countries, for example, will show that they tend to pass on
facts, train skills and instill certain attitudes. This is also
reflected in the books' linguistic and physical presentation.
The books contain an increasing proportion of personifying
and emotional texts, while technology has made it possible to
package this growing body of facts in an attractive, often
compact form. Almost all modern textbook publishers
constantly strive to strike a balance between facts, skills and
attitudes.
  The same uncertainty may be exemplified by a relationship
that deals with researchers', teachers' and pupils' evaluations
of textbooks. Countless evaluative systems have been devised
at individual schools as well as at the national level. But none
of the models have yet achieved authoritative status or been
declared normative or routine by any of the bodies that select
school books, for example.
  Anyone who answers yes to the question of whether the
passing on of knowledge can be measured will have to deal
with one last discussion before starting to take measurements.
This second discussion involves the relationship between
textbooks and teaching as a whole: Is it at all possible to
isolate books as a separate, measurable part of the classroom
process? Although systematic classroom observation has some
rather spurious aspects (Clark 1991), it is possible to measure
a variable such as time quite accurately. Some people cite
such measurements as their only grounds for contending that
teaching is textbook-centered. But if we move one step
further and measure knowledge acquisition, we encounter
methodological problems deriving from our point of
departure: What is knowledge - in and/or outside textbooks?
Our means of measuring will provide the answer to the
question and to some extent determine the nature of the
knowledge to be measured.
  As regards the terms "teaching", "textbook" and
"knowledge", we see a trend which places distinct, fairly
narrow constraints on our perspective if the studies are to
represent reliable measurements on the one hand and, on the
other, there is the question of whether we shouldn't take a
broader approach if such investigations are to perform any
 valid function.
  The following survey divides the studies into three groups,
depending on whether they deal with control, accessibility or
effects and effectiveness. The question of textbook control is
dealt with first because the answers are so significant to the
results achieved in the other two groups. Extensive use and
strict control will amplify any damage done by poor
accessibility and improve otherwise bad teaching if
accessibility is good. Under certain circumstances, a
comparable correlation will apply to the relationship between
control and learning.
  There is also another important distinction between the
first group and the other two. Studies of textbook control
have often focused on teachers and publishers, i.e., on teacher
training and practice and on publishing house traditions and
ideologies. Many critical analyses of biases in material
selection and presentation are based solely on the assumption
that the books are read, trusted and used by many teachers.
The other groups tend to focus more on the books' suitability
for use as teaching materials, i.e., their focus is on the pupil.
To some extent the studies examine the language and/or the
methods used in the books (accessibility), and to some extent
they try to pinpoint what pupils can or may have learned
from them (effectiveness).
 
 

Control
 

Control per se is a complex concept. It can have both positive
and negative connotations. In principle, reading and using
literature may entail freedom and dialogue as well as control.
Ideally, control would be deliberate and positive if well-
qualified, experienced teachers opted to remain loyal to the
content, structure, methods and philosophies of a textbook
because it was consistent with their own and other talented
colleagues' mature views on educational goals in all these
areas, and because their pupils like to read it. If a less well-
qualified, less discerning teacher in another class chooses the
same book and the same course of action, control will still be
a positive factor, albeit for different reasons. However, if the
teachers at a school, out of necessity or ineptitude, base their
teaching on books (approved or not) which pupils reject, we
see a negative form of control ascribable more to the school's
budget than to the books' status or quality. We see an even
more blatantly negative form of control when teachers fail to
prepare their pupils for the pitfalls found in many of the
textbooks the pupils use as independent sources of
information for reading on their own, working in groups,
using workbooks or doing homework. In isolation, textbooks
may also contain a potential for negative control due to
unreliability, incomplete contexts, structures inconsistent
with the subject's basic premises, methods that conflict with
important objectives for work routines and/or
imprecise/vague use of language.
  In a separate report for the DsU 1980:4 project (see page
385), Christina Gustafsson makes a clear distinction between
positive and negative control. She found that in theory and to
varying degrees also in practice, textbooks:

  (...) exercise negative control:
- as regards material selection and sequencing
  when the teaching medium is not based on the logical
  structure of the subject
- as regards language
  when the pupils don't need to understand the language of
  the teaching medium
- as regards learning
  when pupils do well merely by memorizing the words in
  the textbook when basic skill like speaking and writing are
  not called for
 - as regards ideology
  when there is no discussion about the values expressed in
  the teaching medium
- as regards methods
  when the teaching medium "forces" users into certain
  routines. (Gustafsson 1982, p. 96.)

Considerable Control?
What is possible in the field of use studies, is to examine
finite situations and uncover tendencies toward control. This
has been done, though only on a modest scale. The following
investigations are mainly arranged according to the different
aspects of teaching they have measured. The dominant factor
in this context is the consumption of time.
  Birger Bromsjö examined social science teaching and
textbooks in Swedish primary and lower secondary schools in
the 1950s (Bromsjö 1965). First, he had teachers make lists,
ranking the subjects included in the curricula and noting how
much classroom time they wanted to devote to each of them.
He subsequently compared these data with measurements of
time consumption and schedules in relation to the books. For
example, when the distribution of teaching time actually
followed the book even when the teacher had different
priorities (time/scope) than those expressed in the book, the
teaching was defined as textbook-centered. Bromsjö found no
strong, general tendency in that direction, but he did find
that some factors which had received little priority on the
teachers' lists dominated the classes to an extent that
corresponded to the priority they had been given in the
textbooks.
  Certain other similar time measurements point in the same
direction without posing a problem of method. One obvious
question concerns the extent to which the amount of time
devoted to a topic can serve as a measure of influence.
Bromsjö and others have used quantitative methods to show
that the sequencing of material in the books can affect
teachers' own sequencing. But the question of how much the
material selected, presented and read controls teaching in
terms of influencing pupils' attitudes and knowledge
acquisition will remain unanswered because time cannot be
used as the only valid or absolute measure in such contexts.
  Ten years later Ulf P. Lundgren studied textbook control
in mathematics, history and social science in upper secondary
school (see Johansson 1988, p. 10). He compared three factors:
the number of pages in the textbooks, the lesson units as
planned by the teachers and the classes as they were actually
taught. He found the correlation to be greatest between the
books and the lesson plans. Still another observation led
Lundgren to assume that teaching was significantly textbook-
centered. He compared the sequencing of topics and points in
curricula, books and classroom teaching. In mathematics, only
three of 22 classes deviated from the  pattern suggested by
the books. In social science, eight of 17 classes deviated from
the books and in history 25 of 39 classes departed from the
books' sequencing. These figures led Lundgren to conclude
that the books exert control, but that the control varies
according to the topic structure of the subjects themselves.
   Time use and the organization of teaching time were
employed as units of measurement by Danish scientists in the
1970s, when they studied the teaching of knowledge subjects
at the middle and lower secondary level in a total of 125
classes in Copenhagen (Jensen 1975, Jensen 1977). The classes
were observed for two-week periods over a considerable
period of time. The measurements distinguished between class
instruction, group work and individual work. Moreover, they
distinguished between categories of textbook functions which
were in turn related to the three main types of teaching. This
allowed Jensen to operate with four types of control: teacher-
centered, pupil-centered, teacher and pupil-centered and
textbook-centered. The last category proved to occupy more
than one-third of the time used in middle school and about
15 per cent at the lower secondary level. Control was defined
on the basis of what the pupil was actually doing, which once
again entails a quantification based on the pattern used by
Bromsjö. The question was, and therefore still is, whether it
is possible to measure any control other than the directly
observable control physically manifested in the classroom.
  The tendency to perceive textbooks as exerting a great deal
of control is confirmed in one of the two largest Nordic
studies in this field: Ingvar Sigurgeirsson's "Inquiring into the
nature, role and use of curriculum materials in Icelandic
schools" (Sigurgeirsson 1990). The measurement of time use
plays an important part in this study as well, although
Sigurgeirsson also incorporates other important factors.
  The study was motivated by the authorities' desire to renew
an outdated book stock. The aim was as follows: "The
teaching materials were to serve as a foundation for teaching
methods, where the main emphasis was to be placed on the
pupils' independent study, a stimulating school environment,
inter-disciplinary teaching, various inductive teaching
methods and creativity." In other words, the material
included teaching media in all subjects, new teaching media
developed according to new curricula and national teaching
media that dominated in virtually all the country's schools.
  The method involved observing 20 classes at 12 schools (for
a total of about 1000 hours), interviewing teachers and pupils
and sending out questionnaires.
  Some of the main categories were time use, types of
teaching aids, the questions posed by teachers and the texts,
types of pupil activities, the interdisciplinary nature of the
work and teaching methods; then the categories were further
subdivided.
  Here are a few of the most important conclusions as Sigur-
geirsson himself has summed them up:

  Published material plays a decisive role in most classrooms.
  Approximately 60 % of teaching time is spent working
  directly with the material. In English: 96 % of teaching
  time. In mathematics and social studies: 75%. Teaching
  materials aren't used at all in 20% of the classes.
    The material is generally reviewed page by page.
    Teachers' manuals are not used much. No more than 5-
  10% of the teachers use teachers' manuals to any great
  extent.
     Only limited use is made of materials such as overheads,
  video, slides, games and other teaching aids.
    The teaching methods are one-sided.
    The material controls what happens in the classroom to
  a large extent - only not in the ways envisioned by the
  authors when they wrote the material! (Sigurgeirsson 1990,
  pp. 2-3.)

The DsU 1980:4 investigation, one of the most comprehensive
studies made in the Nordic countries besides Sigurgeirsson's,
is part of a project entitled "The Market for Educational
Materials" (see page 385). The mandate was "to investigate
the control exerted by teaching media during the teaching
process and to discuss how such control can be avoided or
decreased." (P. 15.)
  The investigation was set up in two stages. The first
consisted of preliminary studies and was mainly intended to
test methods. It consisted of observing and analyzing the
classroom situation during 39 hours of instruction in different
subjects and classes and at different levels in the two
municipalities that were later to take part in the main studies.
Christina Gustafsson, head of the project and author of the
report, described the preliminary studies in a separate report
(Gustafsson 1978). Two important observations were noted:
it proved to be very difficult to separate the role of teaching
media from other factors involved in teaching and, in
contrast to the plan inherent in the mandate for the study,
there was little point in summarily defining textbook control
as a negative phenomenon. Teaching media are aids and, as
such, they have a function that is positive rather than
negative, both in principle and in practice.
  Like Sigurgeirsson, Gustafsson relied on observations and
interviews. The main studies comprised observations of 217
classes in Swedish, English, natural science, social science and
electrical teletechnology at different schools in the two
different municipalities. Although few classes were
investigated in each subject, several lessons were monitored
in each class. For practical reasons, it was decided not to use
tape recordings and transcripts, but to employ a printed form
on which the observers made notes during the classes,
arranging the material into five categories: Time (spent on
different types of teaching media and other activities), pupil
grouping (organization of the class and teaching), materials
(classifying all the teaching media used), content (of the
teaching media), and comments (mainly evaluations of the
textbook control effect).
  On some points the results show a certain correlation with
Sigurgeirsson's findings. Gustafsson can also rank the subjects
on a scale based on the role played by the teaching media in
a certain subject (English) in terms of time, method, content
and learning. In other subjects (social science), on the other
hand, she describes this role as being very modest in the same
 areas.
  Like Sigurgeirsson's, Gustafsson's investigation concludes
that the teaching media probably have little effect on the
methods employed in the classroom. However, an example
will show how difficult this factor is to measure. Like all
other studies in this field, Gustafsson's investigations
concluded that reliance on textbooks is most dominant in
foreign languages; according to her observations of English
classes in the sixth grade, 71 per cent of all classroom time
was spent on the textbooks. One of the English teachers, who
was observed for a total of 12 hours, used books almost all the
time. She relied not least on workbooks, in which a typical
drill consisted of dialogues intended for the pupils to work
together in pairs. But the teacher chose her own method; she
played the one role, while the whole class played the other.
This was obviously the teacher's method of preference; she
disregarded the most important other drill models in the
workbook entirely. How controlled is such a teacher, and how
much control does the book exert on the pupils when such
procedures are followed? (This example is taken from
Johansson 1988, p. 12.)
  Insofar as it is possible to talk about a general control
effect across the boundaries between disciplines, levels and
teachers' personalities, Gustafsson also concludes that this
effect is primarily a factor of content as expressed by the
selection and sequencing of material. Yet she is far more
cautious than Sigurgeirsson in naming teaching media control
as the force that generally predominates in the classrom: "The
question is not whether but how teaching media exercise
control and that is an effect of the interplay between the
teaching media, the environment and the individuals
involved." (P. 196.)
  Norwegian researcher Svein Lorentzen studied the
development of social studies at the lower secondary level
(Lorentzen 1984). Lorentzen, who based his findings solely on
teachers' own responses to questionnaires, found that the
majority of them made extensive use of textbooks; most of
them also responded that they viewed textbooks as "very
important" in social science, history and geography. In
contrast, supplementary literature and AV equipment were
not used much. One of Lorentzen's conclusions was especially
interesting in light of what was said earlier about the
importance of taking the point in time and lapse of time into
account when evaluating the results of use studies (see page
158):

  Of the teachers who had teaching experience prior to the
  introduction of the National Curriculum, nearly two-thirds
  stated that they had made "some changes" in their
  selection/use of teaching materials in social science after
  the implementation of the National Curriculum.
    Only some 6 % responded that they had made
  "significant changes", while nearly 20 % answered that
  they had made "a few changes" in this aspect of their
  social science teaching.
    As far as teaching materials are concerned, a
  significantly larger percentage of women than men
   reported that they had made "signficant changes" in their
  social science teaching following the introduction of the
  National Curriculum (Lorentzen 1984, p. 326.)

Lorentzen's report may also be viewed in conjunction with a
more recent Swedish report: Knowledge of the World around
Us: Knowledge Subjects. Background, Description of
Teaching, the Fostering of Democracy (Svingby - Lendahls -
 Ekbom 1990). The report is part of a major national school
evaluation project based mainly on knowledge and skills.
  As in Lorentzen's study, the Swedish questionnaires asked
teachers about their actual use of and attitudes toward
textbooks. The responses showed a strong degree of
agreement in both areas. Nearly half of the 157 teachers
surveyed stated that they used textbooks every day, while
about the same number answered that they used textbooks
weekly. Almost 80 per cent felt that textbooks were either
"very important" or "important" in history and geography.
By contrast, only about 20 per cent felt the same way about
social science textbooks. This agrees with the comparable
results of all the studies referred to in this chapter. A similar
degree of correlation is also apparent in teachers' manuals and
workbooks, which were not used much. The results of the
Gothenburg studies confirm the results derived from other
studies on still another important point. The extended
teaching media concept has not caught on in practice in the
area of AV equipment (radio/TV). New technology has not
had the planned effect on pupils' "discovery" of their local
environment.
 

