Bilingual and Monolingual EFL learners’ Use of Writing Metacognitive Strategies and Writing Performance

Introduction

Writing is a versatile tool which can be employed to achieve different purposes, from creating imagined worlds and telling stories to sharing information (Graham, Gillespie, & McKeown, 2013). However, based on Naghdipour (2016), almost all Iranian learners of English have difficulty in writing skill. One of the ways which help improve writing ability is using writing strategies. As Riduan and Abdullah (2009) state, the most important factors in producing good writing or compositions are the types and amount of strategies used. The role played by strategies in writing is so important that Luchini (2010) viewed writing as a wide range of cultural, social, cognitive, meta- cognitive, interpersonal, and linguistic strategies. In this study writing metacognitive strategies are examined. These strategies, based on Richards and Schmidt (2002, p. 329), are “a category of learning strategies which involves thinking about the mental processes used in the learning process, monitoring learning while it is taking place, and evaluating learning after it has occurred”.

Emig (1971) emphasizes that writing needs to be learned only with formal and systematic training. Learning to write in a new language is not just learning to put down ideas in a new code (Shen, 2003). In other words, writing in one’s mother tongue may not be similar to that of their L2 and FL (Kroll, 2003). In fact, writing processes in L1, L2, and L3 and their requirements may be different. However, research in ESL contexts has generalized and applied findings of L1 studies (Cumming, 1998; Silva, 1993). Based on Raimes (1987) and Arndt (1987), it is not advisable to use L1 findings in and generalize them to L2 contexts on the grounds that writing processes and strategies used by L1 and L2 learners are different. Therefore, along Rimes’s (1987) and Arndt’s (1987) line of discussion, it may not be appropriate to use the findings of studies on L1 and L2 writing to address writing problems in L3, especially if the L3 is learnt in an EFL context. One of the sources of difference in the underlying processes of L1, L2, and L3 may be the number of languages one knows; bilinguals have better cognitive development (Cenoz, 2003), higher metalinguistic awareness (Herdina & Jessner, 2002; Kuile, Veldhuis, Van Veen, & Wicherts, 2011), increased syntactic awareness (Davidson, Raschke, & Pervez, 2010; Foursha-Stevensonv, 2011), and improved communicative skills (Kuipers & Thierry, 2015; Dewaele, 2010). The main aim of this study is to examine if bilingualism makes any difference in learners use of writing metacognitive strategies and writing performance.

To date, an abundance of studies on L3 learning/ acquisition has been conducted in different parts of the world (see Cenoz, 2003 for a summary of studies in these places). Recently, some Iranian scholars have shown interest in investigating differences between monolinguals and bilinguals in learning English as an additional language. They have investigated different areas of L3 including, learning imperative verbs (Sobhani & Vaysi, 2015), reading comprehension and its relation to strategies (Afsharrad & Sadeghi Benis, 2017; Tafarroji Yeganeh & Malekzadeh, 2015; Keshavarz & Ghamoushi, 2014; Maghsudi & Talebi, 2009; Modirkhamene, 2006), relative clauses (Moghtadi, Koosha, & Lotfi, 2015), English dictation performance (Poorsoti & Assadi Aidinlou, 2016), vocabulary learning (Keikhaie, Khoshkhoonejad, Mansoorzadeh, & Panahandeh, 2015; Zarghami & Bagheri, 2014; Zargosh, Karbalaei, & Afraz, 2013; Zare & Davoudi Mobarakeh, 2013; Kassaian & Esmae’li, 2011; Keshavarz & Astaneh, 2004), pragmatic awareness (Rahimi Domakani, Hashemian & Mansoori, 2013), grammar (Yeganeh, Ghoreyshi, Darabi, 2013), and language learning strategy use (Sa’di, Sa’di, & Zarin Shoja, 2013; Seifi & Abdolmanafi Rokni, 2013). Despite the abundance of studies on bilingualism, this important skill (writing) has not been addressed properly in Iranian context, a gap that this study is set out to fill.

