Nindya et al. Cohesive devices in argumentative essays by Indonesian EFL learners

Cohesive devices are essential elements that have long been recognized as important features of good writing. However, making use of them appropriately is viewed as problematic for learners. This descriptive qualitative study aimed to investigate Indonesian EFL learners' ability to use cohesive devices in their writings. Twenty EFL learners of a state university in Malang, East Java, Indonesia were required to write and send one piece of argumentative essay consisting of five paragraphs to be analyzed. Twenty essays were analyzed by identifying the number of correct and errors of grammatical cohesion. The errors were then evaluated to know the possible causes that contributed to the learners' errors. The results showed that the total number of grammatical cohesion used by the learners was 2386, while 175 of them performed errors. It was also found that there was a heavy reliance on the use of particular cohesive items. The intralingual transfer mostly caused the errors made by learners. This implied a need for writing teachers to apply strategies for the development of EFL learners' writing ability, such as exposing students to exercises, using collaborative learning, giving feedback, and using the learning management system.

The aforementioned previous studies mostly examined how EFL learners encounter problems in applying cohesive devices. However, little is unknown what causes cohesive devices errors made by the learners. Meanwhile, figuring out what causes the errors is necessary to be done by applying clear procedures to investigate what sources contribute to the errors. One way to reveal the learners' difficulties in applying cohesive devices is by analyzing the errors (Kotsyuk, 2015;Widiati & Cahyono, 2006). By knowing the errors, Kotsyuk (2015) suggests that learners will reveal what is difficult to understand. If the learners know what triggers them to make errors, they can pay more attention to dealing with the errors. In addition to that, referring to Kotsyuk (2015), Martínez (2015), and Murtiana (2019), figuring out the cause of errors also helps teachers to find the most appropriate remedial teaching to overcome the problems. Hence, the teachers are likely provided with ideas of what they need to do to help EFL learners make better use of cohesive devices. Brown (2007) has distinguished the causes of errors in different types. He suggests two main sources contribute to the learners' errors: interlingual and intralingual transfer. The interlingual transfer occurs due to the interference of the learners' native language (L1). For instance, the errors due to interlingual transfer can be seen in the sentence. On the other side, teachers' reward does not significantly influence teacher professionalism (Patriana et al., 2016). The writer used the phrase, on the other side, as Indonesian-English word-by-word translation for "di sisi lain". The intended meaning was, on the other hand. However, due to the lack of lexical repertoire and practice, the writer translates the phrase into on the other side, instead of on the other hand. Patriana et al. (2016) further suggested that this kind of mistranslation problem occurred when the writers transfer their L1 writing practices to their L2 writing. According to Murtiana (2019), L2 learners generally commit interlingual errors in the same vein because they use their L1 as a tool to learn L2 and often transfer structures from L1 to L2. Therefore, in learning L2, these interlingual errors might occur when learners are not able to separate L1 from L2, and rely on a word per word translation of L1 structure.
Another source of errors mentioned by Brown (2007) is an intralingual transfer, often called a developmental error. The learners' target language competence influences this error. It occurs due to the learners' incompetence in applying the grammar rules of the target language. Referring to Al-Khresheh (2016), intralingual transfer errors encompass four aspects: over-generalization, ignoring rule restriction, incomplete application of rules, and false concepts hypothesized. An example of an intralingual transfer error taken from the empirical finding of Murtiana (2019) study can be seen in the sentence. There are so many organizations on the UIN. The identified intralingual error happened because the learner used the article "the" improperly before the name of an institution UIN when it is not needed (Murtiana, 2019). Regarding this, Brown (2007) pointed out that most of the intralingual errors are affected by the learners' experience in learning a new system of second language learning. It means that their existing knowledge might manifest new errors when applying a new grammar rule in composing a text.
