Fraction Sense: An Analysis of Preservice Mathematics Teachers’ Cognitive Obstacles

• Research on cognitive obstacles related to fraction sense in preservice mathematics teachers is significant, because their success depends on their skills. The acquisition of fraction sense is a complicated problem requiring a strategy to solve it. This study presents cognitive obstacles with fraction sense tests in preservice who will teach in secondary schools. It focuses on the following categories of cognitive obstacles: epistemological (language representation, tendency to generalise and rely on intuition) and didactic (less meaningful learning, and strategy). This paper takes a qualitative descriptive approach to examine 20 preservice mathematics teachers. The preservice teachers who encountered cognitive obstacles related to fraction sense testing were then grouped based on the similarity of their answers, and seven of them were selected to be interviewed. The research findings showed that five preservice teachers had overlapping obstacles: language representation and tendency to generalise; tendency to generalise and less meaningful learning; language representation, tendency to rely on intuition and trial and error strategy in; language representation and trial and error; and language representation and tendency to rely on intuition.


Introduction
Number sense is defined as a good understanding of numbers and the ability to use them flexibly (Akkaya, 2016;Sa' dijah, 2013;Yaman, 2015); fractions and decimals are parts of number sense (Way, 2011). Number sense applied to fractions is called 'fraction sense' , which is the understanding of fraction concepts, including the relationship between fractions, various representations of fractions, and flexible skills for work with fractions (Fennell & Karp, 2017;Way, 2011;Woodward, 1998).
One of the causes of students' cognitive obstacles with fraction sense is less meaningful teaching by teachers, so that the students could not complete the fraction sense tests (Cortina, Visnovska, & Zúñiga, 2014; and found the fraction topic difficult (Lortie-Forgues, Tian, & Siegler, 2015;Ormond, 2012;Yoshida & Sawano, 2002). Therefore, students' cognitive obstacles related to fraction sense tests are induced by teachers' inability to assist students in solving the obstacles.
Based on the explanation above, research on cognitive obstacles in fraction sense assignments was conducted on both students and teachers. However, studies on cognitive obstacles with fraction sense assignments in preservice mathematics teachers have rarely been carried out (Newton, 2008;Olanoff et al., 2016;Whitacre & Nickerson, 2016). Some researchers studied the preservice teachers' difficulties with fractions (Manfreda Kolar, Hodnik Čadež, & Vula, 2018;Osana & Royea, 2011;Son & Lee, 2016). These researchers analysed preservice teachers' cognitive obstacles in fraction sense assignments, focusing on the didactic and epistemological categories. The ontogenic category was not analysed since, theoretically, obstacles caused by children's mental development should diminish with age (Brousseau, 1997). Fraction topics are studied since elementary school, so the researchers assumed that there were no preservice teachers with ontogenic obstacles.
Research on cognitive obstacles related to fraction sense assignments is crucial, because it aids in understanding how these obstacles can be minimised. Meanwhile, most of the previous decade's research on preservice mathematics teachers related to fraction sense tests studied the fraction sense strategy used, the preservice teachers' low performance in the fraction domain, their understanding of representation and computation skills, their reasoning of fraction amounts, and their mental calculation (Courtney- Clarke & Wessels, 2014;Lemonidis, Tsakiridou, & Meliopoulou, 2018;Şengül, 2013;Son & Lee, 2016;Whitacre & Nickerson, 2016;Yang, Reys, & Reys, 2009). Consequently, a set of information is further required to acknowledge preservice teachers' cognitive obstacles on fraction sense tests.