Little Control?
The tendency to want to look at textbooks as relatively
dominant control instruments is not universal. A number of
studies have also been made that point in the opposite
direction. Relatively extensive measurements were
undertaken in the Netherlands recently in a study covering
more than 700 primary and lower secondary schools (Reints -
 Lagerweij 1989). The results indicate to some extent
significant teacher independence in relation to teaching
media. The time teachers spent on books when planning and
in the classroom varied from an average of 62.9 per cent in
geography to 31.3 per cent in social science. Ten per cent of
the schools did not use textbooks in mathematics; the
percentage was higher in several other subjects, particularly
foreign languages. (This last observation deviates substantially
from the results published in DsU 1980:4; see page 167.)
  The Dutch investigation also examined how teachers used
the books in four core subjects in primary and lower
secondary school. They divided methods of use into three
categories: Eleven per cent of the teachers followed the books
page by page. Nearly half allowed the books to dictate the
structure of much of their teaching, but they supplemented
the books with their own materials. For some ten per cent of
the teachers textbooks were just one of several equally
 important sources. However, many different combinations of
the last two methods of use were practiced by roughly 20 per
cent of the teachers. Ten per cent of the teachers did not use
textbooks at all.
  The most comprehensive investigations of textbook control
were conducted in the USA. In 1988 Donald Freeman and
Andrew Porter presented a study entitled "Does the Content
of Classroom Instruction Match the Content of Textbooks?"
(Freeman - Porter 1988). They worked with seven
mathematics teachers in primary and lower secondary schools
in three school districts. The teachers kept teaching logs and
were interviewed. On the average, they spent about one-third
of the lessons on textbooks. More than 80 per cent of all
classroom time was spent on the topics/problems found in the
books, but there was a great deal of individual variation on
this point. Although the sample comprised just seven
teachers, researchers found it was possible to divide them into
three main groups according to their use of and attitudes
toward textbooks. One group followed the book faithfully,
lesson by lesson, using little or no time on supplementary
material. Another group followed the plan and progression in
the book, but was selective about the texts. And the third
teacher group broke up both the structure and content by
adding supplementary material. The division into three
groups corresponds to the grouping used by Lee J. Cronbach
(see page 181). Freeman and Porter, who have conducted a
series of studies on the teaching of mathematics in lower
secondary school, distinguish between the main kinds of
choices every teacher has to make. They are referring to the
choice of topic, the time set aside for a topic, which pupils
are to study the material, the sequence of the topics, and the
goal of the teaching (how much should be
understood/learned?). Rather than discussing "control",
Freeman and Porter write about "teachers' adherence to
textbook copies", which they view as a function of the
strength of their perception of or belief in the authority of
textbooks. (This discussion is based on the review presented
in Stodolsky 1989.)
  Based on this point of departure, the question of textbook
function becomes a question of teacher function. In light of
the emphasis Freeman and Porter place on content and topic,
the way in which topics are defined will also be decisive. The
broader the definition, the easier it will be to show a
correlation between textbook and teaching. The narrower the
definition, the greater the distance.
  Freeman and Porter's approach to the problem of control
shares one feature with the other use studies mentioned
above. None of them say much about the relationship
between the authorities of the teacher and the text, as
perceived by pupils/parents. They also overlook or avoid the
"balance of power" between teacher and text: What about the
effect of the "intermediate" language that emerges when the
teachers speak at the level between the book and the pupil,
and which might, for instance, be analyzed from a socio-
linguistic point of view? The question has been posed by
Australian and Canadian researchers who have presented
almost diametrically opposed views on the balance of power
 between teacher and textbook as far as influence is concerned
(Luke 1989 and Olson 1989; see page 173).
  As mentioned above, several researchers point out that
groups or types of teachers use textbooks to different degrees
and in different ways. This situation has motivated several
people to question the reasons for such differences. Some
have discussed teachers' education (Ball - Feiman-Nemser
1988) in this perspective, while others have emphasized the
grade level (Lorentzen 1984; Tournier - Navarro 1985). In
both cases, many studies indicate that independence increases
and the time spent on textbooks decreases in direct
proportion to the grade level being taught. The grounds for
this conclusion are rather vague and uncertain, however. As
early as in 1930, William C. Bagley studied the practices
followed in 539 classroom hours in 30 different states (Bagley
1931). He drew the conclusion that the strongest determinant
in the teacher-textbook relationship was not necessarily
education or level, but seniority. Svein Lorentzen's
questionnaires examined the same criterion, but he arrived at
the opposite result (Lorentzen 1984, p. 372). Adding criteria
such as the teacher's specialties and his/her motivation for
becoming a teacher, level and practice, would probably
complicate matters further.
  Susan Stodolsky (Stodolsky 1989) is probably the one who
has made the most direct claim that textbooks generally play
a fairly modest part in relation to instructional content and
progression. She examined the teaching of six mathematics
teachers and six social science teachers in the 5th grade
(Stodolsky 1988). Her results tallied with Freeman and
Porter's:

  In sum, our cases suggest that teachers are very
  autonomous in their textbook use and that it is likely that
  only a minority of teachers really follow the text in the
  page-by-page manner suggested in the literature. Use is
  much more varied than usually suggested, particularly
  when one considers more than just the topics contained in
  the books. Even with regard to topics, we found, as
  Freeman and Porter did, that what teachers teach is in the
  book, but they do not teach everything that is in the books.
  Thus, math textbook content tends to place something like
  a cap on content coverage in classrooms, although putting
  something in a book does not guarantee instruction will be
  devoted to it. (Stodolsky 1989, p. 176.)

One of the points of the study is that some aspects of the
teacher-textbook balance differ significantly in mathematics
and social science. This has often been taken for granted in
earlier literature, in that studies have assumed that social
science teachers were more independent of the textbooks than
mathematics teachers, who were believed to be bound to a
certain pattern of progression (problem by problem and
assignment by assignment). Stodolsky arrived at a more
differentiated result. While it is true that the use of
supplementary material is far more common in social science
than in mathematics, she found that when a text is used
during a class in social science, it is followed very closely; its
 "topical sequence is usually preserved (...). More skipping
and departure from text sequence is found in math."
(Stodolsky 1989, p. 181.) This page by page faithfulness in
social science is confirmed by Yngve Nordkvelle's
observations (Nordkvelle 1991).
  In her article "Is Teaching Really by the Book?"
(Stodolsky 1989), Stodolsky's conclusion also indicates an
explanation as to why investigations and teachers have so
overwhelmingly perceived textbooks as controlling. She points
out that most of the studies have focused on basic books in
reading skills, and that we may be overly hasty when we draw
general conclusions from those results. Moreover, she
contends that "The faulty assertions have also been bolstered
by a lack of direct observation or other systematic data with
which to verify or refute them." (P. 182.)

The Authority of the Text
The question of the text's degree of dominance is a question
of the authority of the text. In educational cultures such as
those in Western countries in the 1800s and well into our own
century, teaching frequently revolved around reading aloud,
memorization and verbatim repetition from books. The
authority of the text was indisputable and undisputed;
consequently, concepts such as "accessibility" and
"effectiveness" were never on the agenda. Today textbooks
still possess attributes which lend them authority, despite the
fact that they are just one of a multitude of media inside as
well as outside the classroom. Although an author's name is
given, somehow textbooks don't seem to bear what Roland
Barthes has called a personal "signature" (Barthes 1977).
Coupled with their official "neutrality", this lends textbooks
an impersonal authority. Regardless of cultural period, the
fact that textbooks don't consist of discourse elevates them
above criticism and lends them "great authority", as David
R. Olson claims in a number of analyses (Olson 1989). Olson,
too, recognizes that more devotion to dialogue and greater
opportunity for more personal, conversa-tional language have
broken down what he quotes de Saussure in calling "the
tyranny of written language". At the same time, he feels that
the ability to "speak a written language" (Olson 1977, p. 270)
is still our ultimate goal as we advance through the educa-
tional and examination system.
  Olson's view has been challenged by Suzanne de Castell,
Carmen Luke and Allan Luke (de Castell - Luke - Luke
1989), who claim that written language/textbook texts alone
carry little authority today. First of all, there is a teacher in
the picture as long as the anonymous texts need to be
interpreted and second, behind a teacher, there is the system:

  We have argued for a more interactive and pragmatic
  explanation of text apprehension whereby meaning is
  contingent on the interaction between the reader's prior
  knowledge, the institut-ional setting within which the
  reading task is situated, the teacher who teaches the text
  and the distinctive features of the textbooks per se. This
  relationship, we have noted, is delimited and constrained
   by the rules of schooling which position teacher, text and
  student in hierarchical levels of power and authority.
  (De Castell - Luke - Luke 1989, p. 258.)

In a comment to the criticism from de Castell, Luke and
Luke, Olson points out that the most important point is not to
measure the degree of authority carried by the respective
elements, but to reveal the existence of two more or less
parallel structures. "To me the structure of language and
texts is neither subordinate nor superordinate to the social
structure. A change in one is likely to change the other and
vice versa" (Olson 1989, p. 262). This last statement implies
that the evolution of media and communications technology
and its consequences on (textbook) language will in turn
influence social conditions. This warrants new types of text
studies because the language in and style of the texts are
indicative of what schools and society associate with truth
and authority at any given time.
  The point made by de Castell, Luke and Luke has been
carried to the extreme by Maingueneau and Gordon, who
consider school language as being somehow superior to
textbook language - the school is its own language (see page
147 and page 148). Jean Anyon's studies indicate that one and
the same text can have vastly different meanings and
interpretations, depending on social context (Anyon 1981). In
Sweden, Ulf Lundgren deals with this question in a book
called Organizing the World (Lundgren 1989). It is a study of
curricular theory which discusses the language used in
curricula, schools and textbooks. Lundgren's work is based on
theories put forth by Bernstein (Bernstein 1971) and Halliday
(Halliday 1973). He asks how evaluations and knowledge are
selected and organized; parts of the analysis examine how a
specific curriculum and a specific teaching medium compare
with organizational parameters with regard to control of
teaching. At the end of his book, Lundgren writes that
studies of linguistic interaction in the classroom (teacher-
book-pupil) provide a foundation for the development of
theories about which premises control which teaching. But
such a theory "cannot be placed in a simple, rigid structure.
The complexity inherent in the various relationships is far too
great." (Lundgren 1989, p. 227.)
  Like other researchers referred to in my book (Michael
Apple, Pierre Bourdieu, David R. Olson, Ian Westbury),
Lundgren strongly underlines the necessity of tying the
analysis of ecucational processes to general theories of
education and its function in society. However, if such
overall and holistic theories in the sociology of knowledge,
culture and education are mingled into pieces giving each
party only its kaleidoscopian fragment in the textbooks
themselves, with constant incorporation as an issue stronger
than that of change, one might question the starting point for
textbook research. Should such research be organized
primarily in relation to categories established by theorists in
the sciences of sociology and education? Or would it be
functional to start looking more systematically also for
categories more directly connected with the textbook as a
 general as well as a special form of literary expression? It is
my opinion that the latter possibility has not been exploited
to the same degree as the former. I try to follow up this point
of view in the two next chapters (Accessibility, Effects and
Effectiveness), as well as in my final chapter (Conclusion).
  John A. Zahorik demonstrates the teacher-book complexity
in his investigation of the relationship between textbooks and
teaching styles (Zahorik 1990 and 1991). He based his work
on that of K. Hinchman (Hinchman 1987) and D. Alverman
(Alverman 1989), both of whom distinguish between three
types of usage that also display a strong correlation to three
teaching styles. The textbook may be perceived and used as
a) a source of facts to be learned ("coverage"), b) a source of
different types of activities ("textbook based activities")
and/or c) a basis for interpretation and discussion ("higher
level interpretation/reference"). Hinchman and Alverman
based their work directly on observation of teachers, while
Zahorik used questionnaires. All the investigations confirm
that textbooks are used extensively and conclude that the
ways of using them range from a) to b) to c). Zahorik goes
one step further by investigating the connection between
teaching style and the books on the one hand, and teachers'
own "ideological belief" on the other. He cautions against
attaching too much importance to conclusions drawn on the
basis of questionnaires, but, referring to the teachers in group
c), claims: "The text-thinking teachers had different beliefs
about students, knowledge, and teaching than the other types
of teachers" (Zahorik 1991, p. 186). Hence Zahorik gives the
following answer to the question of textbook dominance: "If
and how the teacher uses the textbook follows teaching style
rather than dictates teaching style." (Zahorik 1991, p. 195.)
For Zahorik, the issue is not whether people are for or against
using textbooks, but how teachers can adapt the books to
their own teaching style. Indirectly, this view supports
Suzanne de Castell's claims about the necessity of working on
the attitudes of teachers and pupils to textbooks (see page
201).
  Work done in the field of readability has tended to pay
increasing attention to pupils' qualifications as readers, while
focusing on the division of responsibility between textbook
and teacher. Analyses conducted by linguists and educators
such as Maingueneau, Lundgren, Olson, de Castell, Luke and
Luke have placed textbook, teacher and pupil in a socio-
linguistic triangle where the question of authority, and
consequently that of control, becomes one of several rather
mercurial forces that work together or in opposition at any
given time.

Perspectives
Most of the investigations mentioned here, as well as the
majority of a number of smaller-scale investigations not
mentioned in this survey, conclude in one way or another that
 textbooks have had and continue to have a strong controlling
position in the classroom. It is essential, however, to be aware
that such conclusions are based on observations, polls and
other kinds of surveys, each of which somehow has its own
built-in limitations, especially practical ones. In addition,
both positive and negative connotations are conferred on the
key word "control" when it is used to describe approaches to
problems.
  The answer to the question of textbook dominance may be
summed up in five points:
  First, in terms of volume, textbooks do dominate classroom
instruction if one measures the total share of classroom time
during which textbooks of one kind or another are in use.
  Second, one might speak of a general dominance if one
accepts teachers' information on the use of textbooks and
their general attitude toward them, as expressed in polls.
  Third, textbooks can only be assumed to have a certain
degree of systematic influence on teaching methods.
  Fourth, the question of textbook control is especially
difficult to answer with respect to its implications for the
subject matter. (The problem of ideological influence and the
question of learning effect is discussed elsewhere (see page
158 and page 226).)
  Fifth, and this also involves methods, the investigations
often overlook the question of textbook type. This may be
due to practical problems, but no one can claim that
researchers haven't been aware of the problem. It was
addressed in the very first known example we have of an
English-language, educational periodical which devoted an
entire issue to the topic of textbooks. In a special issue of Phi
Delta Kappan, Francis St. Lawrence first dealt with the
question in 1952 (St. Lawrence 1952). The subject was high
school biology and 170 teachers were interviewed. The results
were particularly interesting in a time perspective. A great
deal of supplementary material was used and applauded,
while attitudes to textbooks were generally negative. In other
words, the investigation pointed in a different direction from
the use studies conducted in the 1980s.
  It can be difficult to distinguish between the use of
whether and that in Freeman and Porter 1988 and Stodolsky
1989. Right from the start the stated objective - to determine
whether specific, analyzed material controls the teaching -
may seem to imply a general complaint of textbook
dominance. Other analyses in the same area take a less clearly
expressed position on the use of textbooks. This applies to
some of the studies mentioned earlier, including Bromsjö
1965 and Lundgren 1969 (see Johansson 1988), as well as to
the large-scale Dutch investigation (Reints - Lagerweij 1989).
Lorentzen 1984 and Svingby - Lendahls - Ekbom 1990 also
maintain neutral positions. Note that in the last two studies
the use of books was only addressed as one of many elements
within the framework of more comprehensive investigations.
The objective was to examine many elements of control in
context, something which may mean right from the start that
each individual element receives less attention than it would
get if it figured more prominently in the investigation.
   Regardless of their opinions about the phenomenon of
dominance and regardless of their point of departure,
investigations have been affected by the notion that textbooks
can be improved as subordinate teaching aids. They are there
to serve the controlling teacher who enjoys freedom of
choice, and they are not intended to stake out too much of
the teacher's course. In a few exceptional cases, textbooks
have been examined with the clearly stated intention of
making them better suited to control teachers' methods of
working (Sigurgeirsson 1990). The studies are otherwise
colored by the prevalent educational view that the teacher,
independent of controlling forces outside the school (with the
exception of laws, plans and rules), is or should be the
principal determinant of the teaching. In such analyses,
normative and descriptive passages therefore intertwine, but
the problem is always more or less clear: How can
presumptively neutral descriptions, i.e., the analyses, be used
to ensure high quality teacher-controlled teaching?
  In this context, it is vital to bear in mind that control has
mainly been measured as a quantitative factor, with time use
during classroom teaching as the primary unit of
measurement. Time use also plays a part in studies that focus
on teaching plans (the teachers' plans as opposed to the plans
set out in the textbooks/ teachers' manuals), the selection and
sequencing of material (teachers' as opposed to textbooks'),
method of reading (controlled by teacher/group/individual),
type of textbook and subject (the three last groups have
rarely been investigated).
  Accordingly, the material associated with the textbook
texts has largely consisted of classroom observations based on
taking notes and/or completing forms, personal interviews
and questionnaires. In total and in relative terms, teachers are
more strongly represented than pupils in the interview and
questionnaire material. All the investigations seem to be based
on the generally implicit assumption that the teacher is the
principal determinant of what the books do for the pupils or
what the pupils do with the books. If so, this must be related
to the fact that almost all the studies use time as their unit of
measurement. The teacher decides how classroom time is to
be organized. Further along in the system, on the other hand,
the teacher has no say about the curricula or examination - an
examination where the books may threaten to eclipse the
control of the teacher in terms of work and content.
  Larger scale investigations, such as DsU 1980:4, Reints-
Lagerweij 1989 and Sigurgeirsson 1990, use all the above-
mentioned types of material. Each of them also uses a
combination of "hermeneutic", "quantitative" and
"qualitative" methods (see page 139).
  This may be exemplified by the Swedish investigation. One
of the main tables (DsU 1980:4, pp. 154-55) distinguishes
between five types of control effects (see page 163). Each of
the groups is ranked according to a four-step scale: Strong
 influence/tendency toward influence/weak, irregular
influence/very weak influence. These degrees, which were
given for seven different subjects, are based on very precise
calculations which make it possible to determine percentages.
Moreover, they include the researchers' evaluations of
interviews and talks. To some extent, however, assessments
made initially preclude attaching any importance to later
observations. The table covering the distribution of working
methods (class instruction, group work and individual work)
indicates that teacher-centered teaching dominates. But there
is no discussion of the degree to which this is related to the
textbook dominance otherwise pointed out by the
investigation because it is assumed that no such relationship
exists.
  There is a conflict inherent in this, as in the other
investigations that combine measurements and interviews,
involving the relationship between what teachers say they
think/do, on the one hand, and what they do, on the other.
Apart from Bromsjö 1965 and Lundgren 1969, who both
point out that classroom time may turn out differently than
expected, the investigations appear to assume a high degree
of concordance between statements about teaching plans and
the teaching itself. If such concordance actually exists, it
would indicate that researchers should use less time on
resource-intensive classroom observation and more time on
talks and interview analyses.
  The DsU investigation was based on notes taken in the
classroom. The idea was originally to complete the notes with
tape recordings, as was done in Sigurgeirsson 1990. The goal
was to supplement the early experience gained from such
recordings. After a test run using only notes, the research
group came to the conclusion that this method in itself was
sufficient. The observers felt convinced that their presence in
the classroom did not influence the teaching. In addition, the
group gave up the idea of focusing on statistical analyses and
focused instead on the style and content of the reports from
individual lessons (217 lessons adding up to 8,680 minutes).
Instead of attempting to produce quasi-scientific
documentation, they decided to describe their own individual
reports, which were subsequently computerized, in a way that
might be called semi-documentary:

  The form of qualitative report we decided upon is mainly
  of a descriptive nature. Based on the notes jotted down
  during observations, we've produced a brief, verbal
  description of the progression of each and every lesson we
  observed. The focus of these qualitative descriptions has
  not been so much on teaching materials as on describing
  the context in which the various teaching materials are
  used. Without overstating our case by using too many
  evaluations, we have intentionally tried to maintain a
  personal style in these reports. (DsU 1980:4, pp. 21-22.)

