A Cognitive Gap between Arithmetic and Algebra

Author(s): Nicolas Herscovics and Liora Linchevski

Source: Educational Studies in Mathematics, Vol. 27, No. 1 (Jul., 1994), pp. 59-78
Published by: Springer
Stable URL: https://www.jstor.org/stable/3482666
Accessed: 20-03-2020 21:11 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/3482666?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Springer is collaborating with JSTOR to digitize, preserve and extend access to Educational
Studies in Mathematics

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 tNICOLAS HERSCOVICS AND LIORA LINCHEVSKI

A COGNITIVE GAP BETWEEN ARITHMETIC AND ALGEBRA1' 2

ABSTRACT. Serious attempts are being made to improve the students' preparation for algebra.
However, without a clear-cut demarcation between arithmetic and algebra, most of these undertakings
merely provide either an earlier introduction of the topic or simply spread it out over a longer period
of instruction. The present study investigates the upper limits of the students' informal processes in
the solution of first degree equations in one unknown prior to any instruction. The results indicate the
existence of a cognitive gap between arithmetic and algebra, a cognitive gap that can be characterized
as the students' inability to operate spontaneously with or on the unknown. Furthermore, the study
reveals other difficulties of a pre-algebraic nature such as a tendency to detach a numeral from the
preceding minus sign in the grouping of numerical terms and problems in the acceptance of the equal
symbol to denote a decomposition into a difference as in 23 = 37 - n which leads some students to
read such equations from right to left.

INTRODUCTION

The question regarding the demarcation between arithmetic and algebra is

than just academic interest. Teachers who want to prepare students for an e
course in algebra need to have some idea of what such a demarcation mig
otherwise they are bound to present them with an earlier introduction t
rather than with a preparation to algebra that might enhance their readi
pupils' lack of readiness may explain the dismal results achieved in algeb
secondary schools. The best known data which reflect these poor results or
from the United States where a National Assessment of Educational Progress
is carried out systematically every four years. In their Second Mathematics
Assessment, Carpenter et al. (1981) reported that 67% of the 17-year-olds in their
sample had completed a first course in algebra and that only 35% had taken at
least a half year of a second course. Eight years later, Swafford and Brown (1989)
reported that for the same age group, 75% had completed first-year algebra while
40% had completed a second year. Thus, only about half the students emerge
from a first course in algebra, traditionally involving one unknown, with sufficient
motivation to enroll in a course dealing with algebra in two variables. We should
also keep in mind that these results do not take into account the number of high
school drop outs. Considering that even those students who envisage a technical
vocation need to have some knowledge of algebra, one can only conclude that
many will not be prepared by the end of secondary school.
Possible explanations for the difficulties that students experience with algebra
involve both the pace at which it is covered and also the formal approach often used
in its presentation. More importantly, it seems that many teachers and textbook
authors are unaware of the serious cognitive difficulties involved in the learning

Educational Studies in Mathematics 27: 59-78, 1994.

? 1994 Kluwer Academic Publishers. Printed in the Netherlands.

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
60 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

of algebra. As a result, many students do not have the

intuitive basis for the ideas of algebra or to connect the
ideas they have developed in primary school; they fail t
the new symbolism and are reduced to performing inea
symbols they do not understand.
Failure in algebra may not reflect the students' learn
simply be induced by a certain type of instruction. Thus
cognitive accessibility do arise: Is algebra within the rea
answer to this question, the National Council of Teachers
published Algebra for Everyone (Edwards, 1990) and, as the title implies, it
argues that all students ought to be introduced to this subject. This is not a
vacuous discussion since in reality a large group of students is prevented from
entering algebra. As shown by Lodholz (1990) there are three streams of varying
mathematical content by the end of Grade 5 in primary school:

First we find those students who master the traditional curriculum ... and
proceed to a course called pre-algebra in grade 7. Next we find students
not quite as successful, and they spend the middle grades reviewing some
arithmetic in more complex exercises while they wait to enroll in algebra in
the ninth grade. Finally, we observe the unsuccessful students, who stay in
school and are relegated to a complete review of arithmetic. Typically, these
students will never enroll in a course called algebra.
.. The data in the recent international assessment (Travers, 1985) indicates
that about 10 percent of the students in the United States enroll in algebra by
grade 8 and that about 65 percent are in a regular track in algebra by grade 9.
The concern for guaranteeing success in algebra for all students is then directed
towards the lower one-fourth of the student population, who presently do not
even think about algebra.
(Lodholz, 1990,24-25)

We are thus concerned with two main groups of students: those who ente
Algebra 1 and fail or encounter major difficulties, and those who do not even
enroll in an initial course. The first group does reasonably well in arithme
but experiences problems with algebra. For them, the traditional curriculum
arithmetic may not be sufficient and a bridge between arithmetic and algebra
to be constructed. The second group is rejected on the basis of poor performa
in arithmetic. The assumption here is that elementary arithmetic is a necessar
and sufficient prerequisite. Nevertheless, questions arise as to whether or not th
should be exposed to some other mathematical experience, such as pre-algebr
instead of just repeating their unsuccessful one in arithmetic.
In order to achieve better results and reach a greater population, several e
ucators have raised the possibility of spending some of the time allocated to
mathematics in the seventh grade to introduce pre-algebra, that is, the transit
phase between arithmetic and algebra. One such example was the ALGEBRID
project initiated by Educational Testing Service and the College Entrance Exam
ination Board (1990) whose objective was to bridge the gap between arithmeti