Fraction sense
Preservice mathematics teachers with good fraction sense usually have the skill and expertise needed to manage fractions. However, the fact is that most preservice teachers' skills in fraction sense remain low (Courtney- Clarke & Wessels, 2014;Iuculano & Butterworth, 2011;Lamberg & Wiest, 2014;Sengul & Gulbagci, 2012;Whitacre & Nickerson, 2016;Yaman, 2015;Yang, Reys, & Reys, 2009). Such skills are used not only when doing mathematical calculations, but also in daily life. For example: If we have an income of six million rupiah and is used for household needs, is paid as taxes, and is for children's education, we can calculate the income left over. Fraction sense is also useful as the starting point for understanding the concepts of measurement, geometry, algebra, and data analysis (Purnomo, Kowiyah, Alyani, & Assiti, 2014). Someone who calculates through algorithms does not have fraction sense (Chattopadhyay, Sarkar, & Koner, 2017). Most preservice teachers used algorithms in completing mathematical calculations without attempting to give answers that make sense (Olanoff et al., 2016), which is inappropriate, since mathematical calculation should be completed not only by using an algorithm, but also using a strategy that makes sense.
According to Way (2011), fraction sense has the following components: a) understanding fractions as a language and as written symbols, b) understanding the relationship between the number of parts and the relative size of the parts, c) having a sense of the size of the fractions in relation to the whole, and d) being able to visualise and make fraction representations with diverse models. In Figure 1, we adapted the idea of number sense problems and numeration to fraction sense. The five ideas of number sense and numeration (counting, operational sense, quantity, relationships and representation) shown in Figure 1 are related each other conceptually. In number sense, counting is the reading of a series of numbers in sequence. The conceptualisation of symbols in counting is a representation of quantity. In fraction sense, counting is generally left implicit where there are still other fractions between and , such as, , and others (Simon, Placa, Avitzur, & Kara, 2018). Fraction as a quantity is an arrangement of , in which the whole is arranged into n identical (super-imposable) parts, and m of those parts are designated as the measure, amount, or quantity (Simon, 2006). For example, is defined as having the size of 4 parts out of 7 equal parts.
Relationships between numbers that are well understood help to make mathematical connections. For example, the relationships between 2 and 6 are similar to the relationships between 12 and 16, since 6 -2 = 16 -12. These relationships involve the operational sense of reduction. Relationships also involve counting, quantity, and the operational sense, which ultimately refers to distance or sequence representations. An example of a relationship in fraction sense is the following: the relationship between and is the same as the relationship between and , since -yields the same result as -. To find the difference of and , as well as in and , involves operational sense, which in fractions is the understanding of fraction operations, the characteristics of fraction operations, and the relationships between them (Alenazi, 2016). The counting, quantity, relationships, and operational sense result in distance representation, since and have equal distance between them as and , namely . In other cases, fractions can also be represented as symbols, figures, and written language (Murniasih, Sa' dijah, Muksar, & Susiswo, 2018). Cornu (1991) differentiates obstacles and learning limits into four types: cognitive, ontogenic, didactic, and epistemological obstacles. According to Cornu (1991), cognitive obstacles occur when a difficulty arises in the learning process; ontogenic obstacles occur in accordance with the child's mental development stages; didactic obstacles take place because of less meaningful teaching by the teacher; epistemological obstacles are caused by difficult mathematics concepts. Brousseau (1997) challenge Cornu's definitions of obstacles. He stated that knowledge acquisition occurs in a complex interaction between students, teachers, and the knowledge system (Brousseau, 1997). In this model, it is difficult to divide obstacles to learning into types. Obstacles can be caused by ontogenic development, a difficult lesson, and less meaningful teaching conducted by teachers. Brousseau (1997) stated that cognitive obstacles could be ontogenic, didactic, and epistemological.

Obstacles and fraction sense
The classifications of obstacles according to Cornu and Brousseau are shown in Figure 2.  Cornu (1991) and Brousseau (1997).
Cognitive obstacles in fraction sense tests are most likely influenced by didactic and epistemological obstacles as well Purnomo, Kowiyah, Alyani, & Assiti, 2014;Sbaragli et al., 2011;Yoshida & Sawano, 2002). Didactic obstacles in teaching fractions could happen because of the teacher's practices (Pinilla, 2007). The research results of  indicate that the didactic category was used to analyse conceptual change obstacles when moving from integers to fractions. For example, How the teacher observes the obstacles of students' thinking process when teaching changes from the concept 2 7 5 7 3 7 6 7 1 7 of integers (there are no numbers between 5 and 6) to the concept of fractions (there are many fractions between and ). According to Purnomo, Kowiyah, Alyani, and Assiti, (2014), epistemological obstacles in learning fraction sense occur when someone does not understand fraction density. For example, the students' mistakes related to the density of fractions: the students stated that there was no fraction between and ; this was because the students linked their pre-existing knowledge of integers to fractions, and said that there was no number between 40 and 41. Meanwhile, according to Bishop et al. (2014), a cognitive obstacle is a piece of useful knowledge for solving a particular problem, but one that generates a contradiction when it is applied in the new context. The tendency to generalise the result of positive integer's multiplication to be bigger than its two factors is not applicable in the positive fraction multiplication. For instance, in the multiplication of multiple integers of 3 x 5 equals to 15, in which 15 > 3 and 15 > 5. Meanwhile, in the multiplication of positive fractions of x equals to , in which < and < . Obstacles play an essential role in learning, since they force the learners to modify and adjust some aspects of their mindset in resolving the contradiction. Bishop et al. (2014) recommended researching cognitive obstacles based on a specific problem that is considered difficult using a test conducted in a low-performance school. An example of a cognitive obstacle that may occur when learning fractions is difficulties with equal-partitioning and equal-whole relationships (Yoshida & Sawano, 2002). Up to that point, the students have been taught fraction representation with similarly partitioned units, which may become an obstacle for students when they encounter fraction representation with unequal partitions.