Moreover, considering diversity in L3 studies (Cenoz, 2003), the results of studies conducted in different contexts and with different languages should be used cautiously. The findings of the aforementioned studies may not be applicable to all Iranian settings because of the combination of variables (e.g., gender, level of proficiency, and especially the background language of the participants), which have major effects on the findings. This reveals the need for more studies in different areas related to L3 learning, which is one of the aims of this study.

Besides the scarcity of research on L3 writing in Iran and diversity in L3 studies, which indicate the need for more L3 studies, there is another reason to conduct research like the current study. The number of studies in different areas of L3 is limited and the findings of these studies are not consistent enough, making it hard for teachers and decision makers to use the findings for the purpose of decision making. A number of related studies are reviewed in the next section.

Review of literature

As mentioned in the previous section, L3 studies conducted in the context of Iran have produced inconsistent results. The following review of literature reveals the inconsistency. Seifi and Abdolmanafi Rokni (2013) compared 25 Turkmen-Persian bilinguals with 25 Persian monolinguals, in language learning strategies. The results of their study revealed that bilinguals used more cognitive and metacognitive strategies than monolinguals. They stated that the difference might have been due to bilinguals’ experience with and success in mastering two languages.

Afsharrad and Sadeghi Benis (2017) compared 50 Turkish-Persian bilinguals with 36 Persian monolinguals across gender for their cognitive, metacognitive, and total strategy use as well as reading comprehension performance. Their findings revealed that bilinguals used more metacognitive strategies and had better reading comprehension scores than monolinguals. However, the two groups were not different in terms of cognitive and total strategy use. They related the insignificant difference of bilinguals and monolinguals in using cognitive strategies and yet significant difference in metacognitive strategies to the nature of the bilinguals’ L2, which lacks written modality. The bilinguals of their study, like the bilinguals of this study, had not received more written input than the monolinguals and, as a result, did not have a better performance in written cognitive strategy use. They explained the better performance of bilinguals in using metacognitive strategies by their experience with learning two languages which improve their ability of planning and monitoring the process of L3 learning. They believe that the better performance of bilinguals in reading ability may be mediated by their more effective use of metacognitive strategies.

The results of the study by Sa’di, Sa’di, and Zarin Shoja (2013) produced similar results. They compared 50 Turkmen-Persian bilinguals and 50 Persian monolinguals in their use of learning strategies. The participants were male high school students. The results of their study showed that there was no significant difference between the two groups in general. But bilinguals reported to use more metacognitive strategies than their monolingual counterparts. The researchers have not discussed why bilinguals of their study were better than monolinguals in using metacognitive strategies and also why there was no difference between the two groups in other strategies.

In a longitudinal survey, Modirkhamene (2006) compared reading achievements of 42 Persian monolinguals and 56 Turkish-Persian bilinguals in three phases. The results of her study revealed better performance of bilinguals in all phases. The author related better performance of bilinguals to their experience in language learning. Bilinguals, who have already accomplished the complex task of learning two languages, have developed a competence to tackle the task of learning an L3. This competence makes language learning process easier for them compared to their monolingual peers. Moreover, greater metalinguistic awareness and using knowledge of two languages facilitates the process of L3 learning for bilinguals. As the author put, superiority of bilinguals may also be related to their effective use of cognitive and metacognitive strategies as well as their active use of their languages in different contexts.

Keshavarz and Ghamoushi (2014) compared 100 Turkish-Persian bilinguals with 100 Persian monolinguals in terms of metacognitive reading strategy awareness. Their results showed that bilinguals were more aware of total reading strategies (supportive, global, and problem-solving strategies). Moreover, bilinguals reported to use more strategies than monolinguals in global strategies. However, in the other two strategies, i.e. supportive and problem-solving strategies, there was no significant difference between the two groups. They suggested learners’ attitude towards reading and proficiency level as variables that may have resulted in differences between the two groups.