The issue that might arise is whether it is significant to investigate errors while they are seen as common and acceptable in a language learning process. Kotsyuk (2015) suggests that making errors is not seen anymore as bad signs in the learning language, and it is an inevitable thing in the world. He further claims that errors are considered as an important part of a language learning process. Amara (2015) also avers that making errors was seen as a sign of mislearning and regarded as an undesirable language learning process in the past. Then, it later claimed that each human being might make errors while they are learning a language. However, investigating the errors remains necessary to do as Kotsyuk (2015) suggests that learners can benefit from the errors they make in the language learning process by obtaining feedback from the teacher. As a result, they will probably not repeat the previous error deployment in the future language learning process. Thus, analyzing errors is necessary to do even though errors are seen as common things that can be accepted in a language learning process.
Not only to know what is still problematic for EFL learners, investigating how they use CD errors is also significant since CDs impact the quality of writing. Admittedly, the presence of errors does not always make an essay not coherent. However, the existence of errors might affect the quality of writing. Previous studies have examined the correlation between the number of cohesive device errors and the quality of writing (Martínez, 2015;Yang & Sun, 2012). Yang and Sun (2012) have examined the correlation between the number of conjunction errors and writing quality across different undergraduate Chinese EFL learners' proficiency levels. It was found that EFL learners at different proficiency levels significantly diverged from each other in their cohesive errors deployment. The result also shows a positive correlation between the use of cohesive devices, the quality of writing, and the EFL learners' proficiency levels. In a different context, Martínez (2015) has also conducted a comparative study to examine the conjunction use and conjunction errors in argumentative essays written by Spanish EFL learners. Findings showed that the frequency of conjunction errors was negatively correlated with the quality of the writing. These findings have implied that the higher the number of errors found in a text, it can lower the text's quality.
A great deal of research has investigated the learners' cohesive device errors in the process of writing English texts (e.g., Hamed, 2014;Mohamed, 2016;Nasser, 2017;Patriana et al., 2016;Rahayu & Cahyono, 2015). However, the aforementioned strand of research has not addressed more attention to the factors affecting such errors. In response to this, the researchers intend to address the gap found in previous studies by some researchers on similar topics. Thus, this study attempts to examine how EFL learners apply grammatical cohesion to build cohesion in a text when they write an argumentative essay. This is done to find out the most frequently used cohesive devices by EFL learners as an indication of their ability to varying the use of cohesive devices when writing an essay. Besides, it also aims at investigating the difficulties of Indonesian EFL learners in applying cohesive devices. The investigation is intended to find out the frequency and the types of cohesive device errors that Indonesian EFL learners make.
Additionally, the study also provides linguistic descriptions of what possible causes that contribute to the learners' errors when they write an argumentative essay. Based on the aforementioned objectives, the research questions are thus formulated as follows, "How do EFL learners use cohesive devices in their argumentative writing?", "What is the frequency of cohesive device errors found in the learners' writing?" and "What are the possible causes of Indonesian EFL learners to make cohesive device errors?"