Research question
In general, this study aims to analyse preservice mathematics teachers' understanding of the size of fractions, which is related to fraction sense. We want to answer the following questions: 1.
How can we identify preservice mathematics teachers' cognitive obstacles in understanding the size of fractions? 2.
What are the factors that cause the preservice mathematics teachers' cognitive obstacles in understanding the size of fractions?

Methodology
This research was conducted using qualitative descriptive data. Preservice mathematics teachers' cognitive obstacles related to fraction sense were analysed based on the accomplishment of a written test and an interview.

Sample
The research participants were drawn from a group of 69 preservice mathematics teachers who will teach in secondary schools in one of the private universities in Malang City, Indonesia, who intend to teach in secondary schools. These preservice teachers were students who had taken the courses Mathematics Principles and the Mathematics Instruction in School I. Both courses were designed to help the participants understand fractions.

Instruments and measures
The instruments employed in this study were 1) a test consisting of one question; 2) a recording device; 3) interview guidelines and 4) a field note. The test was given to establish the cognitive obstacles in fraction sense problems. The test with fraction sense problems was adapted from Clarke, Roche, and Mitchell (2011). The problems were designed to measure preservice mathematics teachers' cognitive obstacles related to understanding the size of fractions. The test was also used to measure cognitive obstacles based on mistakes in answers. The test sheets were distributed to 20 selected preservice teachers. Afterwards, interviews were conducted to investigate whether the cognitive obstacles occurring should be classified as didactic or epistemological. The test that was given can be seen in Figure 3. The numbers 4, 3, 1, 6, 5, and 7 are given. Each number can only be used once. Make two fractions, in which the numerator and denominator each contains one number, so that if the fractions are added together, the sum will be closest to 1, but not actually 1.
The focus of this study was on the preservice teachers' understanding of the size of fractions. The indicators of cognitive obstacles in the fraction sense test can be seen in Table 1, and the categories of cognitive obstacles in Table 1 are adapted from the previous research. They include the tendency to generalise (O4 and O5) (Bishop et al., 2014;Yoshida & Sawano, 2002); the tendency to rely on intuition (O6) (Purnomo, Kowiyah, Alyani, & Assiti, 2014); less meaningful learning (O7) and strategy (O8) (Pinilla, 2007;Olanoff et al., 2016;Prediger & Wessel, 2010). The category of cognitive obstacles (O1, O2, and O3) was added based on the theory shown in Figure 1 (representation). The preservice teachers understand 'closest to 1, but not actually 1' to mean two fractions with an ordered numerator and the same denominator, for example: , or , the closest to (the obstacle of language representation defined as fraction order)

O1
The preservice teachers used a given number more than once (the obstacle of representation from language to symbol)

O2
The preservice teachers chose the addition of a fraction with a distance of 0 from 1 (the obstacle of language representation when interpreting 'distance of 0 from 1')

O3
The tendency to generalise The preservice teachers added the fraction by adding the numerator and numerator as well as denominator and denominator

O4
The preservice teachers relied on pre-existing knowledge of integers and decided that the numbers closest to 1 were 0 or 2.