Maghsudi and Talebi (2009) investigated differences between monolinguals and bilinguals in their awareness and use of cognitive, metacognitive, and total strategies across proficiency levels. Their findings suggested significant differences between bilinguals and monolinguals in cognitive and metacognitive strategies. In addition, more proficient learners excelled less proficient ones in all strategies. However, the interaction effect of proficiency level and language background was not significant for any of the strategies. They have not elaborated on the reasons why their bilingual and monolingual as well as higher level and lower level participants were different. However, their study suggests that bilingualism and proficiency level might be plausible reasons for the differences they found.

Modirkhamene (2011) examined differences in cross-linguistic transfer of trilinguals (knowing Turkish, Persian, and English) and bilinguals (knowing Persian and English), each divided into two groups of high-proficiency and low-proficiency. She found a high correlation between the writing ability of the languages (Persian and English) the participants knew. Moreover, trilinguals were significantly better writers than their bilingual counterparts. However, her qualitative analysis showed that both trilinguals and bilinguals thought more in Persian. In other words, Persian which had been learnt both orally and academically, compared to Turkish which had been learnt only orally, was more frequently referred to as the base language while thinking to write. The author explained better performance of trilinguals to their wider world knowledge and more language learning experience.

To the best of our knowledge, Modirkhamene’s (2011) work is the only study comparing differences between monolinguals and bilinguals in L3 writing. She compared the two groups from a cross-linguistic transfer point of view. The importance of this study is that it considers the issue from a different perspective. We examined the metacognitive benefits that bilingualism may have to the process of L3 writing not the transfer of skill from previously known languages to the new language.

This study examines general effects of bilingualism rather than cross-linguistic transfer. Many studies on the effect of bilingualism have investigated transfer from previously known languages to the target language. Most of these studies have examined languages with alphabetic writing systems (Geva & Siegel, 2000; Gholamain & Geva, 1999; Wade-Woolley & Geva, 2000; Arab-Moghaddam & Senechal, 2001), with some studies examining transfer across languages with different writing systems (e.g. Bialystok, McBride-Chang, & Luk, 2005). However, compared to languages with a writing system, the number of studies investigating languages which lack a writing system and are only spoken is limited. In this study, we examined a different type of language, a language with no written form. Therefore, it can shed some light on the benefits (other than transfer from L1) that bilingualism has in the process of L3 writing. In other words, since the bilinguals’ L1 does not have written modality, nothing can be directly transferred from their L1 (Turkish) to L3 (English) as far as writing is considered. Moreover, if there is anything to be transferred from their L2 (Farsi) to L3 (English), the monolinguals also have had that benefit since both groups had received the same instruction in Farsi. This combination of languages provides a unique situation to determine indirect/ general benefits of bilingualism rather than direct transfer from previously known languages. In this study we examine how bilingualism affects bilinguals’ use of writing metacognitive strategies and writing performance.

Another main importance of this study is, as suggested by Cenoz and Gorter (2011), to focus on both SLA and bilingualism under the terms multilingualism or L3 rather than considering them separately. In other words, this study is an attempt to bridge between and bring together two fields of second language acquisition and bilingualism which have traditionally ignored each other. (Cenoz, 2003, Cenoz, & Gorter, 2011). The latter of these has focused more on the product of bilingualism (difference between bilinguals and monolinguals) and the former on the process of acquiring a second language. These two have been brought together in a new field of study named L3 acquisition. Although both L2 and L3 learning have a lot in common in that both focus on learning additional languages, they are different in that L3 learning process is influenced by the outcomes of bilingualism, what brings the two fields of bilingualism and SLA together (Aronin & Hufeisen, 2009).

In short, this study is designed to bridge the gap in L3 research in Iran; in spite of a growing number of studies on L3, scant attention has been paid to writing skill and metacognitive writing strategies and there is a gap in our existing knowledge about these areas (L3 writing metacognitive strategies and L3 writing). The current study aims to examine differences that may exist between monolinguals and bilinguals across proficiency levels in using writing metacognitive strategies and writing performance.