Sigurgeirsson 1990 observed more than four times as many
classes as DsU 1980:4 and conducted a far greater number of
interviews (official and unofficial), in addition to making
tape recordings and using a special analytical scheme. Despite
 having far more documentation than DsU 1980:4 and Reints -
 Lagerweij 1989, Sigurgeirsson found less variation between
teachers and classes, and from subject to subject.
Sigurgeirsson had, and used, completely different ways of
processing his data than the researchers in the Swedish
investigation, but his analytical scheme was more than 15
years old (The Sussex Scheme for Curriculum Analysis; Eraut
1975). At the same time, a study conducted by Ball and
Feiman-Nemser (see page 311) shows that more
impressionistic procedures are used even when technical
equipment is available: Just six women, teachers selected by
age, sex, skin color and academic background, took part in
the study based on notes and surveys. On the other hand, they
were monitored for a period of two years.
  As mentioned by way of introduction, investigations on
textbook dominance are mainly related to how teachers use
the books in planning and teaching. All investigations stress
that practices differ considerably from teacher to teacher. An
attempt has been made to systematize such differences using
a very old form. It dates from 1950 (Herrick 1950), but it was
the object of more far-reaching theoretical discussion by Lee
J. Cronbach (Cronbach 1955) and later used for more recent
research (Woodward and Elliott 1990). Cronbach distinguishes
between three main levels of lesson planning. Teachers at
level III readily and immediately accept the textbook's
presentation of a subject; they use material from the textbook
and the workbook, and they let their year's teaching schedule
follow the book. Teachers at level I discuss the educational
views inherent in the curriculum with colleagues, they draw
on the local community and pupils for material, and they plan
their year in cooperation with their colleagues and pupils.
Teachers at level II are somewhat less book-oriented than
those at level III, but less progressive than those at level I.
The assumption that these variations exist, if not as clearly as
outlined here then at least clearly enough for each level to
encompass a significant number of teachers, may or should
influence every investigation and evaluation of textbook
control. Logically speaking, traditional teachers' roles should
thus suggest that textbooks control when the teacher wants
them to control.
  However, such a pattern also raises the question of what
group(s) textbooks are actually written for, insofar as
publishers and authors have such an (approximated) pattern
in mind. With the exception of Zahoric 1991, there are no
pure textbook analyses that discuss the control power of
different types of textbooks in the hands of different types
of teachers. Nor do we know very much about the position of
textbooks in pupils' work or in their consciousness. Beverton
1986 identified considerable differences from subject to
subject for seventh graders (see page 199). An investigation
conducted by Sosniak and Perlman confirms Beverton's
findings. Sosniak and Perlman recorded interviews with 24
and 25 high school students in history and
English/mathematics, respectively. They held informal
interviews toward the end of the school year. The students
 were guaranteed anonymity, and they knew that the teachers
would never have access to the material. The authors found
that most of the teaching was based on textbooks, but that the
extent of use as well as the attitudes of the student varied
from subject to subject and to some extent from class to class:
"The data we have collected tell a variety of stories about
high school students' experiences with the core academic
disciplines" (p. 428). They point out that students represent
a virtually untapped source:

  Observing classroom activity without attending to the
  interpretations of the participants for whom those activities
  have been structured may well result in missing, or
  misinterpreting, important elements of practice. Students
  have been sorely neglected as a source of information
  about the practice of schooling in the US. It behoves us to
  become more respectful of and responsive to their
  perceptions of their experiences, perceptions which we as
  adults may be unable to identify or understand without
  students' help. (Sosniak - Perlman 1990, p. 440.)

The absence of pupil reactions in the investigation of the
books' control effect implies more than just a narrowing of
perspective. It may also be seen as symptomatic of the total
absence of a methodical theory in the field of control. One
reason for this is that non-physical book dominance is
extremely difficult to gauge. It is also probable that there is
a two-way cause and effect relationship between little theory
and few investigations. Yet this does not explain the lack of
interest in what the readers think of textbooks; in how willing
they are to let themselves be taught and educated by the texts
incorporated there.
  One rare exception is Lars Sigfred Evensen's doctoral
dissertation on the way in which teachers and pupils
experience problems in language and literature courses at the
lower and upper secondary level (Evensen 1986). Using a
very comprehensive polling apparatus and highly complex
statistical processing systems, the author measures views on
and attitudes toward the teaching situation as a whole. Pupils'
opinions of their textbooks in English and Norwegian is part
of the investigation. The pupils' answers to a number of
questions concerning their opinions on the books allowed the
author to identify certain tendencies:

  The teachers and pupils in both subjects gave teaching
  materials a slightly positive rating. Nonetheless, Chapter 8
  reveals several problems. A majority of the pupils felt
  there was a great deal of dry, boring material in the books
  and that the exercises were monotonous; they wanted to see
  a more pronounced pupil profile that would encourage
  more interest, e.g., by means of more creative exercises.
    The lack of creative exercises is also clear in the
  teachers' materials. In this context the problem of
   differentiation is also central, especially in the materials
  used to teach the subject of Norwegian. Viewed as a whole,
  however, the results on teaching materials indicate that the
  "interest" dimension is the most significant problem.
  (Evensen 1986, p. 435.)

Insofar as homework is assigned and followed up, textbooks
will also occupy an important position outside the classroom.
Depending on current trends and the subject itself, publishers
tend to shift back and forth a bit in the way they adapt their
textbooks for home use. It is remarkable that textbook use in
individual user situations - at home or, for that matter, at
school - has failed to capture the interest of researchers who
have studied the books' functions. Granted, language and
accessibility have been studied, but very few investigations
have focused on the school-home dichotomy. One exception
is T. Sticht, who wrote an article called "Understanding
Readers and Their Use of Texts" (Sticht 1985) in which he
claims that texts function in entirely different ways in the
two use situations, meaning that they call for entirely
different skills in order to be understood and utilized.
 

Accessibility
 

The numerous articles and essays criticizing textbooks have
been and continue to be dominated by two main schools of
thought. One involves exposing erroneous information and
ideological biases. The other involves language and style;
presentations are described as heavy, difficult, bad and
boring. But the two areas should not be regarded as mutually
exclusive. Erroneous information and ideological biases may
be a part of and a result of poor language as well as of wrong
facts. Conversely, it isn't very productive to uncover biases
in texts if in turn the same texts lose influence due to
linguistic inaccessibility. (By these examples I do not imply
that poor language or wrong facts are necessarily the main or
only sources for ideological bias.)
  The majority of the investigations in this section are book
analyses which have no parallel in use investigations. They
are primarily based on other research results, not least from
readability research and learning theory. Most of the
investigations deal with language and to some extent also with
the selection and scope of the material included in the texts.
Some also discuss or deal exclusively with arrangement and
lay-out, methods, illustrations and extra-textual
enhancements.

Vocabulary and Syntax
Non-empirical articles and essays have not been included in
the following survey, which is arranged so that the earlier the
investigation is mentioned, the narrower the scope of the
problem. Investigations that take a broad approach are
mentioned last. This system also ensures a certain chronology
of presentation.
  One of the first vocabulary studies was conducted in 1941
by Arnulv Sudmann (Sudmann 1978). He tested the word
comprehension of 886 children in grades 5 to 7 in Oslo. The
material included 20 words such as
metal/colony/trust/statesman, all taken from a chapter about
antiquity in the children's history book. The tests were
performed after the material had been read and reviewed in
class. The children were given every opportunity: They were
not required to define the words, just to use them in
sentences. They were drilled in the method ahead of time.
Less than half the children proved that they had understood
as many as 50 per cent of the words.
  Kathrine Simonsen broke new ground when she conducted
a comparable investigation in upper secondary schools in
Norway (Simonsen 1947). She claimed that graduates (age 19-
20) mastered little more than half the more uncommon words
 used in the textbooks at the 1st year level (ages 16-17).
  This investigation also dealt with the words in isolation.
One frequently criticized problem with such investigations is
that the words are taken out of context, another is that there
is a difference between passive and active word
comprehension. One of the pioneers in this area, Norwegian
educator Helga Eng, addressed these problems:

  We see that children (grades 3 to 7) have very little ability
  to form a correct picture of an abstract word taken out of
  context. (...) This result comes as no surprise. We tend to
  think it doesn't matter if children don't understand every
  single word, they will get the general idea from the
  context. The investigation showed that context helped
  children understand only a small fraction of the unknown
  or tenuously perceived concepts. (Eng 1912; quoted from
  Sudmann 1978, p. 28.)

Recent research appears to bear out Eng's views on the
limitations of context as an instrument for more or less
instant understanding of the wholly or partially new words
that arise in a context.
  In recent years the Center for Science and Mathematics
Education at the University of Oslo has worked with the
language used in science textbooks. Their efforts have been
based on questions and answers such as these: "Does the word
acid mean anything to upper secondary school students? Most
of them know the definition of acid, "a substance that gives
off protons", but the term carries few connotations for them.
They don't associate many properties with acids. In my
opinion, they haven't really gained an understanding of the
term acid." (Ringnes 1986, p. 15.) One of the most thorough
analyses of concept learning, Dittmar Graf's study of the
subject of biology and biology books at the lower secondary
level, confirms Ringnes' opinion of existing textbooks'
limited potential to serve as conveyors of conceptual
comprehension (Graf 1989).
  A number of English-language investigations from the
1970s and 1980s also concluded that what has been assumed
to be a simple vocabulary, well-adapted to the relevant age
group, may have just as profound an effect on complexity as
sentence structure does. Harold Rosen posed the same critical
question as Vivi Ringnes (see above) about words such as
revolution, therefore and correlation in the history books
(Rosen 1972). Larger-scale British investigations which point
in the same direction are "The Readability of School Biology
Texts" (Gould 1977), Readability in the Classroom (Harrison
1980) and Learning from the Written Word (Lunzer 1984).
  Studies of textbook language have developed both in depth
and breadth in recent decades. The most portentous change
evolved in the late 1980s in the guise of a systematic
discussion of methods and criteria for language-based
textbook evaluation (Davies 1986; see page 219). Further, it
became increasingly unusual to treat vocabulary and syntax
as separate entities. Many people have also finally recognized
the necessity for applying different models to different
subjects, a situation proven by a comparative analysis of the
 language used in textbooks in geography, biology and geology
conducted by linguists at the École Normale in Lille (Darras -
 Delcambre 1986).
  A new attitude is evident already in the title of a collection
of articles edited by Michael Marland, Language Across the
Curriculum. In his contribution, Marland attributes much of
the "problem at school" to the problem of text:

  The first problem at school is the sheer quantity of new
  words and new ways of putting things. As the pupil gets
  older this accelerates - indeed some attempts to lighten the
  load for the immediate post-primary years have the effect
  of making the mid-secondary language incline so steep as
  to be unsurmountable by many. Either way the vocabulary
  expectations of the new specialized curriculum are very
  heavy.
  (...)
  The reading material in different subjects is startlingly
  different, each subject from the other and all from fiction.
  It is not only the vocabularies that are different. The
  predominant paragraph patterns are quite different.
  (Marland 1982, p. 74.)

This expansion of perspective can be illustrated by
investigations conducted in one particular subject. For a
number of reasons, mathematics is an obvious choice. The
subject has a well-developed sign and symbol language that
does not preclude the use of prose texts. Still, the amount of
text has been modest enough to make mathematics easier to
study than most other subjects. The textbooks underwent
fundamental changes in the 1970s and 1980s. The use of
consecutive text increased steadily. Consequently, several
researchers wondered whether declining mathematics results
could be due as much to poor reading skills/poor texts as to
a real decline in mathematical skills. That question has also
been posed in a way that is relevant to all school subjects:
What proportion of math instruction is really language
instruction?
  A report called Language and Vocabulary in Mathematics
in Sweden (Björnsson - Dahlkvist - Edstam - Jarne 1967) was
published in 1967. Consisting of three sections, it describes
the vocabulary used in math books, gives an account of
pupils' comprehension of that vocabulary, and discusses the
issue of how the language used in math problems affects
pupils' ability to solve them. The material consisted of three
popular textbooks used by third graders. The vocabulary
descriptions show wide variation in the amount of text, which
accounts for a significantly lower percentage of one of the
books than of the other two. The study also deals with other
variables such as the ratio of separate words to the total
number of words, the percentage of mathematical terms used
in all the books and in each book separately, the ratio of
words that occur only once (a figure as high as one to four
for mathematical terms), high frequency words (where the
percentage of repetitions is so high that researchers describe
the language in one of the books as "very monotonous" (p.
133)), a comparison of the vocabulary used in the math books
 with the vocabulary used in other children's literature (where
the distinction was especially prominent for pronouns,
conjunctions and negations, which occurred far more
frequently in other literature than in textbooks).
  Finally, the vocabulary was described within the context of
the sentence, using a method that in many ways experienced
its breakthrough with this investigation: The readability index
(LIX), which is based on the percentage of words more than
six letters long (long words) and average sentence length.
Generally speaking, the difficulty of a text is directly
proportional to its LIX value. The books in question scored
27, 27 and 30. The last figure applied to the book with the
least text per problem, the briefest explanations, many single
occurrences of words and the fewest high frequency words.
Researchers found these investigations telling enough to
characterize the textbook in question as not as good as the
two others.
  One important feature of the Swedish investigation is that
the description of the vocabulary was approached from
several different angles, all of which were explained and
tested prior to the main investigation. Another important
feature was the follow up, which encompassed two other
equally comprehensive investigations; one attempted to test
pupils' comprehension of the texts in the same math books,
while the other studied the possible correlations between the
language used in the math problems and pupils' ability to
solve them. In very general terms, the comprehension test
instructed pupils to underline any words they did not
immediately understand. The test was adjusted against
comparable word list tests. Approximately 400 different
words in the books were classified as difficult based on the
criterion that four or more pupils in each class did not
understand them. The vast majority of the words were long,
many (more than one-third) were mathematical terms, and
most of them were abstract. In addition to those 400 words,
900 others were not understood by two or more pupils per
class. One observation that contradicts the view of Helga Eng
(see page 186), was that the pupils in the control group with
(the same) words out of context underlined some 50 per cent
more words. This could indicate that the presence of a
meaningful context eases word comprehension considerably.
However, in light of the more than 1300 "difficult"" words
in the textbooks, the researchers felt justified in deeming the
language used in the textbooks unsatis-factory.
  The last investigation (concerning the relationship between
language and problem-solving) was a "multifaceted
problem" (p. 137). A severely limited procedure was selected:
The test involved two sets of problems in which the problems
and answers were the same, but the text in one set had been
re-written/improved by the researchers themselves to
conform to the experience gained from the first two
investigations. The results may be summed up as follows; 1)
The re-worked version resulted in a larger number of correct
answers, 2) The improvement was notable at every skills
level, but especially among the slowest learners, 3) Changing
text relating to mathematical terms resulted in relatively
 greater improvement than other types of linguistic
intervention. In this investigation, too, pupils were asked to
underline the words they didn't understand. The test
consisted of more than 30 problems, all of which were the
most frequently underlined in the original version. The pupils
marked 669 words in the original texts, but only 275 in the
re-written texts.
  One remarkable aspect of this investigation is its
exceptional thoroughness, which provides immediately
applicable results in a limited field. Another and at least
equally remarkable trait is that the investigation was not
concluded with the writing of the report. It was designed for
teachers and was arranged so that the investigations could be
supplemented and elaborated by individual teachers and by
schools.
  Nearly 20 years later Andrew Rothery wrote an article
called "Readability in Maths" (Rothery 1986). He looks at
the status of readability in light of research and textbook
development in the 1970s and 1980s. He refers to series of
investigations that confirm the results of the Swedish
investigation, e.g., the correlation between vocabulary and
math comprehension. A.R. Nicholson found that pupils at the
lower secondary level all understood the word "multiply",
while only one in five understood the word "product", used
in the mathematical sense (Nicholson 1977). Rothery has
included a summary of readability testing methods in
Children Reading Mathematics (Rothery 1984). The most
familiar method in the USA is the so-called cloze procedure,
which was introduced by W.L. Taylor in 1953 and
subsequently developed by others. The main principle is to
leave out individual words and mathematical symbols in a
text. The reader is instructed to fill in the blanks; generally,
a score of 40 - 60 per cent indicates satisfactory accessibility.
Americans developed the system further, until it became
strongly reminiscent of the Björnsson study. Kane, Byrne and
Hater (Kane 1974) emphasized the distinction between three
classes of words: Ordinary words, mathematical words and
mathematical symbols. Based on this, they designed a
readability formula, which, by the way, Rothery finds
unsuited for the subject of mathematics as taught in Great
Britain. Rothery is generally very skeptical about such
formulas:

  Perhaps the greatest inadequacy of all such formulas is that
  in their attempt to be widely applicable, they lose the
  ability to help with detailed professional problems. For
  instance, if a teacher is to help pupils in the use of a
  particular text, it is important to judge its vocabulary in
  terms of words which are familiar to those particular
  pupils. (...) In fact a sensitivity to the issues involved would
  be more helpful than the knowledge of a readability score
  for a passage. (...) Clearly some significant factors affecting
  readability are such that they cannot be readily qualified.
  (Rothery 1986, p. 126.)
 Such skepticism does not imply that Rothery rejects formulas.
He summarizes his article by pointing out the importance of
certain research results derived from readability studies. In
the field of vocabulary skills, for instance, this applies to the
difference between ordinary words and mathematical words,
but it also applies to the difference between words used in
the ordinary sense and the same words used in a different
way, a way which may be inaccessible within the context of
the subject. In the field of syntax, there is growing
acceptance of the theory that sentence length per se is no
criterion for accessibility; long sentences can be easier to read
than short ones (Perera 1980, Harvey 1982). As regards
mathematical symbols, it comes down to things as common,
but little researched, as the ratio between the concentration
of the expression and readability, or "spatial structure" (e.g.,
the difference between a2 and 2a). As regards extra-textual
enhancements, it concerns the use of lay-out, examples,
tables and illustrations. Rothery compares the presentation of
the same topic in two competing textbooks published the
same year. The language is obviously more clear and readily
accessible in Book A. On the other hand, Book B does a better
job of illustrating the phenomenon in question.
  Like the researchers in the Swedish mathematics
investigation, Rothery does not see the systems for measuring
accessibility as exhaustive, but as a foundation for gaining
insight that must be put into practice. Teachers who strive for
"perfection" must act as researchers in relation to textbooks
as well as to pupils.
  The Institut für Schulbuchforschung was founded in
Vienna in 1988. This was largely thanks to Richard
Bamberger, who until that time had been doing readability
research on children's and young people's literature. In
collaboration with Erich Vanecek, Bamberger has written an
analysis of the problem of readability that provides a
thorough introduction to the measurement systems developed
in the USA and Europe (Bamberger - Vanecek 1988). In the
early 1980s, Bamberger and Vanecek investigated readability
in primary and lower secondary level textbooks in several
different subjects. They developed and used more than a
dozen formulas which took a large number of different
linguistic factors into account - including those beyond the
word and sentence level - factors that were tested in roughly
500 texts for several different subjects and at several
different levels from the fourth through the twelfth grades.
The measurements were summed up as follows: The degree of
linguistic difficulty in the books was often one or two levels
higher than the grade for which the books were intended. The
discrepancy was especially pronounced at the fifth through
the eighth grade levels. Based on their readability quotients,
many books intended for use in the fifth grade belonged in
the eighth grade.
  These results stirred up a public debate, and the formula
model was therefore expanded further. A new round of tests
included factors such as print face, typography, information
 density, reading speed, reading situation and motivation. The
result showed such a strong degree of correlation with the
first investigation that the researchers felt they had shown
that linguistic analysis in itself was a reliable measure of
accessibility (Bamberger - Vanecek 1984). Almost
concurrently, East German researchers presented
measurements of reading difficulty that entailed comparable
criticism of books in the GDR (Starke 1983).
  Sylvia Danielson studied the language used in a total of 15
textbooks in the subjects of physics, social studies and
Swedish (Danielson 1975). The books were used at the lower
and upper secondary levels. Altogether a relatively random
sample was selected, including 19 lengthy texts from the 15
books. An analysis was made of 400 sentences and 4000 words
in each text.
  Danielson had certain reservations about traditional
readability tests. The most important of them may be summed
up as follows: Repetitiousness makes much of the material
that cannot be understood out of context understandable
afterall. Analyses of "incomprehensible" sentences in a
paragraph may overlook earlier explanations that would have
provided sufficient information for accessibility. The testing
situation (testing the pupils) entails such great practical
limitations that the linguistic factors "measured" do not
occur frequently enough in the sample. It is not feasible to
judge the effect of one linguistic factor in relation to other
linguistic factors on the basis of non-fabricated texts. Strictly
speaking, this can only be measured reliably using specially
designed texts in which one factor can be altered, while the
others remain constant.
  Since Danielson's material consisted of printed texts, she
chose another approach, i.e., describing the similarities and
differences between the texts with a view to topic and
educational level. Within this framework, she mainly limited
herself to nouns for studying vocabulary and to passages with
just one sentence (the dominant type of sentence in the
books) to study syntax. At the word level, the investigation
focused chiefly on abstracts and certain types of compounds.
In the area of syntax, the study looked at sentence type and
structure in addition to sentence length. As far as sentences
alone are concerned, Danielson refers to one of the main
views propounded by Schlesinger 1968 and Platzack 1974:
Long sentences, or the intertwining of several (long)
sentences in a passage, are not in themselves sufficient
grounds for proclaiming poor accessibility. On the contrary:

  (...) sentences that are too short can have an adverse effect
  on readability. (...) Among the other texts that I examined
  in the initial phase of the investigation, I found one
  textbook in materials science for the first year of upper
  secondary school which averaged 11 words per sentence. In
  certain paragraphs the average sentence is 9 words long,
  i.e., the average sentence length which, according to
  Platzack's results, is relatively more difficult to read than
  texts with longer average sentence lengths.
   (...)
  If sentence length in itself is not as important as many
  would like to think, then it is unfortunate if accepted
  beliefs about the importance of sentence length entice
  authors to believe that their problems can be solved by
  using shorter sentences, rather than by looking at content.
  (Danielson 1975, pp. 130-31.)

There is reason to emphasize that Danielson's study neither
shakes nor is intended to shake the position of the large body
of documentation produced by readability researchers in the
1960s and 1970s that established once and for all that
formulas such as lix, fog and cloze are well suited to calculate
the degree of readability. It is also clear that the methods
applied in empirical, data-based linguistics, for instance, can
provide detailed descriptions covering not only vocabulary
and syntax, but also morphology (Schuyler 1982, Mikk 1990).
The more linguistic aspects that can be revealed in that way,
the more questions will arise. Nonetheless, one question will
probably remain unanswered no matter how systematic such
descriptions become. Danielson's concluding discussion
contends that such descriptions of existing, printed text in use
are limited by not saying enough about the reasons why a
particular word or sentence earns a particular readability
quotient. To sum up the investigations of Danielson and
others, it might thus be said that measurements of vocabulary
as well as syntax are measurements whose results indirectly
reveal the need to examine these two factors in connection
with other text elements. Such a change in perspective
occurred in the 1980s.
  One of the first to address the problem was linguist and
educator Harold Rosen of the University of London. He
stressed that while a number of models had been developed
to deal with linguistic analyses of textbook language, there
was no analytical instrument for "the psychology of
understanding and using it. We need a dimension which
embraces both the linguistic and the psychological
components. Until we know more about it let us call it the
personal-impersonal dimension." (Rosen 1967, p. 104.) While
investigations made prior to the 1980s were rather
linguistically-oriented, and frequently limited to elements
such as vocabulary or syntax, linguists and reading experts
gradually grew more interested in Rosen's personal-
impersonal dimension. Two publications represent milestones
in this area: Gillham 1986 and de Castell - Luke - Luke 1989;
see page 199 and page 84.

Style and Structure
In the preceding section, the term syntax was mainly linked
to investigations that focused on sentence length and/or the
number of sentences, in a passage for example. Formulas such
as LIX flourished in the 1970s, only to lose ground later.
Some researchers have gone as far as to contend that the only
practical result achieved was to "lix" textbook texts into
showing a relatively uniform level of inaccessibility
 (Johansson 1988, pp. 22-26).
  A more comprehensive text study would also include
analyses of textual cohesion, within each sentence, from
sentence to sentence, from passage to passage, from
paragraph to paragraph and among the larger units which
combine to form the final text.
  Such a method would expand the parameters for evaluating
the language used in textbook texts, but it would also increase
the degree of uncertainty regarding approach and
measurement. Some of the most pronounced problems in this
context are the text concept, the scientific ideal, and the
relationship between form and content in that special
educational context in which the textbook belongs. The
following survey of investigations will demonstrate how, in
different ways, these studies have addressed such problems
by selecting approaches which are sometimes narrow, and
sometimes very wide indeed.

New Trends
The 1970s saw the publication of a number of new books in
"knowledge subjects" (history, geography, social science and
natural science) in Norway. Several of the books proved too
difficult for many pupils. This led some publishing houses to
compile simplified versions, meaning there were two versions
of the same book sets on the market, an A set and a B set.
Anne Hvenekilde compared two such sets that were marketed
by a major publishing house (Hvenekilde 1983). She used
traditional methods to measure vocabulary and syntax, but
she also analyzed the conjunctional and adverbial
characteristics of the texts. Although the revision of the A
texts consisted in length reduction, it also involved language
revision. Hvenekilde showed the result was often a more
difficult text because the well-planned cause and effect logic
of the original text was destroyed and not adequately replaced
in the B-sets.
  While working on a project involving textbooks in
Norwegian for foreigners, Hvenekilde discovered that the
authors had gone so far in their quest for simplicity that they
arrived more or less right back where they started - with an
inaccessible text. She therefore went on to study 4th and 5th
grade knowledge subject textbooks written for children who
have Norwegian as their native language (Hvenekilde 1986).
She studied only one set of books, but it was a set that had
dominated the market for several decades. First, Hvenekilde
proved that the paragraphs were organized and connected to
one another in a way that often impeded understanding
through context - inter alia by repeatedly mixing sentences
and lines as textual units. Second, she found numerous places
in the books where it was unclear whether the reader was
being confronted with texts containing facts (reference texts)
or narrative texts; in this context, she discovered
"simplifications" that entailed breaks in logical text
structures. Third, she found a number of contraventions of
conventions related to information structure (for example,
unmotivated switches between old and new information).
  In conclusion, Hvenekilde makes a comment which is very
 apt for the investigations surveyed in this section:

  However, analyses such as this one can only form a point
  of departure for hypotheses concerning what pupils find
  easy and what they find difficult. To find out for sure, we
  need empirical investigations to study how the various
  features of the texts affect pupils' understanding and
  acquisition of knowledge. (Hvenekilde 1986, p. 25.)

This quotation is reminiscent of a factor which was indirectly
addressed earlier in this chapter, namely that researchers have
concentrated mainly on the books' texts - and not on the
readers' use of them. The following may be interpreted as a
rebuttal to that tradition:

  The fact is, we have little understanding of the effects of
  a uniform, ubiquitous application of readability criteria,
  for what students find difficult to read is influenced by
  what they have read and what they have been taught. It is
  meaningless to say a text is difficult; it is difficult for
  someone. Whether a text is difficult depends on the reader
  as well as upon the text. (MacGinitie 1981, p. 285.)

The above quotation is taken from a book based on a
conference held in New York in 1981: Learning to read in
American schools: Basal readers and content texts (Anderson
- Osborn - Tierney 1981). As mentioned earlier, research into
the linguistic accessibility of textbooks has traditionally been
dominated by reading teachers and readability researchers
who have mainly studied the materials used to teach reading
in grades one through nine. The New York conference, which
was attended by the foremost readability experts in the USA,
may be viewed as part of that tradition. At the same time,
however, it represented a departure from the tradition in that
it heralded new directions. This applied to three areas in
particular: The first was the disparity between reading ability,
the content of readers and books in other subjects (knowledge
subjects and workbooks in general), and curricula. The
second involved the constraints in and misuse of readability
investigations ("Davison, Green, and MacGinitie presented
rather convincing evidence for avoiding the use of readability
based and contextually sensitive means for testing texts."
(Tierney 1981, p. 560.)) The third stemmed from the criticism
in the second area and concerned the deficiencies inherent in
the narrative technique and the authorial role in the growing
number of cases in which publishers and authors had tried to
break old patterns.
  One of the main points made at the conference was that the
problem of reading and text comprehension must be removed
from the domain of the reading teacher and put into the
domain of all teachers in all subjects, and that this must be
done on the teachers', pupils' and relevant subjects' own
terms. Exciting stories, glossaries containing many detailed
explanations and workbooks with differentiated batteries of
exercises are not enough in any subject:
   The programs' reliance on story context and independent
  use of the glossary as methods of vocabulary development
  are at best appropriate only for the most motivated and
  competent readers. Children most in need of vocabulary
  development, the less skilled readers who are unlikely to
  add to their vocabulary from outside sources, will receive
  little benefit from such indirect opportunities. (Beck 1981,
  p. 19.)