O5
The tendency to rely on intuition The preservice teachers did not mention the fraction density O6

Less meaningful learning
The preservice teachers forgot the numerator and denominator. O7

Strategy
The preservice teachers used a trial and error strategy, not a fraction sense strategy. O8 During the activity, this study used visual and voice recording devices. The visual recording device was used to record the preservice mathematics teachers' facial expressions while they did the assignment. The voice recording device was used to record the interviews. After that, we took field notes while we observed the preservice teachers working on the questions, and while they answered the interview questions. The selection of interview subjects can be seen in Figure 4.

Processing data
The preservice mathematics teachers were given a fraction sense assignment, and their answers were analysed. Those who answered the question correctly were not interviewed. The preservice teachers who made mistakes were grouped based on their answers, and one representative from each group was chosen for an interview. After that, the answers were categorised based on cognitive obstacles. The researchers followed an interview guideline when they interviewed the preservice teachers, discussing their cognitive obstacles more thoroughly. This was done based on the recommendations of Bezpalko, Klishevych, Liakh, and Pavliuk (2016), who stated that interview guidelines are useful for investigating a particular problem more thoroughly.
The selection of preservice mathematics teachers to be interviewed can be seen in Figure 4. Data analysis in this study followed the stages defined by Milles & Huberman (2009): transcribing the collected data, validating the data, interpreting the data, and drawing a conclusion. Transcribing the collected data included making a verbal data transcript of the preservice teachers' opinions, making an interview data transcript, making a duplicate of the completed assignment, and arranging the data by category. The next stage was examining the preservice teachers' work, recordings, Giving pre-service mathematics teachers fraction sense task Grouped the other participants who gave the same answer Analysing the task results One representative from each group was selected to be interviewed Analysing based on the indicator of cognitive obstacles Not interviewed

Yes
No Is the answer correct? field notes, and interview results. The collected data were reduced to focus on the data that were relevant to the research. The data were classified and coded based on the assigned category. Subsequently, the data were validated using source triangulation. The next step was interpreting the data description results, namely the preservice mathematics teachers' cognitive obstacles in the fraction sense problem. A conclusion was then drawn from the data obtained from all analyses.

Results
The right answer to the question in Figure 3 is + . Based on the estimation and benchmark, the two largest fractions are and . If is chosen, the remaining numbers in the question cannot be used to form a fraction that, when added, yields a number close to 1, but not 1. This means that should be chosen. Then, the remaining numbers are used to form the second fraction, which should be smaller than . The only possible second fraction is , since the addition of + will result in the fraction that is closest to 1 and is not 1. Only 10% out of the twenty participants answered correctly. This amounted to two preservice teachers, X19 and X20, who used a fraction sense strategy by counting the fractions flexibly and efficiently, which shows that they have a strong and flexible understanding of fractions and the relationships between them.
The fraction sense test can be seen in Figure 3. Based on the answers they gave, 18 out of 20 preservice teachers had cognitive obstacles in understanding the size of fractions. The incorrect answers given by the preservice mathematics teachers are presented in the following Table 2. The preservice mathematics teachers with cognitive obstacles were grouped based on their answers. Those who gave similar answers were grouped together. After that, seven representative preservice teachers were selected to be interviewed for in-depth analysis. X1 represented the answer + , X2 represented + , X3 represented + , X5 represented + , X11 represented + , X15 represented + , and X17 represented + . Those preservice teachers were selected as representatives because they had good communication skills and were willing to allocate time for the interview. The interview was conducted outside class time. The results of the interviews with the seven preservice teachers can be seen in Table 3. The preservice teacher X1 chose + because + = .
This teacher used the same numbers more than once. This teacher also correlated with the pre-existing knowledge of integers, and when he was given a fraction problem, he generalised by adding the denominators as well as the numerators. This teacher said that the closest number to 1 but which is not 1 is 0 distance from 1.

X2
The preservice teacher X2 chose + and his reasons were that the numerators should be smaller than the denominators, and that the numerator should be close to the denominator. However, the teacher thought that the numerator was the denominator and vice versa. Preservice teacher X2 also stated that the sum of + is equal to with 1 + 1 = 2 (this teacher stated that the number 2 was close to the number 1).

O5, O7
3 X3 The preservice teacher X3 had initially tried other answers, namely + = = and + = = , and finally decided on the answer + = = since is the closest number to 1. After that, the preservice teacher X3 drew a number line and divided the number line between 0 and 1 into 12 parts, and then positioned to the left of 1.