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNlTIVE GAP 61

and algebra. Several other

The ideaof introducing pr
mathematics teachers since
is spent on reviewing prim
provide, in its present form
pre-algebra textbooks used m
simply spread out over a lo
formal course. While these
of awareness of the cogniti
between arithmetic and alg
Trying to find a demarcati
problem. Some people tend
solving a missing addend p
place holder can be conside
purely arithmetic means su
the other extreme we find
without a prior knowledge
al-Khwarizmi, considered t
who introduced it to the Me
when negative integers wer
A most interesting idea r
Filloy and Rojano (1984) wh
between arithmetic and al
both sides of a first degree equation in one unknown as in ax + b = cx +
d. The students who were studied in their research were subjected to some
instruction regarding the solution of equations involving a single occurrence of
the unknown: they were taught to use inverse operations. Afterwards, in order
to solve equations with the unknown on each side of the equal sign, they were
introduced to the classical balanced scale model and another original model based
on the equivalence of distinct rectangular areas (Filloy, 1987; Filloy and Rojano,
1989, 1985a, b; Gallardo and Rojano, 1987).
It is difficult to judge from this research how seventh graders might respond
to the task of solving equations prior to any instruction in algebra. Yet, in trying
to determine some differentiation between arithmetic and algebra, knowledge
of the range of equations that can be solved by students using purely intuitive
means might prove to be revealing. Of course, knowledge of this range can only
supply part of the answer. More importantly, the solution procedures used must
be assessed. In fact, this is a main drawback in Filloy and Rojano's notion of a
didactic cut. They define it in terms of a mathematical form, that of a class of
equations with the unknown on each side of the equal sign, without any reference
to the mathematical processes involved in their solution. This raises some difficult
questions. For instance, would a student solving such equations by systematically
substituting numerical values be doing algebra or arithmetic? If the unknown
occurs twice but on the same side of the equal sign, would the student's solution
procedure differ?

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
62 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

In fact, we cannot look at an equation without also determining how the

letter representing the unknown is perceived by the learner. In a paper on the
development of formal reasoning, Collis (1975) investigated this problem in the
context of algebraic expressions. He called these letters pronumerals since they
stand for some number or set of numbers and described three distinct levels of
conceptualization:

Lowest level (10-11 year-olds): maps pronumeral directly into a specific

number which seems viable at the time - if one trial does not give a satisfactory
result, child gives up working on that item.
Middle level (12-13 year-olds): those willing and able to map a group of
numbers (each specific on any occasion) onto the pronumeral in a sort of
"guessing and testing" technique.
Top level (14-15 year-olds): seem to have extracted a concept of "generalised
number" by which a symbol "b" (say) could be regarded as an entity in its
own right but having the same properties as any number with which they had
previous experience.
(Collis, The Development of Formal Reasoning, pp. 43-44)

While the ideas presented by Collis are very interesting, his age levels must
be taken with some caution for the algebraic expressions used in his assessments
were formal and somewhat artificial. Nevertheless, the idea of a pronumeral
evolving into a generalized number is quite enlightening. However, it is not
sufficient to endow it with "the same properties as any number", for this can be
interpreted quite passively, as for example "let n be an even number". In fact,
the pronumeral must also be endowed with the operational properties of number;
the unknown must be perceived as a generalized number that can be subjected
to all the operations performed with and on numbers. Perhaps the expression
"operational generalized number" describes this necessary evolution.
That this additional property is far from trivial has been evidenced in some
prior research. For instance, clinical evidence of this problem appears in Davis'
interview of an exceptionally bright 12-year-old student (Henry), enrolled in an
experimental class following an enriched algebra program (Davis, 1975). After
being taught how to solve equations such as 3x + 2- 5x + 6 = 9x - 3x + 23, the
class was introduced to rational forms such as 3/x = 6/(3x + 1). Upon being told
to multiply both sides by x, Henry seemed to agree that by multiplying 3/x by x
he would be left with 3. However, when told to multiply the right-hand side by x,
he replied: "How can we multiply by x when we don't know whatx is?" Although
this kind of evidence is anecdotal, it is worth mentioning for it so clearly indicates
the cognitive problem at hand. Although Henry may view the literal symbol as a
generalized number, he cannot operate with it.
More substantial evidence can be gathered from a large scale assessment study
carried out in Great Britain involving 3000 secondary school students in their
second, third or fourth year aged 13.3, 14.3, and 15.3 respectively (Kuchemann,
1981, 1978). Only results obtained from the 14-year-olds in their second year of
algebra have been published. When asked to add 4 onto 3n, 36% gave a correct

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNITIVE GAP 63

answer 3n + 4, but 31% answered 7n while 16% gave 7 as the answer. When
asked to multiply n + 5 by 4, 17% gave correct answers 4n + 20 or 4(n + 5),
while 19% answered 4n + 5 or 4 x n + 5, 31% gave n + 20 as the answer, and 15%
simply wrote 20. The last two results indicate that at least 46% of the students
will not perform the required operation on the literal symbol.
Of course, some of the difficulties faced by the British students are specific to
algebraic expressions. One cognitive problem identified with this mathematical
form is what Davis (1975, p. 18) called the "name-process" dilemma by which an
expression such as 6x is both an indication of a process ("What you get when you
multiply 6 by x") and a "name for the answer". Sfard and Linchevski (1993) have
suggested that the term "process-product dilemma" better describes this cognitive
problem. A somewhat different slant is provided by Collis' theory of the student's
Acceptance of the Lack of Closure (ALC) which describes the level of closure at
which the pupil is able to work with operations (Collis, 1974). He observed that
at the age of seven, children require that two elements connected by an operation
(e.g. 3 + 2) be actually replaced by a third element; from the age of 10 onwards,
they do not find it necessary to make the actual replacement and can also use two
operations (e.g. 6 + 4 + 5); twelve year-olds can refrain from actual closure and are
capable of working with formulas such as Volume = L x B x H; between the ages
of 13 and 15, although students are not yet able to handle variables, they have no
difficulty with symbolization as long as the concept symbolized is underpinned by
a particular concrete generalization. Collis' ALC theory is particularly relevant
to the teaching of algebraic expressions since the operations performed on the
pronumerals cannot be closed as in arithmetic. This problem was evidenced in
a teaching experiment aimed at constructing meaning for algebraic expressions
(Chalouh and Herscovics, 1984, Herscovics and Chalouh, 1984). Even after
instruction, some students could not accept 8 x a as the area of an indicated
rectangle unless it was inserted in the formula "Area of rectangle= 8 x a".
Thus, if even after an introduction to algebra, students experience difficulties
in performing operations with or on a letter representing an unknown or a gener-
alized number, one can hardly expect them to do so spontaneously without any
instruction. Although the letter in an equation or an algebraic expression may
have a numerical referent in the pupil's mind, this does not necessarily render
it operational. Meaning for these operations with literal symbols still has to be
constructed. In fact, it is to supply such meaning that Filloy and Rojano felt the
need to provide some specific instruction for the solution of equations with the
unknown occurring on both sides of the equal sign.
The inability to operate spontaneously with or on the unknown indicates the
existence of a cognitive gap that can be considered a demarcation between arith-
metic and algebra. Given the students' difficulty with the Acceptance of the Lack
of Closure of algebraic expressions, algebraic equations seem to be more appro-
priate for any further investigation of this cognitive gap. It can be conjectured
that such a cognitive gap would establish an upper boundary to the scope of the
student's informal equation solving procedures. These informal procedures are
limited to inverse operations or numerical substitutions and approximations for