Two of the investigations included in the conference report
stand out in this context. Jean Osborn presented the first,
truly comprehensive analysis of the workbook genre since St.
Lawrence 1952 (page 177) based on documentation which
revealed such widespread use of workbooks in some subjects
that their importance vied with that of the basic textbooks.
Osborn reviewed several hundred assignments in 20
workbooks at different levels in different subjects. As a
readability researcher, she concluded that "workbooks are the
forgotten children of basal programs" (Osborn 1981, p. 130).
Osborn subsequently followed up this work, inter alia through
a recent investigation on what pupils learn from the
workbooks (Osborn - Decker 1990, see page 229).
  At the same conference, Thomas Anderson and Bonnie
Armbruster presented the results of an investigation which
encompassed textbook texts and pupils' work with them. The
researchers sought to reveal aspects of the text which
demonstrated "comprehension and learning differences"
(Anderson - Armbruster 1981, p. 377), with special emphasis
on textual structures which are common in the subject in
question. This project was also followed up in the course of
the 1980s; see page 209.
  The report from the New York conference is an important
manifestation of textbook (language) researchers' propensity
for studying more holistic textual structures. Another
manifestation is a composite work entitled The Language of
School Subjects (Gillham 1986), which is an extension of the
viewpoints in the New York report, corroborated by more
recent British research. The entire book is based on the
attitude expressed in the contribution by Andrew Rothery
(see page 191): Educational text research is also the
responsibility of the teacher and part of the instruction. The
contributions are all written with that in mind. They consist
of research reports adapted for further processing.
  One of the key results and viewpoints in the book is that
teachers and pupils have completely different ideas of what
it means to read a text. There is also a qualitative distinction
made between two groups of pupils (the "purposeless
approach" and the "metacognitive approach"). This was
demonstrated by Marian Tonjes, who works in the USA and
is the only contributor from outside Great Britain. She
stresses the difference between learning to read and reading
to learn. The approach to the two activities varies, both when
authors write and when pupils read the textbook texts (Tonjes
1986, p. 69). Teachers are generally not aware enough of this
in individual subjects, and that represents quite a problem
inasmuch as it was long since proven that each subject has its
own special textual pattern and that such patterns, in turn,
 call for special reading strategies (Smith 1964).
  Sue Beverton reports on a "reader-centered", i.e., not
"text-centered", investigation of the status of textbook
reading and pupils' motivation (Beverton 1986). One hundred
and sixteen seventh graders, all of whom attended the same
school, kept logs of what and how they read for a period of
five weeks. Most of the subjects were textbook-centered:
English, geography, history, mathematics and natural science.
Altogether, the pupils' reading material comprised 108
different titles, 72 of which in the subject of English, and six
of which in mathematics. The grade level is at the crux of
this study. Beverton discerned a general attitude among the
pupils which coincided with an attitude noted in other
investigations (Lunzer - Gardner 1979, Richmond 1979,
Heather 1981, Ingham 1982): "The fun goes out of reading
(...) between primary and secondary schools" (p. 35).
Beverton believes this is related to the textbooks' change in
character at this level and to differences in school reading
policies which are manifested at the same level. Most of the
reading in primary school was fiction, and it was often a
group activity. The situation was rapidly reversed in lower
secondary school, where emphasis was placed on non-fiction
- then too as the most common form of "whole-class-work"
(p. 36). At the same time, the reading of fiction at this level
gradually became less a class activity and more an individual
activity, often tied to the school library. The total number of
titles read, fiction and non-fiction alike, was clearly reduced.
  If it is correct that the status of (textbook) reading declines
as the educational level rises, and that the critical point lies
in the transition between primary and lower secondary
school, then this, on the basis of Beverton's observations,
should offer a starting point for interdisciplinary
investigations and discussions of genres and textbook
language in all subjects. Mother-tongue instruction is
especially interesting in this context. Eighteen of 72 titles in
English were textbooks, the rest were fiction books and
supplementary reading from the school library. The pupils
described English as a subject without coherence; the many
textbooks make English "a widely-spread, free-floating
subject; even if highly book-oriented, it is not book-
specific" (p. 28).
  In The Language of School Subjects, Katharine Perera
(Perera 1986) presents an analysis of the language in 25
paragraphs of approximately 100 words each from the genres
of textbooks and narratives, excerpted from books intended
to be used by pupils aged 9 - 13. She analyzes two levels:
"Discourse-level differences between fiction and non-
fiction" and "Sentence-level differences between fiction and
non-fiction", in the order listed here (i.e., the whole before
the part). Perera's book investigations confirm Beverton's
pupil survey. She demonstrates significant deviations, but she
 does not reject the transition to more fact-oriented reading -
or cogitative writing - at higher levels. On the other hand,
like Beverton, she advocates the need for radical changes in
the transition from the lower to the higher level. Such a
transition can only be effected by developing textbook prose,
reading routines and written work.
  Suzanne de Castell (de Castell 1990) points out one
significant condition which is both the reason for and the
potential answer to the problem of reading material transition
from primary to lower secondary school:

  (...) it is with expository rather than with literary texts that
  students have the greatest difficulty.
    School textbooks, however, are not primarily literary
  texts, and this is increasingly so as students move from the
  elementary to the secondary grades. So although there is
  much that is promising in current work on interpretation,
  the authority of the teacher, the author function, and the
  role of the reader, this work applies primarily to literary
  texts, whereas school textbooks are less often literary texts
  and more often what I shall henceforth call "fact-stating"
  texts. These require, it is argued here, a separate and
  different treatment. (P. 76.)

Beverton's and Perera's investigations both point out the
textbook situation for mother-tongue instruction as a
fundamental problem. In a paper delivered at the 15th
International Schoolbook Conference in Köthen in 1990,
linguist Hans-Wolfgang Lesch from the University of
Lüneberg said the same thing. He had analyzed 17 language
books in German for grades three to nine, published by seven
different publishing houses. Lesch confirmed that the
selection is now so large that a single publisher may publish
up to three different grammars for the same level.
  His analyses showed that grammar books differ appreciably
in terms of linguistics, educational philosophy and
pedagogics: "Using such different grammar books has fatal
consequences for pupils, especially if they change schools
because they move or advance to secondary school" (Lesch
1990, p. 1). Like Beverton and Perera, Lesch pointed out the
consequences this could have for reading and writing in
general. He drew the conclusion that either all members of
the teaching profession in Germany had to be molded into a
corps of linguists or that the new Ministry of Culture should,
at the earliest possible date, accord all teachers and pupils in
both East and West "the benefits of using just one school
grammar". (This last point falls outside the parameters of this
chapter, although the question of its practical consequences
is obvious when many results point in the same direction.)
  To illustrate new and broader approaches to linguistic
usage in textbooks, I have chosen to take a closer look at
studies related to the following three topics: The "reality" of
the text, its metadiscourse and the structure of causal
 explanations.

The "Reality" of the Text
The question of which "reality" textbooks represent or are
intended to represent is especially germane in the context of
grammars. Their content is not given or limited in the same
way as in textbooks in chemistry or geography, or in books
for foreign language instruction (the country and culture) or
in mother-tongue books in language history, literary history
or anthologies. Grammars deal with themselves insofar as the
examples they use are primarily intended to illustrate
linguistic structures. Consequently, a relative clause is
acceptable in the book to the degree that it illustrates the
typical or atypical characteristics of the sentences to be
explained in the paragraph in question. Theoretically, this is
the only absolute criterion, but grammar book authors' field
of vision is not quite so limited in practice. Nonetheless, the
selection and use of examples in school grammars have
followed certain conventions. If we ignore the older tradition
which was largely based on fictional role models, the
examples are to be simple and ordinary, preferably timely
and adapted to a level the pupils can understand and with
which they can identify. The fact that these conventions have
withstood the test of time may be because grammar is the one
discipline among all the primary and lower secondary school
disciplines which, despite considerable scientific
development, has remained most constant during the past 100
years. (The differences Lesch refers to in German grammars
(see page 201) primarily concern educational methods;
otherwise he confirms Chervel's thesis on the constancy of
the school subject (Chervel 1977). The same is true of
Fossestøl 1987, Huot 1988 and Hertzberg 1990.)
  The man who may have been the first to label such
conventions rigid and artificial was not a researcher, although
he apparently embarked on a brief career as a linguist before
suddenly changing his mind. On several occasions, author
Eugène Ionesco mentioned that the starting point for his
absurd drama was the mechanical remoteness and
contentlessness in school grammars. This connection is most
clearly expressed in the drama La Leçon (1951), in which
some of the teacher's monologues paraphrase vocabulary and
structures from the grammars. In contrast, there is one rare
example of deliberate employment of a quasi-literary solution
intended to draw attention to and discuss the artificiality of
such examples. Linguist Marianne Haslev published an
introduction to syntax in which the protagonists in all the
examples are animals (Haslev 1975).
  Åke Pettersson has studied the selection and use of example
words in popular school grammars (Pettersson 1987). The
material involved the use of nouns, adjectives and verbs in
three books for grades 4 and 6 in primary school. Pettersson
found 243 different forms of nouns. After eliminating
doubles and proper names, he was left with 170 common
nouns. These were compared with the vocabulary used in a
total of 955 essays written by 191 different pupils in grades
4 and 6. Eliminating doubles and proper names, that left 109
 common nouns. The comparison clearly revealed that the
vocabulary used in the grammars was more concrete than the
pupils' vocabularies. The pupils' language abounded with
abstracts, but there were few in the textbooks. The books'
repeated use of "childish" and familiar names for animals
and plants was especially striking. Words like tablecloth,
kitten, pillow, cake, leaves and violets occurred only in the
textbook material; so often, in fact, that Pettersson believes
many pupils will receive the impression that this sphere of
meaning is in itself a criterion for being a noun. While
temporal and causal expressions occurred with remarkable
frequency in the pupils' essays, they were seldom used in the
textbooks. Pettersson also compared the two groups with a
survey of the vocabulary used in adult newspaper language.
The scope of the nouns used by the pupils was far more
"adult" than the textbook language.
  With certain reservations due to the modest scope of his
investigation, Pettersson drew the following conclusions:
Pupils are not allowed to see how words and word classes are
parts of authentic texts; the perspective is that of the word
list. They do not see the connection with their own language
usage. The abstract/mental world of ideas is kept outside the
realm of pupils of this age. The grammars distance themselves
from the prose used outside the classroom.
  One might then ask about the extent to which the negative
observations made about grammars are applicable to the view
of the world presented in textbooks on other subjects. One
well-known example of distortion involves the math problems
used in arithmetic books for generations. Some of them
accidently foiled their own effectiveness as a tool: "If 11 men
dig a ditch for 4.5 days, how many men would it take to dig
the same ditch in 2 3/5 days?" Others turned the world
upside-down by providing the logical answer and asking for
a calculation that no one needed to calculate: "A clerk wrote
8 addresses in 5 minutes. How many addresses did he write in
1.5 hours?" Yet Grevholm, Nilsson and Bratt have
demonstrated that this tradition still exists, through other
varieties of unreality, in modern math books (Grevholm -
Nilsson - Bratt 1988, pp. 272-88).
  In theory, textbook authors may advocate vastly different
approaches to the question of how one should represent the
"world" and "life" in textbooks intended for particular age
levels. In light of objectives such as molding attitudes,
selection and individuality, which characterize the curricula
of many countries today, it should be legitimate to write a
chemistry textbook which is primarily an introduction to
environmental protection. Using pollution as the general
theme, an author could conceivably cover a considerable
share of chemistry's most important topics, while relating the
topics to the reader's everyday life.
  Granted, no books that have followed that recipe have met
with any measure of commercial success as yet. The most
prevalent approach today is still based on the tradition of the
broadest possible coverage of a subject, sub-divided into
topic units, like in the grammars. Moreover, today's textbook
authors still write texts intended to take account of grade
 level and neutrality. The majority of all textbook research
conducted thus far encompasses one or more levels - in terms
of subjects, grades and/or neutrality. Meantime, most of the
negative textbook criticism, whether in the wake of the
investigations or independent of them, charges that deference
to so many different factors makes the text inaccessible
because the language becomes "boring," "diluted" or
"dead" (see Evensen 1986, page 183). In this context, the
focus has generally been on the issue of neutrality, but
several investigations discussed in this chapter show that in
many cases educational adaptation has not had the intended
effect. Åke Pettersson's study even indicates that deliberate
attempts at educational adaptation may have an alienating
effect, since we know too little and consequently assume the
existence of children's language and a children's "world" -
for ten to twelve year olds, for example - which does not
exist at all. Of and by itself, adaptation to a particular level
could suddenly become just as controversial as ideology, for
example, were it not for the fact that the concept is probably
already closely related to ideology.

Metadiscourse
Investigations and criticism in the form of debate have had
ramifications. A great deal of public lip service has been paid
to striving for personality and certain profiles in textbooks
(Sanness 1987, California Board of Education 1988, Barth
1991). In this light, the question of the textbook author's own
presence, i.e., of the books' metadiscourse gains importance.
  Metadiscourse was discussed by Avon Crismore in an
article entitled "The Rhetoric of Textbooks: Metadiscourse"
(Crismore 1984). Crismore compared texts from nine
textbooks in social studies with nine non-textbook adult texts
in the same subjects (history, politics, geography and
economics). The textbooks ranged from third to twelfth grade
levels; the texts for adults were largely written for the general
public; they included articles and excerpts from books. The
starting point was "written authorial commentary
(metadiscourse)". Crismore shows that research into the voice
of the author has long traditions in literary research, but that
it has not been investigated in textbooks (see de Castell, page
201). Crismore wants to "create a typology of metadiscourse
based on the functions of language and rhetorical techniques,
and then examine how it is used by writers of American
Social Studies texts." (Crismore 1984, p. 282.)
  Crismore defines metadiscourse as part of many, but not
all, informative texts. There is a "contentless level" which
consists of the author's entry into the text; of his explicit or
implicit "discoursing about the discourse" (p. 280). This
might involve factors such as comments about the progress or
design of the text, personal views concerning the material,
 information about sources or direct petitions or appeals to the
reader. Although one could divide these factors into a number
of sub-categories, Crismore distinguishes between two main
groups. The one is "informational", the other "attitudinal".
The author subdivides the latter group into four sub-groups
which may be exemplified as follows: Even more important
than the reform was ... (assertion); This is an exaggeration, of
course, but ... (emphasis); Perhaps hunger was the worst ...
(reservation); Unfortunately, far too few turned out to vote
... (evaluation).
  The occurrences of metadiscourse were measured
automatically per 1,000-word text unit (beyond 1,000 words,
the text lengths varied all the way up to 12,000 words). On
the other hand, a review of the "presentation style and
patterns of use", as well as the direct, personal manifestation
of the author, called for the extensive use of quotations and
a qualitative analysis.
  One important, but rather unusual point concerning
Crismore's choice of material was the way he distinguished
between typical and atypical texts. Very popular textbook
texts were typical, while little-used ones were atypical. Eight
of the adult texts were typical in the sense that they were
popular and one was atypical in the sense that it was scholarly
and written for a special audience.
  Crismore discovered significant differences between
typical and atypical textbook texts. The former group used
very little formal metadiscourse such as discussion of the
book's objectives or an explanation of the plan and structure,
while such metadiscourse was very common in the latter
group (58 of a total of 82 such elements were found in two of
the three atypical textbook texts). There was strong
correlation between the atypical textbook texts and the
ordinary adult texts insofar as it was common to explain and
define the texts' plan of action, something which virtually
never occurred in the typical textbooks. Crismore found
similar patterns in summaries, or so-called "post-plans".
  Might one find the same correlation in attitudinal
metadiscourse? The investigation revealed that all four groups
(see above) were used far more frequently in adult texts than
in those intended for pupils. To the extent that such elements
occurred in the textbooks, they occurred primarily in the
atypical texts. One striking feature was that this type of text
far more frequently used these tactics to try to establish a
"we" relationship in the text. Crismore's conclusion is cited
here because it suggests an answer to the question about the
relationship between ordinary prose and textbook prose (see
page 217):

  Textbooks seem to use attitudinal metadiscourse to refer to
  concrete people or happenings in the primary discourse
  while non-textbooks use it to refer to abstract concepts as
  well as concrete phenomena. Another difference is the
  tendency of non-textbook writers to be present in text with
   a first person for expressing attitudinal metadiscourse
  while the textbooks prefer more distance and use second or
  third person. A third difference is the large amount of
  emphatics and hedges used by non-textbook writers to
  argue their points. The final difference is that textbook
  writers use simple evaluatives only (and very few of them)
  while non-textbook writers use both simple and complex
  evaluatives. (Crismore 1984, p. 295.)

Indirectly, this quotation raises the question of the
justification for leaving out or cutting down on the use - in
textbook prose - of a number of stylistic instruments that are
important in ordinary prose. According to Crismore, personal
argumentative sequences with reference to abstracts are
virtually never used in textbook prose. In other words,
Crismore's investigation of metadiscourse confirms Åke
Pettersson's discovery about the vocabulary of grammars: The
absence or lack of abstraction is greater in the textbooks'
"reality" than in the pupils' and/or the non-textbook
"reality". An investigation mentioned by Woodward - Elliot
- Nagel 1988 substantiates and puts some perspective on
Crismore's view:

  Starting with the premise that textbook prose is dull and
  boring, the authors asked three groups - text linguists,
  college composition instructors, and editors of Time-Life -
   to rewrite a passage on the Vietnam War taken from a
  grade 11 history textbook. High school students recalled 40
  % more of the passage rewritten by the Time-Life editors
  compared to minor gains for the other rewritten passages.
  (Graves - Slater 1986.)

Åke Pettersson's observations are limited to one area in one
discipline of one subject. One might ask whether the question
is not just as much one of type of language and type of
abstraction as one of abstract - less abstract - not abstract.
This last point was made by J.R. Martin of the University of
Sydney (Martin 1988). Martin studied and described history
and geography as two inherently separate linguistic
expressions at both the university and the textbook level. The
relationship between abstract academic discourse and
pedagogic discourse presents a problem because the latter
type of language, which Martin calls "secret English", has its
own rules and constraints, of which few pupils and only some
teachers are aware (see Tonjes, page 199). Nor is the solution
as easy as ignoring textbooks and basing instruction on pupils'
own personal experience, because "the danger of moving too
far in this direction is that students are denied access to the
major tools of the humanities and sciences; technicality and
abstraction." (Martin 1988, p. 169.)
  The problem is to explain the special nature of pedagogic
discourse to teachers and pupils so that they can consciously
work with the language used in textbooks and thus lay a solid
foundation for the independent acquisition of knowledge
implied in the curricula. Certain research results give us
reason to assume that it can be very difficult to pass such
 insight on to pupils, particularly because they have different
points of departure for reading the books at home - and at
school. Some pupils already understand the textbooks'
"ideational schemata"; while others do not understand it
(Anderson - Armbruster 1984; see also page 198).
  The abstract-concrete problem has been formulated even
more dramatically by Danish researcher Sven Sødring Jensen,
who writes in his doctoral dissertation that:

  By aspiring to a high level of abstraction, numerous
  persons, relationships and actions are put into short
  formula. By using an ordinary linguistic expression, one
  ensures that the formulas can be told to and read by the
  pupil. However, at the same time one finds that the
  explanation cannot be understood, i.e., acquired by the
  pupil. (Sødring Jensen 1978.)