Research Design
This study employed one of the approaches in qualitative analysis, that is, discourse analysis. Peter and Wetherell 1994 in Punch (2005) point three features which make discourse analysis pertinent for qualitative research, one of which is that it concerns talk and texts as social practices. As such it pays close attention to both linguistic content (meanings and topics) and linguistic form such as grammar and cohesion, which is the focus of the present study. The qualitative approach also enables the researchers to analyze the data indepth, particularly since this study involved written texts as the primary source of data (Murtiana, 2019).

Participants
The data source was a collection of twenty argumentative essays composed by twenty undergraduate EFL learners enrolled in the argumentative essay class of the English Language Teaching (ELT) program at a state university in Malang, East Java, Indonesia. The number of argumentative essays was twenty due to the consideration of the writing teacher's access, which was only one writing class. A group of learners produced the learners' writing with the following homogeneous characteristics: they learn English as a foreign language, they were from the same academic year, they have passed the following pre-requisite courses: intensive course, paragraph writing, and essay writing.

Data Collection
The data collection was conducted by requiring twenty EFL learners to send their final project of an argumentative essay to the researchers through email. All of the learners' essays consisted of five paragraphs and the number of words varied from 600 to 800 words. Concerning this, the number of words did not determine the number of grammatical cohesion and errors. The topics were also varied since the learners were allowed to decide the topic by themselves. Twenty argumentative essays were collected from the learners without looking at any specific criteria for every sample of the aforementioned characteristics.

Data Analysis
The data in this study were analyzed through two procedures. First, the data were analyzed by identifying the correct use of all grammatical cohesion found in the learners' text. This study concentrated on the frequency to know the learners' tendencies in using reference, substitution, ellipsis, and conjunction when they wrote the essay. Second, the number of cohesive devices errors were examined. In analyzing the data, other cohesive devices that were not categorized as grammatical cohesion were ignored. The analysis only focused on the use of items that function as grammatical cohesion. For the first and second procedures of analyzing the data, Haliday and Hasan's (1976) cohesion taxonomy was used to find out the types of cohesive device errors found in the texts, whether they were categorized as misuse, unnecessary addition, omission, or redundant repetition. The description for each type of error is figured in Table 1.  Table 1 was used to categorize learners' errors, whether they are misuse, unnecessary addition, omission, or repetition. After the errors were analyzed, the next procedure was evaluating the errors to know the possible causes that contributed to the learners' errors. In this stage, the researchers were required to interpret of what possibly causes the learners' cohesive device errors. To analyze the possible causes of errors, this study adopts a theoretical framework to determine the sources of errors proposed by Brown (2007). The learners' errors were categorized into two sources of errors, namely interlingual and intralingual transfer. The errors were analyzed and categorized into either interlingual or intralingual transfer based on the characteristics of the errors observed by the researchers/raters.
Regarding the distinguish between intralingual and interlingual transfers as the source, Brown (2007) points out that such errors can be detected due to the fluent knowledge or even familiarity with the learners' L1. These could help researchers to detect and analyze the learners' errors. Besides, the repeated systematic observation of the learners' text also helped researchers remove the ambiguity in detecting particular errors.

FINDINGS The Use of Cohesive Devices
The data analysis showed that the learners applied all types of grammatical cohesion. The total number of grammatical cohesion used by the   Table 2 shows that two grammatical cohesion was chosen significantly over the others, namely reference with 1757 occurrences and conjunction with 619 occurrences. It is also presented in Table 2 how the students' tendency in using each grammatical cohesion category in their argumentative essay. The frequency of each category and the explanation of mostly used grammatical cohesion by the EFL learners are further described in the following sections.

Reference
The finding showed that reference was dominantly used, among other grammatical cohesion. The 1757 references occurred in the learners' texts covering the use of personal, demonstrative, and comparative reference. The data analysis revealed that the learners mostly used personal references such as personal pronouns "they", possessive determiner "their", and possessive pronouns "them". The definite article of "the" was the most frequently used of demonstrative reference. While in the use of comparative reference, the learners dominantly used the form of comparative degree "(adjective+er) + than/rather than". The learners' high frequency of reference was because types of reference are commonly and grammatically used as the part of sentences as either subject, modifier, or object.

Substitution and ellipsis
Findings showed that substitution and ellipsis were the two least used categories of grammatical cohesion. It was found that there were only six occurrences of nominal ellipsis and four occurrences of nominal substitution. In the use of ellipsis, the learners omitted the two particular nouns and used 'both' instead to represent the same nouns that have been mentioned earlier. While in the use of substitution, it was found that the learners applied the items 'one' and 'ones' to replace the nouns mentioned earlier in the previous sentence. This finding corresponded with the result of the study done by Adiantika (2015) investigated the number of cohesive devices in the learners' expository essays. The result showed that the use of ellipsis and substitution were found very limited in the learners' essays.
The type learners' writing may characterize the low frequency of ellipsis and substitution, that is, argumentative essays. In argumentative essays, the learners are in attempts to establish claims and take a stand on their arguments to convince the readers using academic language. On the other hand, ellipsis and substitution features are mainly for spoken purposes and may not have been often used in academic writing (Alarcon & Morales, 2011). It is relevant to Haliday and Hasan (1976) that ellipsis and substitution are used more in speech than in writing. Therefore, it is reasonable that inside the circumstance of academic writing, in particular argumentative writing, the use of substitution and ellipsis was rarely found.