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
64 NICOLAS HERSCOVICS AND LIORA INCHEVSKI

equations with a single occurrence of the unknown. For a double occurrence of

the unknown, whether on the same side or on each side of the equal sign, little
is known about the students' spontaneous handling of these problems but our
hypothesis would limit the student to numerical substitution processes.
It is to verify this hypothesis that we designed an experiment consisting of two
parts. The first part dealt with the assessment of arithmetic concepts and skills that
are considered to be essential for the solution of equations such as the cancellation
of inverse operations and the order of operations, or arithmetic concepts that need
to be expanded such as the meaning of the equal sign. The second part was a
systematic assessment of the range of equations that seventh graders could solve
without prior instruction as well as a survey of the spontaneous solution processes
they used.

METHODOLOGY

We decided to interview a whole class of seventh graders in order to

a wider range of mathematical abilities. Of the 27 students in the class, 22
were given parental authorization and were interviewed individually in two 45-
minute session. Nine of these students (mostly the weaker ones) needed a third
session lasting from 15 to 30 minutes. The average age of the 22 students was
12 years and 9 months. The class was chosen in a parochial school in Montreal
on the basis of easy access, a very cooperative staff that allowed individual
students to be interviewed whenever needed, and especially, a classroom teacher
who collaborated with us in avoiding any introduction of algebra prior to our
investigation (except for the meaning of concatenation and numerical substitutions
for literal symbols in the context of formulas) and also in making all the necessary
arrangements. The classroom teacher provided us with the results of the students'
performance in school mathematics in their regular tests and exams: 3 students
had results better than 90%, 8 students were in the 80% range, 5 in the 70% range,
5 in the 60% range, and 1 in the 50% range. Thus the class could be divided into
two groups, 11 achieving 80% or better and 11 between 50 and 80%.
In order to focus the students' attention on the equation solving processes,
we wished to eliminate most of the arithmetic drudgery and wanted also to avoid
the pupils' problem of keeping track of their various attempts. Hence, students
were strongly encouraged to use a calculator provided during the interview, a very
simple calculator that did not take into account the order of operations. Each
equation was presented on half a page with ample space for the students to write
whatever they wanted. Students were asked to "think aloud" and the interviewer,
sitting right next to the subject, wrote down each attempted solution on the side of
the student's worksheet. The student was free to consult the interviewer's notes
at all times. Students were required to go through a verification procedure: once
they thought they had found a solution, they had to write it above the unknown in
the equation and verify its validity on the calculator. Thus, an incorrect solution
would induce the search for a correct one. An observer was present during each

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNITIVE GAP 65

interview in order to take notes and also to participate later in the analysis of
the student's responses. In fact, the observer participated actively by helping the
interviewer keep track of the semi-standardized questionnaire and by intervening
whenever he or she felt that the interviewer was misunderstanding the subject.

Part 1: Preliminary Assessment

The function of the preliminary assessment was to gather some relevant informa-
tion about each student regarding his or her knowledge of the arithmetic concepts
and skills needed for the solution of equations. The first question was "Do you
know the word equation"? "Can you show me an example? ".
Nearly all our subjects responded with examples involving a numerical oper-
ation written vertically. When asked to write horizontally, most of them wrote
operations such as 5 + 7 = . In order to allow the students to become familiar
with the provided calculator, they were next asked to evaluate four arithmetic
operations involving large numbers.
The second set of questions verified their acceptance of the different uses of
the equality symbol. This problem had been investigated by Behr, Erlwanger,
and Nichols (1976) who had shown that most primary school pupils viewed the
equal sign operationally in the sense of indicating the need to perform the required
operation and writing the result as in 3 + 5 =? However, when faced with a
decomposition such 9 = 4 + 5, many children would refuse to accept it, claim that
it was written backwards and re-write it as 4 + 5 = 9. They experienced further
difficulties when the equal sign was used to represent an arithmetic equivalence
such as 3 + 7 = 6 + 4. Many would simply break it up and write two equations:
3+7 = 10 and 6+4 = 10. Several researchers have pointed out that such a limited
meaning of the equal sign persisted even among some students in secondary school
(Cortes, Vergnaud and Kavafian, 1990; Herscovics and Kieran, 1980; Kieran,
198 1a).
We asked our students if 34 = 19 + 15 was written correctly or incorrectly.
Two of our weaker students did not accept the use of the equal symbol for a decom-
position into a sum. In addition, during the latter part of the assessment dealing
with the solution of equations, we found that when the equations represented a
decomposition, 7 students ended up reading them from right to left. This was of
no consequence when only a sum was involved as in 35 = n + 16. However,
in the case of a subtraction such as 364 = 796 - n, three students changed the
intended problem by reading it from right to left "n -minus 796 equals 364".
To verify our students' acceptance of the equal sign as a symbol of arithmetic
equivalence, we asked them if 15 + 7 was equal to 10 + 12 and then how they
would show that they were equal. Most of them wrote down two equations
indicating that each sum added up to 22. Yet, when asked if we could express it
as 15 + 7 = 10 + 12, all of them agreed.
The third set of questions assessed the students' familiarity with the order of
operations. Prior research (Kieran, 1979) had shown that when solving a string of
arithmetic operations, students tended to perform them sequentially, one at a time,