X5
The preservice teacher X5 relied on feeling when he chose a fraction that could be easily represented in decimal form, namely = 0.3. After that, this teacher chose the fraction of . However, since + = 0.8 + 0.3 = 1.1 (closest to 1), he then tried to add and , since + = 0.6 + 0.3 = 0.9. The preservice teacher X5 interpreted 'the closest number to 1 which is dissimilar to 1' as language representation of a fraction order (the order after is ).

X11
The preservice teacher X11 chose + = 0.75 + 0.20 = 0.95. He thought that 0.95 was the number closest to 1 that is not 1. The preservice teacher drew a number line and placed 0.95 to the left of 1. This teacher did not try any other fraction. The preservice teacher X15 chose the first fraction whose value was close to 1, namely . He thought that adding would give a sum of 1, so he decided to select as the second fraction. This teacher stated that the reason that + would result in a fraction smaller than one. He stated that was the fraction located on the left side of 1 on the number line. This teacher said that there was no other fraction between and .

X17
The preservice teacher X17 thought that the addition of + would result in one. The preservice teacher X17 stated that the number 1 with 0 distance was the closest to number 1.

O3
Based on the interview responses in Table 3, the preservice mathematics teachers' cognitive obstacles related to the fraction sense test can be investigated further.

Discussion
Participants X1, X2, and X3 underwent overlapping cognitive obstacles. In Table 3 (1), the preservice teacher X1 generalised the answer by relating it to the addition of integers, so that when he needed to add fractions, he added the denominators as well as the numerators (O4). This happened because the preservice teachers did not understand the characteristics of operations with fractions with different denominators (operational sense). This result is consistent with the research reporting that the tendency to generalise from pre-existing knowledge can lead to a contradiction when facing new knowledge (Bishop et al., 2014). Previous knowledge that is poorly understood will be one of the inhibitory factors in solving the problem (Magajna, 2013). However, the results of the present study were different from the study conducted by Bishop et al., (2014) in that besides the tendency to generalise from previous knowledge, there was an overlapping obstacle of representation from language to symbol in the interpretation of the statement 'each number can only be used once. ' The answer of the preservice teacher X1 used both 3 and 4 twice (O2), even though the question stated that 'each number can only be used once. ' The cognitive obstacles related to representation from language to symbol and the tendency to generalise caused the preservice teacher X1 to answer the question incorrectly. This result corresponds to the research conducted by Murniasih, Sa' dijah, Muksar, and Susiswo (2018), who reported that many preservice teachers made mistakes in representation when transitioning from written language to the fraction symbol. Based on the results of the interview with the preservice teacher X1, he also encountered an O3 obstacle. He also stated that he considered fractions difficult, which is consistent with the studies of Fatqurhohman, Sa' dijah, Irawan, andSulandra, (2017), Lortie-Forgues, Tian, andSiegler (2015), and Prayitno, Purwanto, Subanji, and Susiswo (2018).
In Table 3 (2), the preservice teacher X2 confused the numerator with the denominator (O7). This was caused by less meaningful learning. Meaningful learning should be created in the classroom so that the students do not easily forget a particular concept (Pinilla, 2007;Subanji, 2016;Fatqurhohman, Sa' dijah, Irawan, & Sulandra, 2017). The less meaningful learning obstacle also overlapped with the tendency to generalise (O5). The preservice teacher X2 thought that + was probably equal to 2. This teacher correlated with his pre-existing knowledge of integers by saying that 2 was close to 1.
In Table 3 (3), the preservice teacher X3 said that was the closest number to or 1. The preservice teacher used a trial and error strategy to obtain the answer (O8), which is not wrong, but is a time-consuming way of finding the answer. The researchers asked if there might be any other fraction between and . The preservice teachers answered there was no possibility of this by representing the fraction on the left side of number 1 in the number line. The preservice teacher X3 stated that after , the next fraction was . The researchers concluded that the preservice teacher did not understand fraction density, and encountered the obstacle known as the tendency to rely on deceptive intuition (O6), and also a language representation obstacle with the fraction order of one twelfth (O1). This case is consistent with the study of Purnomo, Kowiyah, Alyani, and Assiti, (2014) reporting that obstacles in fraction sense occurred when someone did not understand the density of fractions. However, the present research differed in that there was an overlapping between the tendencies to rely on intuitive experience when dealing with fraction density.
Based on the answer in Table 3 (4), the preservice teacher X5 faced the obstacle of trial and error (O8). This teacher wrote the answer of + = 0.6 + 0.3 = 0.9 because 0.9 is the only a decimal bigger than 0.9 and less than one located right on the left side of one (O1). However, this reason is incorrect due to the existence of many decimals between 0.9 and one, such as 0.91, 0.93, 0.97, and so forth.
The results of the interview showed that the preservice teacher X11 encountered an obstacle of language representation when interpreting order, and stated that + = 0.75 + 0.20 = 0.95 was the fraction whose position was on the left side of 1 (O1). The pre-service teacher X15 stated tha + = . The researcher asked whether there were fractions between and , and the teacher stated that there were none, and thus encountered an O6 obstacle. The preservice teacher contended that was the fraction order before . This means that the teacher faced the obstacle of language representation when interpreting the fraction order (O1, X15 also stated). Meanwhile, the preservice teacher X17 answered + = . According to the result of the interview, X17 thought that is the closest number to 1 since the distance between them is 0. Thus, it can be concluded that X17 experiences a language interpretation obstacle (O3).
The results show that the seven preservice teachers encountered cognitive obstacles in understanding the size of fractions. Therefore, building fraction sense for preservice teachers is essential (Whitacre & Nickerson, 2016). The cognitive obstacles of seven preservice mathematics teachers presented in Table 3 are also shown in Figure 5. The research findings are further described in Figure 6.