Conjunction
The use of conjunction was the most frequently used of grammatical cohesion after references. There were 619 occurrences of conjunction found in the learners' texts. The subcategories of conjunction, additive, adversative, causal, and temporal were used to decrease order. The additive, 'and' and 'in addition', were found with the highest frequency. In the use of adversative, the learners dominantly used 'but' and 'although'. To show causality and result, 'because' and 'so' were dominantly used. While in showing sequence, 'first', 'second', and 'in conclusion' were the most frequently used temporal conjunction.
Based on the findings illustrated above, it is noticeable that the high frequency of grammatical cohesion by the learners resulted from the use of reference and conjunction. In the use of reference, the dominant use was under the circumstances that references, particularly the subcategories of personal and demonstrative reference, are needed as a part of the sentence. While in the use of conjunctions, the learners tended to use particular conjunctive items repetitiously. Instead, the learners might be able to use other grammatical cohesion which has the same meaning and function to avoid the overuse of particular items. As the alternatives, for instance, instead of using 'and' and 'in addition' many times in the text, the learners could apply 'likewise', 'similarly', 'furthermore', or 'in the same way' to show addition. In the use of adversative, the learners could expand the use of 'but' and 'although' to other alternatives such as 'yet', 'however', 'nevertheless', and 'nonetheless'. The use of 'because' and 'so' as the causal can be substituted by the use of some alternatives such as 'since', 'for', 'due to', 'as', 'hence', or 'thus'. Lastly, in the use of temporal, there was no significant problem appeared in the texts. It was because one particular temporal conjunction is only used once in a text.
The analysis showed that in the use of conjunctions, the learners tended to rely on the use of certain grammatical cohesion that can be possibly affected by some factors. This might result from inadequate exposures obtained by learners, which led to their limited knowledge of how to apply various grammatical cohesion. This was similar to the result of the study by Rostami et al. (2013) that insufficient exposures of using CDs from writing instructors may contribute to the minimal amount of knowledge and cannot use various CDs in their text. Besides, referring to Iseni et al. (2013), it is stated that when the learners lack choices in applying conjunctions, they will tend to use their sense. This might lead the learners to use inappropriate conjunction, including repetition.
Another possible factor is that the application of conjunction can be affected by the learners' L1. This corresponded with the study by Nugraheni (2015) suggested that the wrong application of conjunctions might be the impact of L1 which appears in the target language (L2). In addition to that, Saif (2012) found that the dominant use of certain grammatical cohesion was prompted by the learners' reliance on their experience in applying the same CDs. This can also be affected by the teachers' instruction on using cohesive devices which are not explicitly spoken. This is also in the same line as the study by Faghih et al. (2013) outlined that instructions on cohesive devices can help EFL learners to advance their ability to apply the CDs. Therefore, it is possible for the learners not to apply various grammatical cohesion because the teacher does not require them to explicitly expand the use of grammatical cohesion.