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
66 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

going from left to right. This tendency might become

of operations had to be observed. We investigated this by asking our subjects to
evaluate 5 + 6 x 10 =? 17-3 x 5 =? and 8 x (5 + 7) =? Out of 22 students,
17 (77 %) gave 110 as the answer to the first string of operations, indicating that
they had first performed the addition. Regarding the second string, only 3 of them
performed the subtraction first, the other 19 immediately finding the correct answer
2. No errors were evidenced in the use of brackets. Students were then challenged
by the interviewer who referred back to the first string by explaining that "another
student had performed the multiplication first, found 60, and then added it to 5 to
obtain 65. Who was right, you, the other student, or are you both right?" At that
point all students remembered that they had been taught the order of operations
and had learned the mnemonic BOMDAS (Brackets first, or Multiplication and
Division, then Addition or Subtraction). They immediately corrected their initial
mistake and went on to evaluate a fourth string, 6 + 5 x 4 + 7 x 3 =?, correctly.
Another element we wanted to assess was the student's ability to perceive a
string of operations more globally, by somehow distancing themselves from the
string and taking so to speak an overview of it. We felt that this could be related
to a more deliberate approach to equation solving. One indicator of a global view
is the students' perception of cancellation. We asked them "Can you work out
17 + 59 - 59 + 18 - 18 =?" Only 5 students (three of them among the weak
ones) saw that no operation had to be performed because of the cancellations, 4
more students perceived only one of the two cancellations, but 13 (59%) did not
realize at all that they were present and proceeded to perform all the operations
sequentially. The interviewer pointed out the cancellations and presented another
string: 237 + 89 - 89 + 67 - 92 + 92 =? Results were somewhat better with 8
students perceiving both cancellations, 8 more perceiving the first one (+89 - 89),
but not the second one (-92 + 92), and still 5 subjects not being aware of any
cancellation.
The final part of the preliminary assessment ascertained the students' familiar-
ity with concatenation, that is, the juxtaposition of the letter and the numeral as
in "3n". In their mathematics textbook the topic of symbols and substitution had
been introduced early on. The students were told

"... 15 + n is called a variable expression. Symbols, such as the letter n, are

called variables or placeholders. Often 0 or other letters such as x, y, b, etc.
are used as placeholders. (Remember, each variable represents some number).
Variables are combined with the operations +, -, ?, and x to construct
variable expressions. m + Sm, Sy - y, p . 2, 2 x a (usually written 2a).
(Ebos, Robinson, Tuck, 1984, Math is/2, Sec. edition, p. 27)

Hence not surprisingly, all our students knew that 3n represented 3 x n and
when asked what they would get if they replaced n by 2 or 5, all of them answered
6 and 15, respectively. This is in sharp contrast with the results obtained with
another group of students who had previously been exposed to concatenation in
an introduction to algebra, but in a post test reverted back to an arithmetical frame
of reference and therefore thought that the substitutions would yield 32 and 35

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNITIVE GAP 67

respectively (Herscovics and Chalouh, 1985).

Finally, we questioned our students about their preferred form of an equation.
Carpenter et al. (1981) had found that among 13-year-olds, 91% succeeded in
solving an equation such as 4 x 0 = 24 but that only 65% succeeded when the
numeral and letter were concatenated as in 6m = 36. Among our 22 students,
12 chose to work with the concatenated form (3n), 8 chose the form using the
multiplication sign (3 x n), and 2 preferred to work with placeholders (3 x 0).

Part 2: Range of Equations and Solution Processes

As mentioned earlier, Filloy and Rojano's research on the students' solution of

equations involved prior instruction. Thus one cannot infer from it how students
might handle equations without any formal instruction. Although partial answers
to this question have been provided by Kieran (1984, 1981b), both her number
of subjects (6 and 10 respectively) and the number of equations tested (14 and 9
respectively) were too small to bring out any general trends. In prior assessments
dealing with this topic, little attention was paid to equations in which the unknown
appeared as the subtrahend or the divisor (e.g. 273 - n = 164; 525 + n = 15)
or to equations in which the unknown appeared twice but on the same side of the
equality symbol (e.g. I ln + 14n = 175). Although Kieran used one of each type
in her work, her results did not indicate whether or not the students operated on
or with the unknown.

Selection of Equations

Knowing the difficulties that students experience with rational numbers, we de-
cided to restrict our equations to natural numbers. Thus, the obstacles encountered
by our subjects would be of an algebraic nature, not due to a weakness in arithmetic.
We wanted to submit to our students equations involving all four operations, as
well as variations in the position of the unknown. It was conceivable that when
faced with equations in which the unknown was the subtrahend or the divisor,
the students might solve them by operating with or on the unknown. The size of
numbers was also taken into account in order to determine if we would witness a
shift in solution strategies. This was a reasonable assumption since in many cases,
students will use primitive processes on equations they find very easy and use
their more sophisticated procedures when they feel they are warranted. Whenever
the same type of equation was repeated, the one with the smaller numbers was
presented first in order not to discourage the subject.
Another variable that was taken into account was the direction of the equation.
Prior to our preliminary assessment, we had no idea how well the students accepted
the equality symbol for the representation of either an arithmetical decomposition
into an operation or an arithmetical equivalence. We also varied the number of
operations to verify one of Kieran's early results (Kieran, 1984) showing that
some students were having difficulties in solving an equation involving several
arithmetic operations (e.g. 4 + n - 2 + 5 = 11 + 3 - 5). We included in our list

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 68 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

TABLE I

Coding of solution procedures

C = Complementary subtrahend or factor

Div.2 = Divides by 2
G = Grouping of numerical terms
GU = Groups terms in the Unknown
I = Inverse operation
NF = Number Fact
RS = Random Substitution
SS = Systematic Substitution

IAA = Inverse Addition Algorithm

ISA = Inverse Subtraction Algorithm
SFT = Substitutes and succeeds on First Trial
S + Ap = Substitution and approximation

8 equations in which the order of operations used by the students was specifically
examined.
The last and most critical area of our investigation was the testing of equations
in which the same unknown occurred twice. We included 10 equations in which
the unknown appeared on the same side of the equality symbol in order to assess if
students would group the terms involving the unknown and thereby exhibit some
ability to operate with or on the unknown. We also included 3 equations in which
the unknown appeared on both sides of the equal sign since 5 or sometimes all 6
of Kieran's pupils had not been able to solve such equations (Kieran, 1984).

RESULTS

Since all equations were solved by nearly all our students, we will focus o
attention on the solution procedures that proved to be the most frequently u
as well as on the specific difficulties encountered by the students with some
the equations. Our very high success rate might be due to optimal conditions
mentioned earlier. Table I describes the coding used to identify the differen
solution procedures observed during the interviews.

Coding

The code C represents the procedure used in solving an equation like 37 - n =

where most students performed 37 - 18 = 19. This procedure uses the differe
as a subtrahend in the complementary form 37 - 19 = 18. We describe it as us
the Complementary subtrahend. A similar interpretation was applied to equati
in which the unknown was the divisor.