Research Questions

1. What are common differences between monolingual Persian and bilingual Persian–Turkish EFL students in their use of writing metacognitive strategies across proficiency levels?
2. What are common differences between monolingual Persian and bilingual Persian–Turkish EFL students in their writing performances across proficiency levels?

Method

Participants

The participants of this study were freshman and senior students of TEFL. The freshman had already passed a grammar course. The seniors had passed two courses on grammar and one advanced writing course. They were doing their essay writing course in the semester the data collection process took place. A total of 230 freshman and senior students took part in this study. Data obtained from 54 of the participants were excluded since they had not answered some questions of the questionnaire. Finally, data from 176 participants were analyzed. 88 of these students were bilingual (male= 29 and female=59) and 88 were monolingual (male=13, and female=75). The number of participants is presented in Table 1.

In this study senior students were considered as higher academic level and freshman as lower academic level participants. Based on the data collected through background information questionnaire, the senior participants were 22-25 years old and the freshmen were 18-21 years old.

The bilingual participants were from the University of Tabriz, the University of Marahgheh, and Shahid Madani University in Tabriz province. The monolingual participants were students from the University of Isfahan and Sheikhbahaee University in Isfahan. Monolingual students studying in Tabriz were assigned to the group of monolinguals and bilingual students studying in Isfahan were assigned to the bilingual group. Since there are different ethnic groups in Iran speaking different languages (Turkish, Kurdish, Arabic, etc.), having a sample which represented all bilingual population in Iran was almost impossible. Therefore, the participants of this study were chosen through convenience sampling, based on their accessibility to the researcher. Data were collected from three major universities in Tabriz province, where the researcher teaches and lives, and two major universities in Isfahan, where the first researcher is doing her PhD.

The bilingual participants of the study had learnt spoken Turkish as their native language in natural settings. They had started learning both spoken and written Farsi as their L2 in the first year of primary school. On the other hand, the monolingual participants knew only one language, Farsi. They had a good command of spoken Farsi before going to school. Therefore, they learned only written Farsi in the first grade of primary school, unlike the bilinguals who had to learn both spoken and written Farsi in their first grade.

Instruments

The following three instruments were used to collect data from the participants:

1. Background information questionnaire, which was developed by the researchers in order to identify participants’ language background, proficiency level, socioeconomic status, gender, and self-evaluation of English proficiency level.
2. Writing metacognitive strategy questionnaire (see Appendix), which was adapted from Petric and Czarl’s (2003) writing strategy questionnaire and also the questionnaire used by Peñuelas (2012). As suggested by Dornyei and Taguchi (2010), to make sure that the participants, especially those with lower levels of proficiency, had no problem understanding the questionnaire, it was translated into Persian (see the questionnaire in the Appendix). The reliability of the Persian questionnaire was 0.71, which is, based on Jackson (2006), deemed to be a strong reliability.
3. The writing prompt. In order to assess the writing performance of the participants, they were asked to write a passage of about 250 words about “social networking applications and websites: benefits and disadvantages”. This topic was chosen on the grounds that it is a common concern of many people and students are familiar with their advantages and disadvantages.

The questionnaires were pilot tested and problematic items were either changed or deleted.

Procedure

In order to ensure that an appropriate proportion of bilinguals and monolinguals took part in the study, data were collected from universities of two provinces in Iran: three universities in Tabriz, where Turkish is spoken as L1 and Persian as L2 and two universities in Isfahan, where people mainly speak one language, Persian. The final decision on language background of the participants, however, was made based on the data elicited through the background information questionnaire. There were some participants in Tabriz, who reported to be monolinguals and some in Isfahan who reported to be bilinguals.