Cohesive Device Errors
The aspects of error analysis in applying cohesive devices were elaborated in four grammatical cohesion categories, namely reference, substitution, ellipsis, and conjunction. The result shows that the learners' texts perform varieties number of errors. The occurrence of errors are shown in Table  3.  Table 3 shows the information about the number of errors made by learners in each grammatical cohesion category. Based on the data analysis, the learners' total grammatical cohesion in their argumentative essays was 2386, whereas the total number of errors was 175. The percentage of errors compared to the total number of cohesive devices used was 23.8% of the total cohesive devices correctly used by the learners. The errors demonstrated by learners were in the use of two categories of cohesive devices only, namely reference and conjunction. The following excerpts of the examples illustrate the errors. It is important to note that all of the examples presented in the findings are excerpts extracted verbatim from the learners' argumentative essays without any changes or revision. 1) Trusting teens makes them become powerful and gives positive impacts on their future.
2) The more students have chances to take, the more good learning process they will get.
In both excerpts, the learners' errors were described as misuse. In example 1, the learner used the personal reference "they" inappropriately. Instead, the correct use of personal reference should be "them". In example 2, the error was on the use of comparative reference "good". Meanwhile, to show the comparative degree of "good", the correct use of comparative reference should be "the better".
3) Although other forms of contact are possible, the use of mobile phones is still crucial for teens.
The sentence in example 3, shows an error, particularly the unnecessary addition type. The learner should not use both "although" and "but" to express an adversative statement. In this case, the unnecessary addition was on using "but" because it does not change the meaning if it is omitted. The sentence, therefore, became ineffective due to the use of two adversative conjunction. To make it useful, "but" should be omitted. 4) That way, the student might plan easier and sooner. 5) When both the left and right brain are working together, it will be easier to store information.
In both excerpts, the errors were described as an omission. In example 4, the error was in the omission of "the". The article "the" is needed to put before "student" because it is a single countable noun. Additionally, the article "the" is also needed because "student" in the sentence refers to a certain student that has been previously mentioned. In example 5, the error was on the omission of punctuation comma (,) to separate two clauses. By putting a comma between the two clauses, it helps the readers to understand the ideas easily. 6) However, songs also contain some words or contents that are not easy to be conveyed to the students, and teacher should carefully choose the appropriate songs. And, she also should prepare everything before the process of learning and teaching is started.
The error in example 6 was categorized as a redundant repetition of "and" that disrupts the sentence's coherence. The first "and" in example 5 was considered not really appropriate with the context. Instead of using "and", the learner could write "therefore" or "hence" to show the clause "teacher should carefully choose the appropriate songs and prepare everything before the process of learning and teaching is started" was the result or consequence from the previous clause "However, songs also contain some words or contents that are not easy to be conveyed to the students". Additionally, the second "And" in example 6 could be substituted with other additive conjunction such as "Besides,", "Additionally,", or "Moreover".
Based on data analysis, it was found that the errors were the result of reference and conjunction. In the use of reference, it is interesting to note that the most frequent error found in the learners' texts is the inappropriate use of "the" as the definite article. There were two types of error descriptions that occurred, namely omission and unnecessary addition. This result is similar to the other researchers' findings of errors in using the definite article that is frequently used inappropriately by EFL learners. For instance, Yang and He (2016) conducted a corpus-based study examining the errors on the use of definite articles and the possible causes of the errors committed by Chinese learners. The result showed that there are still a lot of errors found in the learners' texts. There are three categories of misuse of the definite article: omission of "the", redundancy of "the", and definite article confusion.
In the use of the conjunction, the most frequent error on the use of conjunction found in this study was the inappropriate use of punctuation commas (,). It is worth noting that commas' inappropriate use frequently occurred in all subcategories of conjunctions: additive, adversative, causal, and temporal. Therefore, this type of errors dominated the occurrence of errors in the use of a conjunction. There were two types of error descriptions of commas' inappropriate use, namely omission and unnecessary addition. Omission errors occurred when the writers did not put commas in the situation where commas are required. On the contrary, unnecessary additional errors occurred when the writers put unnecessary commas in the situation where they are not needed. Consequently, these errors negatively affect the coherence of the text.