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNITIVE GAP 69

TABLE II

Equations involving addition and subtation

Equation Success rate Most frequent procedures

(1)14+n = 43 100% I 82% S+Ap 9% IAA 4%

(2) 35 = n + 16 100% I 77% S+ Ap 14% IAA 4%
(3) n + 596= 1282 100% I 86% S + Ap 9% IAA 4%
(4) 437 + n = 984 100% I 87% S + Ap 9% IAA 4%
(5) 1269 = 693 + n 100% I 87% S + Ap 9% IAA 4%
(6) 37-n = 18 100% C 73% SS 14% NF 4%
(7) n-13 = 24 100% I 77% S + Ap 9% ISA 9%
(8) 17 = n- 15 100% 1 77% S+ Ap 9% ISA 9%
(9) 23 = 37-n 95% C 72% S+Ap 14% ISA 9%
(10) 273-n = 164 95% C 72% S + Ap 14% ISA 9%
(11) n-872= 167 95% I 77% S + Ap 9% ISA 9%
(12) 235 = n-163 100% 1 86% SS 4% S + Ap 4%
(13) 364 = 796-n 91% C 73% S+ Ap 9% ISA 9%

TABLE III
Equations involving multiplication and division

Equation Success rate Most frequent procedures

(14) 16n = 64 100% I 68% NF 18% SS 9%

(15) 32n = 928 100% I 82% SS 18%
(16) 2088 = 174n 100% I 82% SS 18%
(17) 84 *. n = 4 100% C 78% SS 18% I+NF 4%
(18) 525 + n = 15 100% C 77% SS 23%
(19) n - 6 = 13 100% 1 85% SS 11% NF 4%
(20) n . 8=57 100% I 86% SS 14%
(21) 15 = n-7 100% I 86% SS 14%
(22) 23 = 115 * n 95% C 72% SFT 14% SS 9%

Inverse forms of the addition algorithm (I

(ISA) were used by a few students. For in
they would write it vertically and work out t

693

1269

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
70 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

TABLE IV

Equations involving the grouping of numerical terms

Equation Success Most frequent procedures

rate

(23)n+15+27=61 100% G +I 63% I+I 18% G+NF 9%

(24) n+34=29+38 100% G+I 73% G+NF 14% G+Ap 4%
(25)14 + n + 17 = 50 100% G + I 68% I + r 23% SS 4%
(26)23+n+18=44+16 100% G+G+I 64% G+I+I 18% G+NF 14%
(27)n + 15-9 = 61 100% G + I 77% G + I + 1 4% G+IAA 4%
(28)39+n-12 = 74 95% S+Ap 41% I +1 32% G +I 18%
(29)4+n-2+5= 11+3-S 50% G+SS 37% G3+I 9% G+NF 4%

Equations with Only One Operation

Tables II and III show the procedures used to solve equations involving one
operation. The results indicate that essentially two types of procedures are used.
For equations in which only addition or multiplication are present, students
will overwhelmingly revert to solving these by using the inverse operation, but the
numbers in the equation must be large enough to go beyond known number facts.
These results are very similar to those obtained by Kieran (1981) but differ from
those of Gallardo and Rojano (1987) who found "some confusion between the
various operations, interpreting addition as subtraction, subtraction as division"
which might explain some of their students' problems with inverse operations.
For equations involving only subtraction or only division, one must distinguish
between cases where the unknown is a minuend or a dividend from those cases
where it is a subtrahend or a divisor. In the former cases, over 77% of the students
use the respective inverse operations. In the latter cases, students respectively
subtract the given difference from the given minuend or divide the given dividend
by the given quotient. At no time did we see any evidence of students directly
performing operations on or with the unknown. Thus we can conclude that students
solve these equations by working around the unknown at a purely numerical level.
Some difficulties are worth mentioning. Five students read some of the fol-
lowing equations from right to left (eq. 9, 10, 11, 13 and 22). Two students had
some problem in accepting the equal sign as a symbol for decomposition and felt
the need to re-write equation 9 as 37 - n = 23, and did so again with equations
16, 21 and 22.

Grouping of Numbers

The presence of several numerical additions or subtractions in an equation does

not seem to have any major impact except in one special case. Table IV shows
that in general, students have no problems spontaneously simplifying the given

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNITIVE GAP 71

TABLE V

Equations involving both additive and multiplicative operations

Equation Success Most frequent procedures

rate

(30)4n+ 17 = 65 100% I +I 41% I+NF 36% SS 14%

(31)13n + 196 = 391 100% I + 1 77% SS 14% I +SS 4%
(32) 3n + 12 = 33 100% I + 1 50% SS 27% I + NF 14%
(33) 16n-215 = 265 100% I + I 73% SS 18% I + SS 4%
(34)420=13n + 147 100% I + I 73% SS 18% I+SS 4%
(35) 188= 15n-67 95% I + I 68% SS 23% NF 4%
(36) 6 + 9n = 60 100% I + 1 41% I + NF 18% NF 14%
(37)63- 5n = 28 86% C+NF 41% C +I 27% SS 18%

equations, the majority performing the ind

a comparison between equations 25 and 28 involving an unknown between two
numerical terms indicates that the presence of a subtraction induces a major shift
in the choice of solution procedures.
It is equation 29 which gave us the most startling results. Nevertheless, they
compare with those of Kieran (1984) who found that 3 of her 6 subjects provided
incorrect solutions. In our study, 11 of our 22 students failed (5 in the upper half
and 6 in the lower half of the class) and did so as a result of grouping the numbers
on the left incorrectly. They ignored the subtraction sign preceding 2 and simply
focused on the posterior addition sign, adding 2 + 5! Having thus obtained 7
they ended up solving 4 + n - 7 = 9 and found n = 12 as their solution. Even
after inserting the solution in the equation in order to verify, they maintained their
erroneous grouping and hence were not in a position to detect their mistake. This
unexpected detachment of a number from the subtraction symbol preceding it also
occurred in our preliminary assessment and we will deal with it later on.