Causal Explanations
Explanation is a key word, for both J.R. Martin ("... both
understand and challenge explanations"; p. 171) and Sødring
Jensen. Causal relationships play a very large part in history
and social studies books. A few large-scale investigations
have examined this situation.
  All readers perceive new texts on the basis of their own
personal qualifications for understanding, learning and
retaining what they read. Such "reader's schema, or
organized knowledge of the world" (Anderson - Armbruster
1984, p. 181), will vary considerably. Now the text also has its
own schema, and the factor that determines how well the
pupil will understand what is explained in the book will be
the concurrence between the pupil's schema and the text's
schema. Bonnie Armbruster and Thomas Anderson's
investigation thus involves an entirely different text element
from Avon Crismore's. Yet in principle, they are based on the
same hypothesis: That linguistic and stylistic analyses of an
element on the structural level can be highly indicative of the
entire text's possibilities:

  (...) much of the content of the disciplines, or subject
  matter areas, can be formulated in a relatively small
  number of generic structures or generalized plots, each
  with its own set of content categories or types of
  information. These structures reflect typical patterns of
  thought or ways of conceptualizing the content of the
  subject-matter area. (Anderson - Armbruster 1984, p.
  182.)

Armbruster and Anderson call such "generic structures"
frames. The subject of history is largely based on explaining
events in light of man's motives. Hence it is valid to speak of
a psychologically-based frame. The authors refer to the so-
called "story grammars". They are psycholinguistic models
of common structures in stories which are easily understood
and readily retained: Something happens to the protagonist
which makes it necessary to undertake some action to achieve
 a goal, and there are very specific conventions regarding how
people behave, problems are solved and goals are achieved.
Anderson and Armbruster found that this model lent itself
well for use as the main frame for history books; as a "goal-
frame" consisting of the elements "plan", "action" and
"outcome". They found that the texts contained a very
limited number of superordinate goals such as conquering and
keeping countries or power, maintaining order (history and
politics), ensuring food, water and warmth (biology), using
and distributing resources, and ensuring economic growth
(economics/social science). Such frame-goals could be
subdivided into many groups of elements which could be
subdivided again. The authors discuss several such main
frames and indicate, for example, the existence of a very
specific "war-frame", whose rather inflexible development
is followed in book after book.
  The investigation itself was limited to the application of
the goal-frame to explanatory descriptions of
psychologically-based courses of event. The topic was the
completion of the first railroad connection between east and
west on the North American continent. The texts were taken
from three different 5th grade books, each of which was the
subject of four questions: Does the text explain its objective?
Plan? Action? Outcome?
  Only one of the three texts answered as many as three of
the questions. The two others answered just one question
each. Anderson and Armbruster also analyzed the answers in
light of elementary linguistic rules for logical textual
cohesion/logical connection, and found repeated
contraventions which corroborated the negative result of the
goal-frame's questions.
  (Anderson and Armbruster are among the very few
researchers who have tried to incorporate new insight into
textbook writing. The year after they completed their
investigation of textual structures, they presented an "ideal"
chapter of a textbook, with a well-grounded, uniform,
practicable internal plan, called "Americans Develop Plans
for Government" (Anderson - Armbruster 1985). The
chapter was evaluated by several researchers. The reactions of
Gary M. Schumacher (Schumacher 1985) give some indication
of the difficulties involved in evaluating even a scientifically
prepared textbook text:

  At the beginning of my comments I indicated that the
  process of designing an ideal text is analogous to computer
  simulation in that it forces us to translate concepts into an
  actual product. Now we need to consider this analogy more
  fully. In doing computer simulations, a simulation is not
  complete until we have run the program and determined
  how well it fits human performance. Similarly, there is an
  additional step which needs to be carried out in the design
  of texts - we need to have students use them and determine
  how well they work. Unfortunately this task presents an
  interesting problem - how do we measure how well they
  work? In the past our principal approach would have been
   to have students use the materials. We would then ascertain
  either how well they did on tests about the information
  (retention measures) or how much they liked them. As
  Schumacher and Waller have argued, however, outcome
  measures such as these provide limited information about
  the effectiveness of text design. Retention measures, for
  example, provide an especially narrow window through
  which to view the usefulness of text. In fact it can be
  convincingly argued that retention of the material should
  not be our major concern. (Schumacher 1985, p. 266.))

Other research has corroborated the impression of
argumentation as an element of fundamental importance to
the accessibility of textbook texts (Bamberger - Vanecek
1988; see page 192). Toulmin - Rieke - Janik 1979 used a
model containing six "categories of argumentation" grouped
according to logical function: "Thesis" (statement,
conclusion), "reason" (material which makes the thesis
relevant), "warrant" (justifies the connection between thesis
and reason), "backing" (data that supports points under
warrant), "modal qualifier" (defines and distinguishes), and
"rebuttal" (reservations). Mauri Åhlberg (Åhlberg 1990)
applied the model in an analysis of two Finnish biology texts
for the lower secondary level. In the one text, which dealt
with animal tissue, he found nine theses, six reasons, four
warrants and one modal qualifier. In the second, which was
an introductory discussion (to and for the pupil) of the
question of why and how one works with biology, he found
24 theses and no other argumentational types of statements.
Both texts were approximately 100 words long. (The
investigation was based upon and confirmed observations in
a study by M-L. Julkunen (1990).)  Thus the unilateral
dominance of theses may appear to confirm the randomness
entailed in the construction of causal structures. R.W. Paul
has undertaken laboratory analyses which indicate the same
dominance of (unexplained) theses (Paul 1987). This means
that pupils do not learn to argue. And if they don't learn to
argue, they don't learn to distinguish between important and
less important material. And if they can't make that
distinction, it becomes more difficult to retain knowledge.
  Like Anderson and Armbruster 1984, Staffan Selander
(Selander 1988) studied causal explanations in history books
in an effort to compare how different authors presented the
same topic. He chose a number of radically different topics
and analyzed, among other things, chapters from two books
at the middle school level, one published in 1925 and one in
1975. Based on several detailed analyses of the selection of
material, the style and the arrangement of the texts, Selander
claims that although personal interests, contemporary trends
and current curricula may provide some variation, it appears
that textbook authors consistently seem to write books based
on patterns established by their predecessors. Meanwhile,
several other comparisons between the two books showed that
it was arbitrary whether or not causes were brought into the
presentation of individual events - an observation which
would, in theory, probably have resulted in several 0's on
Armbruster and Anderson's goal frame schema. Insofar as
there were any causal explanations, Selander found religious
 and political motives to receive far more emphasis than
economic motives, and he found that results were more often
attributed to individual than to collective efforts.
  Mauri Åhlberg has investigated what he calls "concept
mapping" and "concept matrices" as tools for the analysis of
the structure of textbook texts (Åhlberg 1990). In this
context, we are dealing with conceptual or topical units which
might comprise a word, a paragraph, or a chapter, and which
are connected to other related or non-related concepts in
many ways. Åhlberg refers to Staffan Selander, who writes
that "Every sensory impression and every perception are
transformed through a process of codification to a lasting
memory, provided that they are incorporated into a network
of conceptual connections" (Selander 1988, p. 128). A
narrative text is in itself a concept, a narrative structure
which in turn consists of other concepts arranged in various
types of hierarchies. This is also true of cogitative texts, in
which the concepts are used as the cornerstones of a
presentation, but not in the same way as people or actions.
According to Åhlberg, both text types often overlap and
analyses of conceptual connections will give an indication of
quality, especially if they are examined as part of an analysis
that takes account of argumentation and content (Åhlberg
1990, p. 9). He refers inter alia to Starver and Bay, who
analyzed "the conceptual structure and reasoning demand of
elementary science texts at the primary level" (Starver - Bay
1989). Åhlberg, who, like Armbruster and Anderson calls the
concept elements "nodes", has developed a method for
producing a graphic representation of their role in the text,
and he is working on the development of a system applicable
to larger text units.
  Åhlberg's colleague at the University of Joensuu, Marja-
Liisa Julkunen, has conducted a large-scale investigation of
conceptual relationships and concepts in Finnish textbooks
for the lower secondary level and the first year of upper
secondary school. Two hundred and twenty five paragraphs
of approximately one hundred words were selected from 49
different books in history, religion, geography and biology.
Julkunen found that the number of concepts per se was not
necessarily any indication of accessibility, but that their
cohesion was crucial. The most important connection from
the concept world of the text to that of the pupil was assumed
to be the pupils' experience at the lower levels. At the higher
levels, on the other hand, the main connection comprised
examples supported by illustrations and tables. The systematic
processing and development of definitions was less common.
Otherwise, Julkunen registered numerous contraventions and
shortcomings in hierarchies and levels:

  When the text and concept analyses are compared, it
  becomes obvious that the analyzed textbooks do not
  enhance learning in the best possible way. The results
  suggest that the textbook writers do not use the rather
  simple linguistic devices that could make learning from
  texts easier. (Julkunen 1990, p. 19.)
 It will always be difficult to label a text's structural elements
because demarcation will have to be effected at several levels
(theme, style, mood, composition). In any model, the most
important aspects of every labelling attempt will involve, first
of all, the question of how precisely the model describes the
subject's and grade level's textbook text, second, the degree
to which the description facilitates relating specific text
patterns to specific subjects/plans/grade
levels/authors/publishers, and third, whether it is these
particular patterns - the variations and collocations in
question - which best serve the objective regarding the book's
accessibility.

Perspectives
These investigations are also mainly characterized by the
more or less explicit desire to improve textbooks. To some
extent, they are all descriptive and normative; some conclude
with schemas that the readers themselves presumably can
apply to test the quality of the books they use. Relatively
speaking, this category has a far greater number of evaluation
criteria than any other group.
  The dominant constituent is the linguistic expression. As
far as method is concerned, the majority of the investigations
consist of book investigations based on assumptions about
how the texts will work. These assumptions are in turn based
largely on readability theories which first evolved in
educational research during the 1950s and have since seemed
to multiply exponentially. The title of readability researcher
Mogens Jansen's English-language presentation of the topic
in Denmark is indicative: The Teaching of Reading without
Really any Method (Jansen - Jacobsen - Jensen 1978).
Following the first investigations of what has come to be
called the micro-level, vocabulary for instance (Sudmann
1978), in the course of the 1980s such research approached
macro-levels such as literary genres (Gillham 1986), the
authorial role (Crismore 1984), ideology (Selander 1988) or
logical/educational structures (Åhlberg 1990). A
corresponding expansion has occurred from textbook
investigations conducted by reading teachers toward text
analyses in all subjects; analyses conducted by experts in the
subjects and/or experts in fields such as psychology,
linguistics, education or philosophy. The trends have moved
from the traditional, largely age-related requirements for the
simplification of vocabulary and syntax toward an
increasingly psycho- and sociolinguistic recognition of the
scope of the problem of communication in the textbook texts.
  Although it is possible to refer to macro-levels such as
genres, the authorial role, logical structures and ideology, a
survey would show that very little has been done in these
fields in connection with accessibility. Nor have what might
be called educational and methodological structures been
examined anywhere near as extensively as linguistic
structures. That is, they have hardly been investigated at all:
How chronological should the contents of a history textbook
be for a 10-year-old? How should natural science and social
 science be integrated in countries in which so-called
knowledge subjects comprise a single curricular unit? What
are the reasons for and the consequences of the fact that the
amount of written text in mathematics books for grades 1 -
9 has increased considerably relative to the content of
symbols?
  Peter Weinbrenner summarizes his discussion of "Text
Type, Text Structure and Text Clarity" (Laubig - Peters -
Weinbrenner 1986, pp. 317-19), by referring to S.-O.
Tergan's comprehensive survey of the field (Tergan 1983),
and places the remaining problems into six categories:

- There is no generally accepted readability theory which
  can be applied to the learning effect.
- Thus far, investigations have almost exclusively quantified
  knowledge acquisition (What is retained?).
- "Superficial structure" has received a disproportionate
  amount of attention compared with "deep structure".
- Too much focus on the text has undermined necessary
  investigations into other aspects of the teacher's role and
  the classroom situation.
- The testing situations have been spurious ("laboratory
  situation").
- Sentence structure has dominated at the expense of the
  semantic structures (which Weinbrenner calls micro- and
  macro-structures, respectively).

No attempt will be made to evaluate the various readability
theories in this survey. Several recent surveys not only review
the theories, they also point out the need to apply
psycholinguistic insight more systematically in connection
with educational and not least popularization problems
involving (individual) subjects:

  Simplified material can often be used in educational
  programs intended to train pupils in reading
  comprehension gradually and in a controlled manner. To
  select reading material of the simplified variety and then
  adapt it to the rest of an educational program, one must
  know how and the degree to which this reading material
  differs from authentic material, i.e., material which has not
  been simplified or revised for educational purposes. Only
  with this sort of insight does one have a sound basis for
  attempting to ensure further development in respect of the
  ability to read authentic texts, which is of course the
  ultimate goal for reading training in most long-term
  educational programs. (Simensen 1986, p. 1.)

The above quotation was taken from an investigation
conducted by Aud Marit Simensen on people's intuition about
text simplification: Text Simplification. An Experiment. The
author provides a review of previous work done in this field
and describes the models used for the investigations. Her own
experiment is based on the desire to break down the barrier
between form and content in readability studies. This work
subscribes to the series of recent experiments intended to
 shed light on texts in more total communicative contexts
(Källgeren - Sigurd - Westman 1977), and supports the view
that it is possible to learn far more about how "a particular
cognitive content can best be conveyed to particular groups
of readers" (p. 104).
  Ferdinand de Saussure's classical distinction between
langue (meaning collectively developed, normalized language)
and parole (meaning more individual or situation/function-
specific language; see Dale 1960, p. 31) will, when applied to
the genre of textbook texts, raise two types of questions. The
one concerns the degree of normality in textbook texts' so-
called ordinary prose. The other applies to the special nature
of textbook prose, which at one and the same time is a
manifestation of scholarship, popularization, educational
philosophy, educational method and - possibly - of the
author's personality.
  Since textbook texts are written for school and educational
purposes, it would be difficult to associate their "normality"
exclusively with the descriptive ("Textbooks are written in
neutral prose"). It would be equally difficult to tie their
special characteristics exclusively to the normative
("Textbooks tell what is correct and democratic"
(grammar/social science)). In a very special way, textbook
texts are conveyors of information and attitudes. Perhaps one
might even say that by virtue of their institutional authority,
they neutralize norms in a language and within a context
which really belongs only to the school. This dual nature is
also reflected in the fuzziness of the questions posed in many
investigations. A hypothetical example may provide a useful
illustration. General questions may be posed in a way which
may invite descriptive approaches (1): "How much of the text
consists of a cogitative representation?" Detailed questions
may be posed in a way which may invite descriptive
approaches (2): "Which argumentational functions do the
examples in the paragraph about foreign aid have?" General
questions may be posed in a way which may invite evaluation
(3): "How comprehensible are the cogitative segments of the
text?" Detailed questions may be posed in a way which may
invite evaluation (4): "How familiar are the examples used in
the paragraphs about foreign aid?" If applied to authentic
texts, such questions would quickly reveal that the text was
more complex than the measurement system. Questions 1 and
2 invite quantitative measurement and the classification of a
body of text which most probably consists of both descriptive
and normative elements. Questions 3 and 4 are also in a
borderland area; in formal terms, they belong among
investigations based on registered observations. Moreover, the
criterion of "advance insight", which may in itself be
controversial (depending inter alia on the occurrence of the
same topic other places in the same textbook), is built into the
question. Such borderland cases are inevitable. As regards
analyses that claim to be scientific in the orthodox sense of
the word, this will impose constraints on fields of vision and
possibilities. Thus if one wants to investigate the language or
content of larger textual units, one encounters one
 inescapable question: Could such duality be discovered and
discussed without a system that has enough latitude to allow
subjectivity?
  Several researchers have analyzed textbook language on the
basis of specific assumptions about ordinary prose;
assumptions which may or may not derive from general
theories postulated by language and readability researchers.
However, very few have ever written analyses which are
holistic in the sense that they simultaneously relate linguistic
analyses to patterns for evaluating ordinary prose, to
discussions of the special nature of the relevant subject's
content and to the textbook genre's extraordinarily large
number of different premises. (Bamberger - Vanecek 1988,
Ahier 1988, and especially Selander 1988 go quite far in this
direction.) This is understandable in light of the requirements
for such analyses devised by modern textual linguistics. Bernt
Fossestøl, author of what might well be the 1980s most
exhaustive international survey over this field of research,
has shown how chosing the entire (textbook) text as a point
of departure can uncover new connections (Fossestøl 1980).
  One of Fossestøl's students, Norunn Askeland, compared
20 textbook texts in religion (Askeland 1984). She applied
several methods; her analysis is divided into three parts: a
LIX analysis, a syntactical analysis and an analysis of textual
cohesion. Askeland concludes that:

  The LIX analysis and the syntactical analysis mainly
  showed that there were inherent stylistic differences
  between the textbooks and the texts. The analysis of textual
  cohesion was the first to expose the difference in degree of
  difficulty, and which provided a point of departure for a
  discussion of the relationship between syntax, semantics
  and degree of difficulty. The analysis of textual cohesion
  also illuminated the relationship between composition and
  textual cohesion.
    The analyses brought forth several examples in which
  form, content and communicative situation were
  inseparable. One must view instructive texts from the
  perspective of interaction. Texts in religion appear to
  function best when they are cohesive and when the texts
  provide latitude for a dialogue with the reader, whether he
  is a teacher or a pupil. (Askeland 1984; my emphasis.)