Source of Errors
The analysis of what possibly causes cohesive devices errors was contributed by two interlingual transfer and intralingual transfer sources. The distribution of cohesive devices errors based on the source of errors is shown in Table 4.  Table 4 shows that the percentage of intralingual transfer errors is much higher than the errors caused by the interlingual transfer. There was 98.29% of the intralingual transfer result, while the percentage of the errors resulting from the interlingual transfer was only 1.71%. It indicates that the errors done by Indonesian EFL learners were mostly caused by the learners' incompetence in applying the grammar rule of the target language. Brown (2007) points out that most intralingual transfer errors are affected by the learners' experience in learning a new second or foreign language learning system. It means that their existing knowledge might manifest new errors when applying a new grammar rule in composing a sentence.
Meanwhile, the low percentage of errors due to interlingual transfer indicates that the learners are rarely interfered with by their L1 (Bahasa Indonesia). There are three examples of errors caused by the interlingual transfer of L1 interference in the learners' texts. They are errors in the use of "not just...but also" instead of "not only...but also", and the use of additive item "also" in the end of the sentence. These errors might be influenced by the wrong choice of lexical items from the learners as a matter of wrong translation. This leads to the inappropriateness of the sentence in the aspect of semantic context. The possible reason for some of the errors in those groups of examples is that the learners interchanged the use of conjunction both from their L1 and target language. In fact, some of them cannot be used with the same application due to the different terms and rules.

DISCUSSION
The aforementioned results related to the investigation of how Indonesian EFL learners apply cohesive devices are closely related to the process of second language acquisition. Findings showed that applying particular grammatical cohesion still becomes problematic for the learners. The major factor contributing to the repetitious use of particular grammatical cohesion is the restricted learners' repertoire of grammatical cohesion. The limited knowledge about grammatical cohesion might suffer due to a lack of exercise given by the teacher. Support this, Rostami et al. (2016) declared that the teacher needs to give sufficient exercise for the learners regarding the use of grammatical cohesion. By doing the exercise, the learners will know whether he/she has enough knowledge and whether he/she has difficulties in doing the exercise. This can motivate them to go deeper into the use of grammatical cohesion in their writing. Moreover, the learners' limited use of grammatical cohesion might be affected by the learners who tend to apply grammatical cohesion that they are convinced how to apply them appropriately. They would not take the risk of trying to use various kinds of grammatical cohesion unfamiliar to them. This is similar to the study by Modhish (2012) that some learners were afraid of making mistakes, so they were reluctant to apply unfamiliar CDs which they have never been used before.
Hence, based on the revealed results, some broad pedagogical implications can be applied by EFL teachers to solve similar problems. Firstly, it was found that the most frequent errors committed by Indonesian EFL learners were on the use of the definite article. This corresponds with the studies by Hasan and Marzuki (2017) and Nasser (2018) that the use of articles still becomes grammatical problems for EFL learners. Referring to Crompton (2011), definite articles inappropriately are the most common error in English language learning. Especially for learners from a country whose language does not have article systems, such as Indonesia, will have increased possibilities to face more definite article use problems. Therefore, even though the dominant source of the definite article is due to the learners' intralingual transfer, there is a possibility that the absence of an article system in Bahasa Indonesia might interfere learners in using the L2 article (Derakhshan & Karimi, 2015). The distinction of the definiteness of whether or not to apply definite articles must be emphasized to the learners from earlier stages of learning the language. Crompton (2011) also suggests that the use of articles should be emphasized through reading and writing development since it provides the repetition of article use that occasionally the learners will generalize them.
The second issue is that the errors in the use of commas have dominated the conjunction errors. Regardless, a comma functions merely as a small unit of composition, yet it is still a necessary writing aspect. This study's finding has a similar result to the study by Husada et al. (2018) that Indonesian students were found dominantly make errors in using a comma in their compositions. Similarly, Awed (2012) and Salman et al. (2017) have also found that comma has become the most common error committed by EFL learners in writing compositions. These studies' findings have provided some broad pedagogical implications, one of which is that a significant amount of classroom time should be devoted to grammar teaching, which directs students to use punctuation explicitly. Referring to Widiati and Cahyono (2006), teaching grammar should allow learners to use linguistics form accurately, meaningfully, and appropriately. This implies that the most important of grammar teaching is not in the learning grammar process but in the acquisition process, where EFL learners can appropriately prevail what they have learned in real communication use. It should be emphasized by exposing the learners with some exercises and continual practices when they are required to compose a text, as suggested by Awed (2012) and Murtiana (2019).