Combined Additive and Multiplicative Operations

Regarding equations involving multiplication as well as addition or subtraction,

our results differ from prior investigations. For equations with large enough
numbers, Kieran (1981) found that only 2 of her 10 subjects were "undoing
twice". Filloy and Rojano (1984) found that for two-step equations their high-
ability subjects used a mixed strategy, "undoing the operation on the first step and
applying 'plugging-in' (non-systematic trial and error strategy) or specific fact
procedures on the second one". Our results show that with large enough numbers
(equations 31, 33, 34, and 35) the percentage of students using inverse operations
consecutively ranged from 68% to 77%. It should be noted that the three students
who failed to solve equation 37 were all reading it as Sn - 63 = 28.

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
72 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

TABLE VI

Double occurrence of unkmown on the same side

Equation Success rate Most frequent procedures

(38) n + n = 76 100% Div.2 68% SS 27% NF 4%

(39)n+5+n=55 100% I + Div. 32% NF 27% I + NF 14%

(40) 3n + 4n = 35 100% SS 55% SFT 36% GU 9%

(41) 9n - 4n = 35 100% SFT 59% SS 36% GU 4%
(42) lln + 14n = 175 100% SS 82% SFT 9% GU 4%
(43)17n -13n = 32 100% SS 86% GU 9% S + Ap 5%
(44) 5n + n = 78 100% SS 86% SFT 14%
(45)7n-n = 108 100% SS 91% SFT 9%
(46)7n +5n+ 7 = 55 100% SS 55% I + SFT 23% SFT 9%
(47)3n +5n+4rn = 19 100% SS 36% SFT 36% I + SFT 18%

Double Occurrence of Unknown on the Same Side

Since none of the students had ever seen equations of this type before, the inter-
viewer had to explain: When we have the letter (or box) twice in the equation,
it means that you have to replace each letter with (or put in each box) the same
number
Equations 38 and 39 must be considered as trivial cases that were meant to
familiarize the students with a double occurrence of the unknown and thus cannot
be seen as representative. In comparing Table VI with the previous tables we
note a major change in the solution procedures used in solving equations 40 to 47.
Whereas some form of substitution (SS or SFT) had been used in the solution of
previous equations, the most frequent occurrence was in the 20% range (23% for
equation 35 and 27% for equation 37). However, in the solution of equations 40
to 47, the frequency of substitution procedures ranges from 87% (equation 46) to
100% (equation 45). These results confirm our conjecture regarding the students'
inability to operate with or on the unknown and thus having to rely on numerical
solution processes.
It is in this set of equations that we found isolated cases of spontaneous
algebraic methods, by which we mean operating with or on the unknown. One
student was able to solve equations 40 and 41 by grouping the two terms in the
unknown, a second student did so for equations 40 and 43, and a third student
solved equations 46 and 47 by first performing an inverse operation and then
grouping the unknown. It is noteworthy that none of these students extended the
grouping procedure to equations 44 and 45 where one of the terms in the unknown
did not indicate a coefficient. The sporadic incidence of this algebraic behavior
prevents us from viewing these students as having achieved any stable algebraic
knowledge. Instead, the overwhelming evidence points to the fact that grouping
of the unknown is not a procedure that is acquired spontaneously.

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNlllVE GAP 73

TABLE VII

Equations with unknown on both sides

Equation Success rate Most frequent procedures

(48) n + 15 = 4n 91% SFT 59% SS 32%

(49) 4n + 9=7n 86% SS 55% SFT 31%
(50) 5n + 12 = 3n + 24 91% SS 86% SFT 5%

Unknown on Both Sides

In the last three equations the unknown occurred on each side of the equal sign. Our
results differ from those of Kieran (1984) whose 6 subjects came up with incorrect
solutions for the equation 4x + 9 = 7x. This same equation (our equation 49)
was submitted to our 22 students and 19 of them found a correct answer either by
systematic substitution or by succeeding in their first attempt at substitution. The
other two equations in this set were solved by 20 pupils. These equations were
more difficult to solve by substitution. Whereas in previous cases the systematic
substitution yielded a sequence of approximations to a given number, this was
no longer the case when the unknown appeared on both sides. The process of
systematic substitution yielded a comparison of the two functions on either side
of the equal sign and at some point the difference between the two was inverted
and the respective numerical values became increasingly divergent.

Detachmentffrom the Minus Sign

As mentioned earlier in the analysis of the procedures used to solve equation 29,
we found among some students a tendency to ignore the minus sign preceding
the number 2 in 4 + n-2 + 5 = 11 + 3-5. In fact, 16 of the 22 students
did at one point add 2 and 5. However, 4 of them corrected their initial mista
during the verification of their solution. The detachment of the minus sign wa
restricted to instances where a number was subtracted from an unknown, for
found evidence of this in the preliminary assessment on the questions dealing
global perception of a string of operations. For the string 17+59-59+18-18 =?
we found that 6 students were ignoring the minus sign in front of 59 and adding
59 to 18. In the second string 237 + 89-89 + 67-92 + 92 =? we found that 5
students were ignoring the minus sign in front of 89 and instead were adding 89
to 67; but even more students (10) ignored the minus sign in front of 92 and were
adding 92 to 92 and then subtracting the sum 184.
The detachment of a number from the preceding minus sign may also explain
the very different answers we obtained on the order of operations. It will be
recalled that 17 of our 22 students had worked out the string 5 + 6 x 10=? by first
performing the addition but in the case of 17 - 3 x 5 =? only three subjects had
performed the subtraction first. Although the two strings have asimilar structure,

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
74 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

the results obtained are markedly different.