After getting permissions from university authorities for collecting data, the first researcher attended the universities at agreed times. Collecting data from each class took a complete ninety-minute session. At the beginning of each session the researcher ensured the participants that the data would be confidential and that it would be used only for research purposes. They were also encouraged to ask questions, if needed. Then, the questionnaires and the paper for writing were distributed. The participants were asked to write one or more paragraphs based on the given instructions. As determined in the piloting stage, they were given 40 minutes to do this task. Then, they completed the background information questionnaire, and finally the metacognitive writing strategy questionnaire. They completed the questionnaires in about 40 minutes.

After data collection was complete, they were submitted to statistical package for social sciences (SPSS version 21) for the purpose of analysis, the results of which are reported in the next section.

Results

In this study there were two independent variables, namely language background and proficiency level. The participants who could speak both Turkish and Farsi were considered as bilinguals and those who could speak only Persian as monolinguals. Seniors were considered as higher academic level participants and freshmen as lower academic level ones.

Considering the variables of the study, ANOVA seemed to be the best statistical procedure to compare the means of the two groups. Therefore, this test was used to compare the groups in using strategies. For comparing the writing scores of the participants, however, two Kruskall-Wallis tests were employed since the data were not normally distributed. The results are reported in detail in the following sections.

Metacognitive Strategies

First, descriptive statistics were computed for metacognitive strategies. The results are presented in Table 2.

The results of the two-way ANOVA, examining the effect of language background and level on participants’ use of metacognitive strategies found a statistical difference for the main effect of language background F (1, 172)=7.27, p=0.008, =0.04. An inspection of the mean scores in Table 2 revealed that bilinguals (M=3.26, SD=0.50, N=88) used more metacognitive strategies than monolinguals did (M=3.05, SD=0.52, N=88).

However, the main effect of participants’ level on their use of metacognitive strategies was not significant; F (1, 172)=0.08, p=0.782, =0.000. Neither was the interaction effect of language background and level on the use of metacognitive strategies F(1, 172)=1.15, p=0.285, =0.007.

Writing Performance

First descriptive statistics were conducted for the four groups. The results are presented in Table 3.

Since there were two independent variables (level and language background), a two-way factorial ANOVA was the appropriate procedure to compare the means. Before running the ANOVA, the data were examined for the underlying assumptions parametric tests. The results of the data analysis revealed that the equality-of-variances assumption was met, however, the data were not normally distributed, meaning that parametric tests would not be appropriate for the purpose of data analysis. Therefore, based on Larson-Hall (2010), two non-parametric Kruakall-Wallis tests were used to compare the groups: one to compare the writing performances of monolinguals and bilinguals and the other to compare that of higher level and lower level participants. In the Kruskall-Wallis test, mean ranks and medians rather than mean scores are used as measures of central tendency.

The results of these descriptive statistics for language background are presented in Table 4.

The results of the first Kruskall-Wallis test revealed a significant difference between the writing scores of bilinguals (mean rank=101.36) and monolinguals (mean rank=75.64), H (1)=11.21, p=0.001, r=.85.

The second Kruskall-Wallis examined differences between higher level and lower level learners in terms of writing performance. Descriptive statistics are presented in Table 5.

The results of the Kruskall-Wallis test indicated that participants with higher academic level (mean rank=101.14) had statistically significantly better performance in writing than lower level ones did (mean rank=75.86), H (1)=10.84, p=0.001, r=.82.

Discussion

The correction of the participants’ compositions revealed that all four groups had very little knowledge about the rhetorical structure of writing in English. They often transferred Persian rhetorical structures while writing in English. Most of the participants developed an idea of writing as a set of sentences which describe something without any understanding of the fact that any piece of writing should be pre-planned and organized with specific rhetorical structures in different language. Most of the learners in the four groups did not have a clear organization and planning for their writing. They did not have separate paragraphs devoted to introduction or conclusion. In fact, most of the compositions were an unorganized list of sentences talking about some aspect of the topic.