The preceding quotation offers a reminder of the one
perspective which must be placed above all those mentioned
above. I'm referring to Suzanne de Castell's contention that
we must also examine the attitudes of teachers and pupils to
the texts contained in the textbooks. She asserts that
traditional comprehension studies will not change much until
the textbook texts are rejuvenated as literary works: They are
not lexical documents. They, in themselves, are
interpretations of the world and must therefore also be
interpreted, like pupils learn to do with fiction texts much
earlier (de Castell 1990; see page 201). That perspective
makes it natural to ask whether it is not new approaches to
 problems rather than new methods that are needed in the area
of accessibility.
  Based on such approaches, new investigations could
progress in directions such as those Florence Davies suggests
in her analysis of traditional criteria for textbook evaluation
(Davies 1986). She contends that the point of departure for
discussions and investigations has been marked by specific,
fixed ideas about function, structure and use, and that this
fact has biased the result:

  The language of a textbook, according to one view, can be
  regarded as impenetrable and jargonistic, or, according to
  another, as necessarily subject specific, and representative
  of a particular genre. (...) The potential of textbooks for
  presenting bias can be considered grounds for teachers to
  censor or censure them, or to use them to encourage pupils
  actively to evaluate the texts themselves. The use of
  textbooks can be passive or active. The adoption of one set
  of criteria can result in global criticisms of textbooks and
  a consequent retreat from print. The adoption of the
  alternative set of criteria could result in a transformation
  in the use of textbooks, and the exploitation of a largely
  untapped potential. (Davies 1986; my emphasis.)
 
 
 

Effects and Effectiveness
 
 
 

A textbook text can achieve specific effects by using simple
textual and typographical techniques. Sophisticated
equipment makes it possible to endow texts with even more
visual appeal, thereby commanding attention and stimulating
reader interest. The question of the extent to which extra-
textual enhancements in themselves contribute to the
effectiveness of a book as an educational tool, or whether
such enhancements accentuate viewpoints and attitudes in the
text, will not be treated here. Granted, for years the
requirements of teachers and publishers in Western countries
have made lavish extra-textual enhancements standard fare.
Teachers and textbook authors in the former USSR are also
interested in more enhancements (Sujew 1986). However,
with the exception of a few major contributions, research on
the relationship between extra-textual enhancements and
effectiveness has been almost inversely proportional to the
growth seen in the technical development of textbooks.
  In both principle and practice, it is impossible to
distinguish between a presentation's visual impact and its
effectiveness. A book will achieve optimal effectiveness if
the material between its covers is written and adapted in a
way that allows most of the pupils, during the time available
to them, with or without the guidance of a good teacher, to
study the book and grasp the knowledge, understanding and
skills specified in the curriculum, as measured by tests and
examinations administered at different levels. (The definition
of optimal effectiveness disregards factors related to
organization of the teaching process and school day, which
might have a negative effect on the unassailable starting
point, the book itself.)

  With such a broad definition of effectiveness, both
accessibility and effects become subordinate factors,
relatively speaking, as they do in ordinary language usage. A
richly illustrated text written in an easily accessible style is
not necessarily motivating, informative, skill-promoting or
easy to remember, and thereby automatically effective as an
educational tool. The concept of accessibility was narrowed
down in the previous chapter and limited mainly to a question
of vocabulary, style and structure. However, factors such as
volume, arrangement, information-density and the text-
picture correlation are also relevant to the question of
effectiveness. One can turn aspects of language, style, and
extra-textual enhancements into a number of measurable
elements and make theoretically based statements about
 probable degrees of effectiveness. Such statements, however,
are rare, and little research has been done in this field. One
explanation for this is the versatility of the concept of
effectiveness itself, another very obvious reason is the
problem of methodology (see page 234).

Illustrations
"The impact of technologies both ancient and modern on
children's learning is either negligible or unknown." This
quotation from David R. Olson (Olson 1974) expresses deep-
seated skepticism to automatically applying technological
innovations to education. But the fact that textbooks may
account for as much as fifty per cent of all books published
on a worldwide basis, measured in the number of copies
printed, is a result of skilled product development and
marketing strategies. First Dewey (Dewey 1938) and later
Freire (Freire 1970) pointed out that these commercial forces
could eventually work against school policy and educational
goals. Michael Apple and Susan Jungck have compared the
electronic media challenge facing today's teachers to the
situation of books 100 years ago (Apple - Jungck 1990). They
maintain that teachers' working conditions in the 1990s are,
in principal, quite similar to their conditions in 1890: Low
status, modest wages, variable working conditions, time
pressure, and curricula they feel unqualified to teach. Just as
the time was ripe 100 years ago for demanding access to
standardized textbooks for all grade levels, today teachers are
demanding access to a good assortment of office equipment
to allow them to satisfy the curricula. These demands did not
come unanimously from all teachers then, nor do they now;
in many schools in many countries, demands are being
reduced to the need for updated textbooks. But teaching aids
are available, whether in the form of books or new
technology. However, this fact may be troublesome since
many teachers feel that teaching aids might deprive them of
some of their autonomy, regardless of the publishers'
assurances to the contrary. It is likely that widespread
skepticism to new technology and sophisticated equipment
still exists. The attitude to illustrations is a case in point:

  It is odd to think, in an age when we can create and use
  pictures on a scale never before seen, that schools' attitude
  toward pictures has retreated into the scholastic Middle
  Ages. It is still applicable, as it was then, that "letters
  make one wise, but pictures make one foolish". Thus we
  have a blind school for the sighted. (Eklund - Hedman -
  Bergquist 1986, p. 5.)

The authors are board members of the Swedish Picture and
Word Academy. Gert Z. Nordström, a professor at the
Stockholm College of Arts, Crafts and Design, writes: "There
is widespread albeit often concealed antagonism to pictures
in society, and this is also evident in schools, education and
teaching" (Nordström 1989, p. 159.)
   These Swedish comments directly contradict the
unambiguous results of all the research studies that have
asked teachers what they look for when choosing new
textbooks. The teachers all strongly emphasize the importance
of good illustrations. (Evans 1987, O'Brien 1988, Woodward
1990.) However, this inconsistency might not be as clear-cut
as it appears. What do teachers actually mean by good
illustrations? Does it suffice that the pictures look like they
might have an inspirational effect? Or do teachers know
exactly what they're looking for and evaluate illustrations as
independent teaching instruments? The ability to answer
these questions requires insight into two areas. First, we must
know whether/how the books' illustrations are used in the
classroom. Second, we must know what educational potential
lies in illustrations: Are opportunities lost in the "school for
the blind"? Hardly any research has been done on the first
point. (Christina Gustafsson's comprehensive use study (see
page 385) demonstrates that the amount of time spent on the
direct use of illustrations is minimal.) As regards the second
point, both theory and research are lacking. François
Richaudeau's introduction to the technology of textbook
development includes a separate chapter treating the use of
pictures, but with the exception of references to standard
picture theories from readability researchers, he quotes only
one study on the implications of using illustrations. Carried
out in Nepal, the study presents evidence that artists'
drawings are more inspiring as well as more informative than
photographs. (Richaudeau 1986, pp. 176-77.)
  In an article from 1991, Arthus Woodward describes the
situation as follows:

  Perhaps most disturbing is the seeming irrelevance of
  research on illustrations to selectors and producers. To
  date, research seems to have had little impact and provided
  few guidelines for practical application. Indeed, the vast
  majority of research on illustration is undertaken without
  regard to actual textbooks or the population that will read
  them. (Woodward 1991, p. 18.)

The question of the instructional value of illustrations may to
some extent be directed toward the effect per se: Are they
inspirational? Such questions lead to new ones concerning
circumstances independent of the books themselves: Is it true
that pupils require extra-textual inspiration? To some extent,
these questions concern effectiveness: What can one learn
from illustrations in textbooks? As far as visual impact is
concerned, extensive research will be required to find the
answer. Many would also contend that this issue is largely
academic - everyone likes a few pictures. In response, others
would maintain that the question becomes less and less
academic as the proportion of illustrations increases. Arthur
Woodward studied the relationship between consecutive text
and illustrations in two natural science series for grades one
through six. In both series, illustrative material covered an
 average of more than 50 per cent of the pages. As expected,
the proportion of illustrations was largest in the books for the
lowest grades, but it was as high as 43 per cent (both series)
in the fifth grade books (Woodward 1989).
  The issue of effectiveness can be broken down into three
parts, as Woodward did in a comparative study of the
chapters about electricity in two sixth-grade textbooks. The
study was cited in the article "Do Illustrations Serve an
Instructional Purpose in U.S. Textbooks?". It treats the use of
picture captions, the effectiveness of illustrations alone as
educational instruments and the way in which teachers'
manuals deal with the use of pictures (Woodward 1991, p.
10).
  Five criteria were used to analyze picture captions: no text,
identification of what is depicted, text that quotes from the
consecutive text, text that supplements the consecutive text,
and questions directed at the reader. While 74 per cent of the
illustrations in the first book had captions, the comparable
figure was only 47 per cent in the second. Further, it was
found that a large proportion of the picture captions were
either simple identifications or quotations/restatements of the
consecutive text. The one book had no supplementary texts or
questions at all. Woodward also examined the references
between the consecutive text and the pictures. Direct
references numbered well under 50 per cent both ways, in
both books.
  Four criteria were used to analyze the true effectiveness of
illustrations alone as educational instruments: Illustrations
with no demonstrable relevance to the content of the
consecutive text, pictures of objects or subjects related to the
text without being directly connected to it, pictures that
exemplified the consecutive text, and pictures that
supplemented the subject of the consecutive text with new
information and new perspectives. Woodward found that the
majority of the illustrations (75 and 86 per cent) were related
to the text in one way or another, and could generally be
placed in one of the first two groups. Only three illustrations
in each book could be characterized as supplementary (fourth
group).
  Teachers' manuals were also examined. The first had 17
references to 20 of a total of 42 illustrations in the book; the
second, four references to four of a total of 36 illustrations.
Advice concerning the use of illustrations was limited to
pointing out that the books had some illustrations. Woodward
concludes that there is definite "evidence that illustrations
are hastily assembled with little regard to their being closely
related to the text" (p. 17). At the same time, he points out
clear differences between the two books that were examined.
 Regarding the illustrations' captions, placement and relation
to consecutive text, one of the books seemed at least to have
an "illustrative strategy" (p. 15). In the other book, however,
the illustrations were "generally instructionally ineffective".
Woodward refers to another study which concludes that
intuition, traditions and market factors decide the illustrative
strategies of publishing houses (Houghton - Willows 1987).
  The impression of inconsistency is strengthened by the
results of other studies. Rune Pettersson interviewed picture
editors and graphic designers in Swedish publishing houses
and concluded that "(...) in practice, procurement time,
accessibility and clarity are the most important factors for
selecting illustrations" (Pettersson 1991, p. 142). He refers to
investigations conducted in the United States and Canada and
claims that the practical work of selecting illustrations is
based on the same factors in foreign publications as it is in
Sweden.
  Pettersson's study is cited from his book Pictures in
Teaching Materials (1991), a survey which examines and
discusses research on the use of illustrations in textbooks. In
addition to offering an introduction into basic illustrational
theory, the survey provides an overview not only of the
relatively few and scattered analyses that exist, but also of
studies on how illustrations are used in teaching. On the
whole, the book reinforces the tendencies pointed out in
Woodward's article. Pettersson's summary of his own and
others' research (pp. 69-70; pp. 93-126) indirectly confirms
the claim that textbooks, in terms of their illustrations, are
aptly described as existing in "a blind school for the
sighted".

Effectiveness
Curricula and measurements in the form of grades and
examinations are formal methods for regulating school
content and setting standards. In principle, the most effective
teaching aids are those which most quickly enable the
majority of pupils to fulfill the requirements. Since a school's
primary function is to help pupils fulfill educational goals
regarding knowledge and skills, one might automatically
assume that the field of textbook research would be
dominated by studies measuring how the pupils use
textbooks, and what and how much they learn from them. It
turns out, however, that the extent of this type of research is
very limited for a variety of reasons.
  First, current governing mechanisms such as educational
objectives and curricula are the result of years of political
compromise. Trends toward teaching guides for individual
subjects and local autonomy have increased the potential for
different interpretations of textbooks, teaching and
measurements. The "result" becomes a mutable concept. This
 will be increasingly true as the objective of individualized
and differentiated teaching gains ground.
  Second, a school is based on teaching and educating. As
superordinate objectives (see page 159), attitudes are as
important as knowledge and skills. This makes it impossible
to measure teaching effectiveness as a single factor. The fact
that the curricula in many countries maintain that the
knowledge/skills/attitudes triumvirate should be embodied in
each individual discipline makes the concept of effectiveness
vague, and, to some, suspect. Debates about "schools for
knowledge" as opposed to "schools for caretaking" can often
be traced to underlying attitudes for or against such multi-
faceted schools.
  Third, in practice it is difficult to measure achievement,
even on the basis of well-defined units. Such research would
represent severe interference with the teaching, and be time-
consuming and research intensive.
  Fourth, it seems unnatural to measure the learning process
as a result based solely on the merit of the textbook. It is a
prevailing opinion that learning involves interaction between
a book and a reader, and that the result depends upon the
reader's qualifications and the way he makes use of them
when he confronts the text:

  It is not enough that certain materials and methods have
  proved effective with other individuals at other times.
  There must be a reason for thinking that they will function
  in generating an experience that has educative quality with
  particular individuals at a particular time. (Dewey 1938, p.
  45.)

The tradition of measuring effectiveness has been most
common in nations with specific ideological educational
objectives, such as, for instance, the former USSR and GDR.
Experiences from these nations are summarized in volume 60
of Informationen zu Schulbuchfragen 1988 (see page 76). In
an introductory article, Werner Jungk writes that studies took
two directions in the 1980s. On the one hand, they steadily
developed more refined ("verfeinerte") methods for
measuring the degree to which specific structural elements in
the books can improve learning. On the other, they became
increasingly more engaged in studying individual
investigations in relation to the complex totality of all
teaching materials ("Gesamtkomplex aller
Unterrichtsmaterialen") (Jungk 1988, pp. 18-19). Jungk
points out that it is untenable that "bei der Analyse von
Struktureelementen eines Schulbuches stehenzubleibed"; one
must combine detail and totality, "the analytical view" with
"the synthetical view". Jungk describes three essential levels
("Ebene") which individual analyses must always take into
account. These are: the class and the classroom situation, the
school and the teaching situation in general, and the
prevailing situation in the nation and the international
community (pp. 19-21).
  The above is exemplified in a representative analysis made
in the (former) East German tradition. Käte Nestler examined
how the topic of the water cycle was presented in four
different fourth-grade textbooks: one Hungarian, one East
 German, one West German and one "textbook" (with two
texts) that was specially developed at the College of
Education in Köthen-Halle (Nestler 1990). The books were
used in separate classes. None of the five classes involved had
previously been taught about the water cycle and their
scholastic levels were about equal. All the pupils worked with
the texts and then with the tests, without any teacher
guidance. Nestler subsequently measured what the pupils had
learned from the textbooks: "ihre Wirkung auf die kognitive
Verarbeitung der textinhalte bei Schülern der Klasse 4". In
the report she points out that in spite of the limitations of this
type of "Wirkungs-untersuchung", research has shown that
differences in the style and structure of textbook texts can
result in differences in their effects on the learning process.
  The primary aim of the study was not to compare the
quality of the books, although it was clearly evident that the
pupils learned more from some books than others, with one
very important addition: No single book was "best" on all
points. There were significant variations on a number of
central points. Most important to Nestler was to identify the
factors within the texts themselves that added to or detracted
from the learning process, measured according to what the
pupils learned from them. In all the texts her conclusions
supported the tendency indicated by the few Western
investigations made in the same field (see page 231): The use
of examples in the texts " their "world" " and the selection
and use of concepts and technical terms are decisive, in a
negative  way. More than fifty per cent of the pupils may
experience difficulty understanding; many of them
misunderstand/do not comprehend the material. (Studies by
Nestler and other East German researchers are presented in
volume 47 of Informationen zu Schulbuchfragen (1983).)
  Nestler's study covers the relationship between book and
pupil, but excludes the teacher. As was seen in the section on
accessibility, in the Western world most investigations of this
kind are purely book studies, often based to a greater or
lesser extent on earlier research which might be based partly
on text analyses and partly on fieldwork. For example when
Harriet Tyson-Bernstein and Arthur Woodward wrote a
critical article entitled "Why Students Aren't Learning Very
Much from Textbooks" (Tyson-Bernstein - Woodward 1989),
they referred to a number of studies and reports that support
their assertion. But the article dealt primarily with a
presumed ineffectiveness because there is a lack of
widespread pupil observations and no tradition for the up-to-
date systematization of such observations within the totality
described by Jungk.