This detachment of the minus sign was somewhat of
high incidence of this mistake indicates that the problem
may well reflect unsuspected cognitive obstacles (Her
this problem is somewhat induced by the introduction o
By learning that multiplication and division take prec
subtraction, some students may conclude that additio
subtraction.
Another plausible conjecture reaches more fundament
nature. Both Booth (1989) and Kieran (1989) have poin
construct their algebraic notions on their previously acq
metic. As such, their algebraic system inherits the structu
with the number systems known to them, Booth (1989
students' difficulties in algebra are in part due to their
various structural notions in arithmetic". Of course, as see
to solve equation 29, these problems are compounded wh
indicates that a number 2 has to be subtracted from
students know that they can use commutativity to o
then use associativity to perform ((4 - 2) + 5), they mu
Since our students had not seen negative numbers, the fir
-2 + 5 may not have made any sense to them. The addit
be seen as an attempt to transform the problem into a mo

CONCLUSION

Of course, we do not claim that the cognitive gap we have identified is

between arithmetic and algebra. Many others are known to exist. Fo
the difficulties students experience in translating word problems
indicate another major cognitive gap most recently illustrated by M
Stacey (1993). Another serious gap exists in the translation of funct
tabular form representation into graphs and equations (Herscovics,
The study of the latter one becomes even more important in view
technological innovations involving graphing calculators and compu
The objective of this study was to identify a possible demarcat
arithmetic and algebra in the context of equations. To this end, w
sequence of 50 first.degree equations in one unknown that were to
a class of seventh graders prior to any formal introduction to algebr
remembered that we made every effort to create optimal conditions
was available to alleviate arithmetical chores; the interviewer's notes
available to the students thereby eliminating their need to keep track;
always reminded that the correctness of the answer was not our pr
since we were interested in the procedures they used; each student was
individually and was the focus of attention of two adults, the interview
observer, who were interested in his/her thinking processes. Thu

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNlTIVE GAP 75

doubt in our mind that ou

conditions. Furthermore,
for its emphasis on acade
upper limit of the pre-al
choice.
That our students were in fact ready for an introduction to algebra was evi-
denced by their repeated inquiries about the existence of more efficient ways to
solve the equations with a double occurrence of the unknown. Quite often they
had tried to find different procedures by using the given numbers in the equation
in what seemed to be random combinations. Since these attempts were seldom
successful, they reverted to their "guess and check" method. But their guessing
was hardly random. They proved to be very systematic in their approximations.
Our study does bring new light to the upper limits of informal solution pro-
cesses. The existence of a didactic cut as suggested by Filloy and Rojano (1984,
1987) has not been confirmed. They defined the didactic cut as "the moment when
the child faces for the first time linear equations with occurrences of the unknown
on both sides of the equal sign". Our results show that most of our students were
able to solve such equations without any prior instruction. But more importantly,
the solution methods they used when the unknown was on both sides of the equal
sign were the same as those used for equations where the unknown occurred twice
on the same side. This leads us to view the demarcation between arithmetic and
algebra in terms of a cognitive gap. This cognitive gap is characterized by the
student's inability to operate with or on the unknown.
Our research was obviously influenced by the work of Kieran as well as
that of Filloy and Rojano. That our results sometimes differ is due to various
factors: different samples of students, different experimental conditions, different
objectives, and a more systematic exploration of the students' informal solution
procedures. In fact, although our study was not a teaching experiment, since
no explicit instruction was involved, there is no question that the sequence of
equations to be solved induced some learning. But this even strengthens the fact
that our subjects could not operate with or on the unknown.
The identification of a cognitive gap between arithmetic and algebra has many
interesting pedagogical implications. It defines more clearly a field of inquiry,
pre-algebra, involving those intuitive algebraic ideas stemming from the presence
of an unknown in a first degree equation, and focuses on the students' spontaneous
solution procedures. But it also brings to light the need to more carefully develop
and expand the arithmetic notions that will prove to be essential in a later course
in algebra.
The need to expand the meaning of the equal sign has further been confirmed
by our results, especially in its use to indicate the idea of decomposition, with
the possibility that this might induce a student to read an equation from right
to left. Our usual way of teaching the convention for the order of operations
may result in some over generalizations by several students who may decide that
addition takes precedence over subtraction in a string of such operations. Even if
such problems occurred with only a few of our students, it must be remembered

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 76 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

that we worked with a fairly gifted group and that these problems may be more
common among average and weaker pupils. On the other hand, the detachment of
a number from the preceding minus sign had a high incidence and this indicates
that evaluating strings of operations is not a trivial problem. These difficulties
indicate that some of the problems in early algebra find their origin in the students'
arithmetic background and warrant further investigation.
It may be difficult for a teacher to appreciate the problems experienced by
weaker or average students at an arithmetic level. But perhaps these are the
difficulties that become major cognitive obstacles in their learning of algebra.
Clearly, more research is needed to determine their prevalence and to find peda-
gogical interventions that might help the learner overcome them. By expanding
the cognitive processes needed to bridge the gap between arithmetic and algebra
many more students are likely to be recuperated. Perhaps even those students who
are presently shunted away from any algebraic experience might benefit from an
appropriate exposure to pre-algebra.

NOTES

1 This research was funded by the Quebec Ministry of Education (Fonds FCAR-92-ER-1032).
2 The authors wish to thank Patricia Lytle for her many helpful suggestions in the preparation of
paper.

REFERENCES

Behr, M., Erlwanger, S., and Nichols, E.: 1976, How Children View Equality Sentences, Project for
the Mathematical Development of Children, Technical Report No. 3, Florida State University,
Tallahassee.
Booth, L.: 1989, 'Grade 8 students' understanding of structural properties in mathematics', in
G. Vergnaud, J. Rogaiski, and M. Artigue(eds.), Proceedings ofPME-XIII, Paris, France, 141-148.
Boyer, C. B.: 1991, A History of Mathematics, 2nd edition, revised by U. C. Merzbach, John Wiley
and Sons, New York.
Carpenter, T. P., Corbin, M. K., Kepner, H. S., Montgomery Lindquist, M., and Reys, R. E.: 1981,
Results from the Second Mathematics Assessment of the National Assessment of Educational
Progress, NCTM, Reston, Virgina.
Chalouh, L. and Herscovics, N.: 1984, 'From letter representing a hidden quantity to letter representing
an unknown quantity', in J. M. Moser (ed.), Proceedings of PME-NA-VI, Madison, Wisconsin,
71-76.
Chalouh, L. and Herscovics, N.: 1983, 'The problem of concatenation in early algebra', in J. C. Berg-
eron and N. Herscovics (eds.), Proceedings of PME-NA-V, Montreal, Canada, 153-160.
Collis, K. F.: 1975, The Development of Formal Reasoning, Report of a Social Science Research
Council sponsored project (HR 2434/1) carried out at the University of Nottingham, University
of Newcastle, NSW, Australia.
Collis, K. F.: 1974, Cognitive Development and Mathematics Learning, paper prepared for the
Psychology of Mathematics Education Workshop, published at the Shell Mathematics Unit Centre
for Science Education, Chelsea College, University of London, U.K.
Cortes, A., Vergnaud, G., and Kavafian, N.: 1990, 'From arithmetic to algebra: negotiating a jump in
the learning process', G. Booker and T. De Mendicutti (eds.), Proceedings of PME XIV: Mexico
2, 27-34.
Davis, R. B.: 1975, 'Cognitive processes involved in solving simple algebraic equations', Journal of
Children's Mathematical Behavior 1(3), 7-35.