04, 05
Tendency to rely on intuition (Purnomo, Kowiyah, Alyani, & Assiti, 2014)  As shown in Figure 6, the cognitive obstacles related to fraction sense in the seven preservice teachers were divided into two categories: didactic and epistemological. Generally, the three biggest obstacles encountered by the seven preservice mathematic teachers are presented in Figure 5. Those obstacles are language representation (O1), the tendency to rely on intuition (O6), along with trial and error strategy (O8). The research also showed that the preservice teachers faced overlapping cognitive obstacles, especially teacher X1 (type O2, O3 and O4), X2 (type O5 and O7), X3 (type O1, O6, and O8), X5 (type O1 and O8), and X15 (type O1 and O5). This is consistent with the discussion of experts, who state that differentiating cognitive obstacles from the other types of obstacle is a complicated thing (Nyikahadzoyi, Mapuwei, & Chinyoka, 2013). In fact, the obstacles discussed here are not easy to distinguish from one another.

Conclusions
We conclude by answering the research questions. First, this study found that preservice mathematics teachers' cognitive obstacles related to fraction sense were epistemological and didactic, which is consistent with previous research reporting that didactic obstacles in learning fractions could happen due to less meaningful learning (Pinilla, 2007;. Meanwhile, epistemological obstacles to working with fractions occurred when someone did not understand fraction density (Purnomo, Kowiyah, Alyani, & Assiti, 2014), and could not solve new problems when contradictors arose with pre-existing knowledge (Bishop et al., 2014). Second, these results show that the types of cognitive obstacles observed were related to language representation, the tendency to generalise, the tendency to rely on intuition, strategy and less meaningful learning which are all defined as cognitive obstacles according to Brousseau (1997). These cognitive obstacles led to an incorrect answer. Based on the written answers and interview responses of the seven preservice teachers, the biggest obstacle was language representation. The obstacle related to language representation corresponds to the research conducted by Prediger and Wessel (2010), who found that a thorough understanding of language is necessary for working with fractions.
We also found that overlapping obstacles occurred in the cases of five preservice teachers. These were language representation with a tendency to generalise, a tendency to generalise with less meaningful learning, language representation with both the tendency to rely on intuitive experience and less meaningful learning, language representation with less meaningful learning, and language representation with a tendency to rely on intuitive experience. Third, the factors causing cognitive obstacles with fraction sense tests in preservice mathematics teachers were the preservice mathematics teachers' low level of skill with fraction sense tests, the fact that fraction sense is considered difficult by preservice mathematics teachers, and less meaningful learning. Based on the result of this study, the cognitive obstacles are overlapping with epistemological and didactic categories. This finding does not support Brousseau's theory postulating that epistemological and didactic categories are clearly separated. Further research is recommended to apply the scheme of the five big ideas from number sense to fraction sense, and use it to analyse the cognitive obstacles more thoroughly.