Workbook Questions and Pupils' Reading
A prime example of how a field of vision can be abruptly
expanded and a perspective sharpened when first one sets out
to observe pupils closely is Mason and Osborn's study of two
"reading periods" in classes from the first to the sixth grade
(Mason - Osborn 1982). The extensive use of workbooks
 persuaded the authors to study them too, in addition to the
basic textbooks. As early as in 1979, using essentially
traditional methods of literary analysis, Agnes Nobel
concluded that the contemporary workbooks used in many
subjects had to be ineffective since they were the result of "a
mechanistic attitude toward teaching materials" (Nobel 1979,
p. 61). Subsequent workbook research has uncovered
weaknesses within the educational system itself and pointed
out new problems (Moosbrugger 1985, Ginsburger-Vogel
1986, Osborn - Decker 1990). For example, a new perspective
appears when Steffan Selander points out that the consecutive
text in a certain history book shows clear signs of having
been written and arranged to provide direct answers to the
questions posed at the beginning and end of each chapter
(Selander 1988, p. 76).
  In 1991, Shepardson and Pizzini reported that:

  The fact that no difference was observed in the proportion
  of question levels among science disciplines suggests that
  students are uniformely exposed to low-level cognitive
  questions throughout their junior high school science
  education experience. The result is that students become
  accustomed to responding to low-level cognitive questions,
  and are limited in their compre-hension of textual
  information. (Shepardson - Pizzini 1991, p. 680.)

Shepardson and Pizzini determined the cognitive level of
questions in junior high school textbooks using an analysis
scheme which classifies the cognitive level of questions as
input ("Name the parts of the flower"), processing ("How
are pine trees different from oak trees?"), and output ("What
would be the best solution to our air pollution problems?").
Their starting point was that since comprehension of textual
information involves extracting and integrating textual
information with prior knowledge, an overabundance of
input-level questions would inhibit the students' cognitive
level of interaction with the textual information. Since they
did find such an extensive use of low-level cognitive
questions, they conclude that this "would appear to restrict
text comprehension" by limiting the prior knowledge
generated to interact with the textual information, by
creating a low-level purpose for reading, and by reducing the
integration of textual information with existing knowledge.
(P. 680).
  In terms of textbooks, the research into pupils' interest has
been limited to the subject of reading training. Numerous
investigations have been conducted to test the significance of
content, text structure and illustration to the reading process,
measured by different, often controversial methods (Jansen -
 Jacobsen - Jensen 1978). In terms of individual subjects,
attempts at measurements are sporadic despite steadily
growing evidence that textbooks are used in different ways
in different subjects (Stodolsky 1988). Beverton's
questionnaire based on journals from 116 pupils showed
typical results concerning the amount and type of material
 that was read (see page 199), but could only evaluate learning
effectiveness indirectly. A second, more qualitative, way to
relate to the pupils' work is to interview them. Lauren A.
Sosniak and Carole L. Perlman talked to 44 lower secondary
pupils about teaching and textbooks in the subjects of
mathematics, English and history (Sosniak - Perlman 1990).
Apart from attesting to strong teacher-control and the
extensive use of books in all subjects, based on the pupils'
own accounts and comments, the authors conclude that:

  (...) work with textbooks in the different subject matters
  suggests dramatically different potential for empowering
  students in their studies of the disciplines, for shaping
  classroom activity in such a way as to engage students
  intellectually and emotionally, and for helping students see
  connections between their school tasks and their life
  experiences. (Sosniak - Perlman 1990, p. 436.)

This investigation does not measure results either. However,
like most of the other approaches mentioned in this chapter,
it does uncover elements that should be included in more
systematic measurements.
  There are a few examples of more comprehensive
approaches. While it is true that none of these measurements
deals with material learned from figures or tables, both
investigations take pupils as their point of departure.
  The first looks at teaching and teaching materials in light
of the pupils' living and working situation both in and outside
school. Yngve Nordkvelle's study of social science instruction
at a Norwegian upper secondary school concludes that
textbooks form the backbone of teaching (Nordkvelle 1988,
p. 94). Teaching material read at home and in school is the
fundamental starting point for classroom discussions and
group work. Should the text in the textbook conflict with the
traditions of the subject in terms of curricula requirements
and teachers' viewpoints, the traditions will win - competing,
however, against media influence. This observation concerns
one particular aspect of learning, namely, the effect of
influence. In terms of the learning process as regards text and
material for examinations, however, textbooks are virtually
sovereign:

  (...) the first thing students look for in their notebooks
  when about to commence their homework is: What will be
  on tomorrow's test? Most students subscribe to the
  following view: The important thing is my ability to
  remember what is in the book, and the speed and clarity
  with which I write my answers. Participation in the
  classroom is something you do to keep your eyes open; you
  get good marks by doing well in the written tests. (...) It is
  vital to "cram" the selected pages, taking notes and
  memorizing the essentials. (Nordkvelle 1991, pp. 9-10.)

Nordkvelle's investigation is especially pertinent to the
subject of Third World countries as they are treated in
 textbooks and in the classroom. A whole different subject
entirely is the part textbooks play in ensuring the
effectiveness of the teaching process in Third World
countries. Certain studies, especially those commissioned by
the World Bank, show that the poorer a country is, the
stronger the correlation between access to/use of textbooks -
more than of the availability of trained teachers - and school
results. The more affluent a society, the more difficult it is to
prove these direct correlations (Heyneman - Farrell -
Sepulveda-Stuardo 1978, Farrell - Heyneman 1989).
  The second study analyzes teaching and textbooks and the
pupils' understanding of them and the subject. It is a
thorough, comprehensive analysis based on educational
philosophy. In his book Spotlight on Science (Sjøberg 1979),
Svein Sjøberg reports on three investigations: The first
analyzed the most commonly used seventh grade
physics/chemistry textbook using a taxonomy of cognitive
demand based on Piaget's theory regarding the stages of
intellectual development. The second tested seventh graders'
general ability, irrespective of subject boundaries, to think in
proportionality, a central concept in many theories of
cognitive development, as well as in quantitative descriptions
of the natural sciences. The third investigation measured how
much pupils had actually understood about key topics.
Sjøberg summarizes his results as follows:

  The textbooks are primarily directed toward an
  understanding of the subject matter based on the
  assumption that students are at the formal operational
  stage. This, at least, is a consequence of the National
  Curriculum's lists of topics.
    The great majority of pupils are, however, at the
  concrete operational stage. They will therefore not be able
  to master the subject matter in the way the book intends.
    But pupils who are at the formal operational stage will
  also have problems. (...) In that case, since many of the new
  words are by no means justified in terms of the pupils'
  concrete world of experience, it must all seem quite
  meaningless. (Sjøberg 1978, pp. 140-41.)

One of Sjøberg's crucial points is the clear distinction
between how much of a text pupils can show they have read
and can repeat, and how much of the same text they actually
understand. Of the research presented in this survey,
Sjøberg's goes furthest in attempting to "measure" the
fulfillment of more than one single or partial goal, in that he
includes the book, the teacher, and the pupil. Like most of
Allard and Sundblad's studies, reading and writing education
researchers may, as has been common practice, analyze words
in textbook texts and demonstrate that many key concepts in
the context of the book are incomprehensible at the intended
age level (Allard - Sundblad 1986, p. 27). Their conclusion is
the same as Sjøberg's. One might say they treat the question
of accessibility and effectiveness on a micro level, while
 Sjøberg works on a macro level - even though he lacks the
dimension of "contemporary society" (See Jungk, page 228,
and discussion on page 204).

Perspectives
The material from this group of investigations is too limited
to make any attempt at systematization. There is a striking
discrepancy between the importance of the textbooks'
knowledge-imparting function and the extent to which the
studies treat this function. Possible reasons for this have been
discussed previously (see page 227); the methodological
problems are especially obvious. The main problem is
classification, which is related to the controversial nature of
the concept of knowledge. Nevertheless, some classification
is already present in the books. Authors and publishers
compile them on the basis of their own ideas about what kind
of text, which pictures and which structures best teach the
pupils. Therefore, it should theoretically be possible to limit
these categories and create measurement systems to
accommodate them. However, this would presuppose access
to virtually unlimited resources.
  Klaus Lange has reasoned that methodological problems
cannot justify omitting the perspective of the books' effect in
textbook analyses: "The reference to the extensive difficulties
involved in an empirical control of the hypotheses on
effectiveness and influence does not justify precluding these
premises." (Lange 1981, p. 17.) Some major theoretical
studies have also been done in this field. The most important
of these studies focus on the question of the textbooks'
effectiveness in a way that also sheds light on other aspects of
"books in use". They will therefore be presented here.
  Theoreticians from such dissimilar political systems as the
former GDR and Switzerland have examined the possibilities
for an empirical check ("Überprüfung") of the knowledge-
related influence of textbook material. The theoreticians are
Manfred Baumann from the Wolfgang Ratke Hochschule in
Köthen and Yanouchka Oppel and Bernard Spreng at the
Institut romand de recherche et de documentation
pédagogiques in Neuchâtel. The two features they have in
common are also seen in other work in this field. First, the
relationship between books and pupils is extremely central
(books and teachers dominate studies about control). Second,
within all the different investigations this is the area in which
the need for a more precise definition of the terms
"textbook" and "teaching methods" is most strongly
emphasized.
  Manfred Baumann's analysis, "Methoden und Probleme
der Schulbuchforschung im Überblick", offers proof of the
need for this thoroughness with detail - but it must be
carefully combined with a holistic perspective - for which
Werner Jungk (see page 228) has also been searching:
   This survey was to show that a large number of different
  factors related to textbook use must be evaluated against
  one another in a holistic perspective in every individual
  analysis. (...) We are special in that we always work with
  textbook material. We must analyze this material as
  thoroughly as possible to discover what has to be processed
  by teachers and pupils, which requirements are posed by
  the textbook material and what actually elicits a particular
  effect. (Baumann 1983, p. 18.)

In accordance with this viewpoint, Baumann has devised a
two-part model that organizes all measured and measurable
dimensions either under "Teilgegenstandsbereiche"
(individual fields) or "Verfahren" (methods of procedure).
The former covers the textbooks, pupil knowledge, activities
(of teachers and pupils) and forms of teaching. The latter
consists of ways to describe these fields as well as ways to
evaluate them. Baumann maintains that the combination of
elements from the two groups will vary with each
investigation, according to a precise definition of its
objective. Baumann points out that the acquisition of
knowledge, the teaching procedure and the motivation for
learning will not automatically correlate if the objective is to
study a particular form of effect, "die entscheidene
Wirkungsrelation". He would like to see methods developed
which would encompass both the cognitive and the emotional
aspects of the process, and mentions the classification of
different types of tasks, for example, as an example of as yet
uncovered areas of research. This is necessary so that: "very
specific effects can be registered and included in a control
strategy or a uniform test of effects. Only then is it possible
to compare whether or not positive effects can be registered."
(Pp. 22-23.)
  Baumann's requirement for separating and defining is also
one of Bernard Spreng's premises (Spreng 1976). With other
researchers at the Institut romand, he analyzed French-
language textbooks from the same starting point that
Sigurgeirsson used in Iceland (see page 180). He did this
because curricula and the book stock were going to be
updated. The Swiss did not make the same type of use-
analyses as the Icelanders, but developed standard systems of
analysis. Spreng insists that all facets of the concept
"textbook" must be defined, and that the books' functions
must be defined within a system of values in which the
components are clearly limited in relation to one other. In his
introduction he therefore discusses systems such as Bloom's
taxonomy, a system of classification to which all other
methods of educational evaluation should be able to adapt.
("(...) il est en fait difficile d'imaginer un comportement
pédagogique qui n'entrerait pas dans cette classification"; p.
21.) Spreng places the individual functions in six main groups
(knowledge, understanding, application, evaluation,
elaboration and creation). As in all other systems, the
triumvirate of knowledge/skills/attitudes recurs here.
 Spreng's main point is not originality in the classification per
se, but the need - in every research situation - to adapt the
classification to precise definitions of what a "textbook" is
or should be in a given situation.
  Spreng's colleague Yanouchka Oppel conducted a study
that provided such a basis for a comprehensive description
and evaluation of the books: "L'analyse des manuels scolaires.
Elaboration d'une grille descriptive" (Oppel 1976). She
analyzes the basic premises for the development of such
descriptions and shows that textbook evaluation forms usually
mix categories so uncritically that their results become
unreliable. The model is flexible; it can be used to evaluate
books in the classroom or as a point of departure for the
definition of research tasks. It may be expanded or contracted
and still provide a clear basis for deciding which overall or
individual factors pertaining to society, school, education and
books should be included or excluded in each case. The
theory is similar to Baumann's: The object of the analysis (the
book) must in each case be decided on the basis of the
characteristics inherent in the book and the situation.
  This model has been tested and modified but its
perspective is consistently that of the textbook. Oppel's work
does not treat the ways in which one might measure the
conveyance of values, it simply points out and systematizes
what one can look for - in the books. There is no "grille
d'analyse" directly adapted to pupils' book use, nor is there
any mention of knowledge acquisition. Yet indirectly, such a
"grille d'analyse" does exist in many subjects and in most
countries, particularly at higher levels, in the form of tests
and examinations. First, the relationship between textbooks
and examinations is among the book's raisons d'être; textbook
critics would say it's their only raison d'être. Second, an
examination can be used to measure what the books have or
have not achieved. For example, there will be some pupils
taking the examination who have not studied anything but the
textbooks, just as there will be some who have not read them
at all. The numerous correlations and possibilities for
measurement opened up by this perspective have not been
utilized, possibly because of methodological problems and/or
more controversial issues concerning school policy. These
measurements will have to be based on experimentation.
Apart from the East German tradition (Informationen zu
Schulbuchfragen, Volumes 47 and 60; see Nestler page 228),
virtually no scientific testing has been done. Opposition to
this might seem to be built into the system - there are rather
strict limits to what school administrators/parents will accept
of special evaluations of their children. The more an
examination system is generally understood to be important
to future careers, the more resistance there will be to
experimentation that might impeach the influence of the very
same system.
  As previously mentioned, function categorization varies
somewhat from theory to theory. As a rule, these differences
are caused by the way the terms are used rather than by real
contradictions. The position advocated by Baumann, Spreng
 and Oppel is also apparent in the surveys of West German
theoreticians such as Hartmut Hacker and Manfred Laubig.
Hacker separates the functions into six groups:
"Repräsentationsfunktion, Steuerungs-funktion,
Motivierungsfunktion, Differenzierungsfunktion, Übungs-
und Kontrollfunktion" (Hacker 1980; pp. 14-27). Laubig
refers to these categories and, to some extent, bases his own
work on them. His report gathers them under the concept
"Verwendungs-zusammenhang", to which he adds another
concept: "Wirkungs-zusammenhang". Thus parallel
approaches turn up in several groups (Laubig 1986, pp. 26-
27). Some researchers make use of variations on the use-
effect (Verwendung-Wirkung) dichotomy. However, as I have
pointed out in this survey, the investigations have not been
dominated by these functions, but by two others: Control and
readability.
 
 
 
 

Go to: | Top of page | Next >