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
 COGNrTIVE GAP 77

Ebos, F., Robinson, B., and Tuck, B.: 1984, Math is/2, second edition, Nelson Canada, Scarborough,
Ontario.
Educational Testing Service and the College Entrance Examination Board: 1990: Algebridge: Concept
Based Instructional Assessment, Janson Publications, Providence, Rhode Island.
Edwards, Jr., E. L. (ed.): 1990, Algebra for Everyone, NCTM, Reston, Virginia
Filloy, E.: 1987, 'Modelling and the teaching of algebra', in J. C. Bergeron, N. Herscovics, and
C. Kieran (eds.), Proceedings of PME-XI, Montreal, Canada, Vol. 1, 295-300.
Filloy, E. and Rojano, T.: 1989, 'Solving equations: the transition from arithmetic to algebra', For the
Learning of Mathematics 9(2), 19-25.
Filloy, E. and Rojano, T.: 1985a, 'Obstructions to the acquisition of elemental algebraic concepts and
teaching strategies', in L. Streefland (ed.), Proceedings of PME-IX, OW & OC, State University
of Utrecht, The Netherlands, 154-158.
Filloy, E. and Rojano, T.: 1985b, 'Operating the unknown and models of teaching', in S. Damarmn and
M. Shelton (eds.), Proceedings of PME-NA VII, Columbus, Ohio, 75-79.
Filloy, E. and Rojano, T.: 1984, 'From an arithmetical thought to an algebraic thought', in J. Moser
(ed.), Proceedings of PME-NA VI, Madison, Wisconsin, 51-56.
Gallardo, A. and Rojano, T.: 1987, 'Common difficulties in the learning of algebra among children
displaying low and medium pre-algebraic profiency levels', in J. C. Bergeron, N. Herscovics, and
C. Kieran (eds.), Proceedings of PME-XI, Montreal, Canada, Vol. 1, 301-307.
Herscovics, N.: 1989, 'Cognitive obstacles encountered in the learning of algebra', in S. Wagner and
C. Kieran (eds.), Research Issues in the Learning and Teaching of Algebra, Reston, Virginia:
NCTM, and Hillsdale, N.J.: Erlbaum, 60-68.
Herscovics, N.: 1980, 'Constructing meaning for linear equations: a problem of representation',
Recherches en Didactiques des Mathematiques, vol. 1, no. 3, Grenoble, France, 351-385.
Herscovics, N.: 1979, 'A learning model for some algebraic concepts', in W. Geeslin and K. Fuson
(eds.), Explorations in the Modelling of Learning Mathematics, ERIC-SMEAC, Columbus, Ohio,
98-116.
Herscovics, N. and Chalouh, L.: 1985, 'Conflicting frames of reference in the learning of algebra',
in S. Damarin and M. Shelton (eds.), Proceedings of PME-NA VII, Ohio University, Columbus,
Ohio, 123-131.
Herscovics, N. and Chalouh, L.: 1984, 'Using literal symbols to represent hidden quantities', in
J. M. Moser (ed.), Proceedings of PME-NA-VI, Madison, Wisconsin, 64-70.
Herscovics, N. and Kieran, C.: 1980, 'Constructing meaning for the concept of equation', The
Mathematics Teacher 73(8), 572-580.
Kieran, C.: 1989, 'The early leaming of algebra: a structural perspective', in S. Wagner and C. Kieran
(eds.), Research Issues in the Learning and Teaching of Algebra, Reston, Virginia: NCTM, and
Hillsdale, N.J.: Erlbaum, 33-56.
Kieran, C.: 1984, 'Cognitive mechanisms underlying the equation-solving errors of algebra novices',
Southwell et al. (eds.), Proceedings of PME-VIII, Sydney, Australia, 70-77.
Kieran, C.: 1981a, 'Concepts associated with the equality symbol', Educational Studies in Mathemat-
ics 12, 317-326.
Kieran, C.: 1981b, 'Pre-algebraic thinking among 12- and 13-year olds', Equipe de Recherche
Pedagogique (eds.), Proceedings of PME-V, Grenoble, France, 158-164.
Kieran, C.: 1979, 'Children's operational thinking within the context of bracketing and the order of
operations', in D. Tall (ed.), Proceedings of the Third International Conference for the Psychol-
ogy of Mathematics Education, Coventry, England: Mathematics Education Research Centre,
Warwick University, 128-133.
Kuchemann, D.: 1981, 'Algebra', in K. Hart (ed.), Children's Understanding of Mathematics: 11-16,
London: John Murray, 102-119.
Kuchemann, D.: 1978, 'Children's understanding of numerical variables', Mathematics in School
7(4), 23-26.
Lodholz, R.: 1990, 'The transition from arithmetic to algebra', in E. L. Edwards Jr. (ed.), Algebrafor
Everyone, NCTM: Reston, Virginia, 24-33.
MacGregor, M. and Stacey, K.: 1993, 'Cognitive models underlying students' formulation of simple
linear equations', Journalfor Research in Mathematics Education 24(3), 217-232.
Sfard, A. and Linchevski, L.: 1994, 'The gains and pitfalls of reification: the case of algebra', to

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms
78 NICOLAS HERSCOVICS AND LIORA LINCHEVSKI

appear in a special issue of Educational Studies in Mathematics.

Swafford, J. 0. and Brown, C. A.: 1989, 'Variables and relations', in M. Montgomery Lindquist
(ed.), Results from the Fourth Mathematics Assessment of the National Assessment of Educational
Progress, NCTM: Reston, Virginia, 55-63.
Travers, K. J. (ed.): 1985, Second Study of Mathematics: Summary Report-lUnited States. Champaign,
University of ilinois.

Department of Mathematics and Statistics,

Concordia University,
Montreal, Canada

School of Education,
Hebrew University,
Jerusalem, Israel

This content downloaded from 62.204.213.86 on Fri, 20 Mar 2020 21:11:13 UTC
All use subject to https://about.jstor.org/terms