Collaborative learning is also believed to help learners improve their writing skills effectively (Challob et al., 2016;Pratiwi, 2020;Rizki et al., 2020;Supiani, 2012). Supiani (2012) has demonstrated results of his study that collaborative learning is beneficial to help learners improve their learners' writing competence, behavior, motivation, and class situation. Other fruitful benefits of collaborative learning techniques are improving learners' selfconfidence and active participation (Pratiwi, 2020), reducing writing apprehension (Challob et al., 2016), as well as enhancing learners' writing ability in composing a text (Rizki et al., 2020). These results of studies have implied that the use of collaborative learning techniques can be used as solutions in solving problems related to the use of grammatical cohesion in writing. Several examples of collaborative learning techniques that can be used in EFL classrooms are group discussions, write-pair-share, group problem solving, round table technique, and team-based learning, focusing on discussing the use of cohesive devices.
Another pedagogical implication that is worth considering is related to using a mapping strategy to improve the learners' skills in applying cohesive devices appropriately. The learners are encouraged to apply the process of writing from pre-writing, drafting, revising, and editing. A study conducted by Sulistyo and Ningsih (2015) proves that semantic mapping strategy was essential to improve the learners' writing compositions. They are also facilitated to find out the most appropriate cohesive devices that might be used to connect their ideas. This strategy can be given in all writing steps, whether in prewriting, drafting, revising, and editing. However, the learners will likely to make frequent errors in the process of drafting. Hence, feedback plays a vital role before the learners revise their drafts. Support this, Rahmawati et al. (2018) suggest that an adequate amount of feedback from teachers and the learners' peers can motivate them to write better. Therefore, EFL teachers should consider giving a sufficient amount of feedback to the learners to be aware of applying cohesive devices in their compositions.
The learners' incompetence in applying cohesive devices can also happen due to the insufficient time devoted to teaching cohesive devices. It often happens because, in one semester, learners are required to learn several topics and, thus, cohesive devices are only small bits of the materials that should be delivered. In regards to this, learners cannot depend on the classroom meetings and materials provided in the textbook due to the limited classroom meeting time. Referring to Wihastyanang et al. (2014), one of the solutions is using the Learning Management System (LMS). It is found appropriate to be applied in teaching English writing by making use of the internet and technology which are considered to facilitate an active and interactive teaching and learning process. The teachers can encourage learners to find out supplementary materials related to the use of cohesive devices in any platform they like from the internet using their own learning strategy. This also corresponds with the  Yulianti (2018) suggested that the learners will be able to study maximally through the strategy which they prefer to use.
The findings have shown that EFL learners find problems in using grammatical cohesion appropriately and effectively in writing an essay. In fact, as a requirement for good writing, the use of grammatical cohesion enables the writer to produce a coherent text and get the message across to the readers effectively. However, due to the insufficient knowledge of grammatical cohesion, the ineffective and inappropriate use of grammatical cohesion may disturb the readers in interpreting the meaning or the message, although the meaning may be successfully interpreted. Even though errors made by EFL learners are considered very common, it would be much better if the number of errors can be minimized to make a better quality of writing. To help EFL learners develop their abilities to apply grammatical cohesion in writing, the aforementioned strategies might be used by EFL teachers' in their EFL writing class.

CONCLUSION
This study concludes that the application of grammatical cohesion in writing becomes problematic for Indonesian EFL learners. There was a heavy reliance on the use of particular grammatical cohesion, and a number of errors were found in the students' texts. The dominant factor that significantly contributed to the learners' error was intralingual transfer related to the students' proficiency in grammar mastery. This implies a need for EFL teachers to address more effective teaching strategies to solve the problems. This study investigates the use of cohesive devices in EFL students' compositions with a limited number of data sources, namely twenty argumentative essays. Hence, conducting a similar study involving a bigger number of sources of data is highly recommended for future researchers. Furthermore, this study also involved only the researchers as raters. A more effective and efficient way of investigating the use of cohesive devices is thus suggested for scholars who have a similar interest and focus using a concordance software program to help in the data analysis process.

DECLARATION OF CONFLICTING INTERESTS
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.