Examining the questionnaires of the participants revealed that bilinguals used more metacognitive strategies than monolinguals. However, there was no difference between higher level and lower level learners in using these strategies. The interaction effect of language and level was not significant either. Also, bilinguals and higher level learners had better composition scores than monolinguals and lower level learners, respectively.

Bilinguals’ more frequent use of metacognitive strategies may be related to their knowledge of and experience with two languages. When receiving language through the receptive skills of reading and listening or producing language through speaking and writing, the process is constantly monitored and checked (metacognitive strategies). Considering that bilinguals have learnt two languages, they have had more opportunities to use metacognitive strategies. Moreover, they have to regularly monitor what they want to produce in order not to mix the languages they know. Receiving linguistic data is also harder for bilinguals since they have to determine to which language belongs what they hear while this is not the case for monolinguals. In short, manipulating and dealing with two languages helps bilinguals improve their ability of planning for their own learning and monitoring their process of learning.

The findings are in line with those of previous research and confirm the idea that bilinguality fosters using metacognitive strategies (Seifi & Abdolmanafi Rokni, 2013; Afsharrad & Sadeghi Benis, 2017; Sa’di, Sa’di, & Zarin Shoja, 2013; Keshavarz & Ghamoushi, 2014; Maghsudi & Talebi, 2009). This seems to be the result of experience with and knowing two language systems. As Afsharrad and Sadeghi Benis (2017) state, bilinguals receive aural input in two languages which provides them with “more food of thought”. Therefore, they are more experienced in planning to act upon, monitoring, and controlling what they hear. This helps them improve their use of metacognitive strategies as compared to monolinguals. Moreover, similar findings of this study and that of Afsharrad and Sadeghi Benis (2017) indicate that the order of acquiring the languages under investigation in these two studies (Persian and Turkish) does not influence the bilinguals use of metacognitive strategies. In this study, Turkish was learners’ L1 and Persian was their L2, while in Afsharrad and Sadeghi Benis’s (2017) work participants’ L1 was Persian and their L2 was Turkish.

Drawing on Modirkhamene (2006), better performance of bilinguals can also be accounted for by their experience in language learning. Bilinguals, who have already accomplished the complex task of learning two languages have developed a competence to tackle the task of learning an L3. This competence makes the process easier for them compared to their monolingual peers. Moreover, greater metalinguistic awareness and using knowledge of two languages facilitates the process of L3 learning for bilinguals.

The findings of this study also confirm those of previous research which suggests that better reading ability of bilinguals as compared to monolinguals might be related to their better use of reading metacognitive strategies (Maghsudi & Talebi, 2009; Afsharrad & Sadeghi Benis, 2017). The findings of this study revealed that such a relationship might exist between writing metacognitive strategies and writing performance. In other words, a close examination of the results (better performance of bilinguals in metacognitive strategy use and writing performance) suggests the idea that the better writing ability of bilinguals might be mediated by more use of metacognitive strategies. This supports Meichenbaum and Biemiller (1998) in that improved self-regulatory behavior contributes to better performance of more successful learners.

Comparing the performances of higher level and lower level participants of this study indicates that careful attention has not been devoted to teaching writing metacognitive strategies in the Iranian educational system. In spite of passing more writing courses, higher level learners were not any better than lower level ones in using metacognitive writing strategies.

Comparison of higher and lower level participants also reveals that better writing scores of higher level learners cannot be explained by their pattern of metacognitive strategy use since higher and lower level participants were not different in using metacognitive strategies. Drawing on Maghsudi and Talebi’s (2009) study, the difference of higher level and lower level participants in this study might be related to differences in using cognitive and total strategies, which were not examined in this study. They found a significant difference between higher level and lower level participants in reading comprehension as well as cognitive and metacognitive strategies. They explained better reading ability of their participants by their better use of strategies. Higher level and lower level participants in this study were different in writing ability. However, this difference could not be accounted for by their use of metacognitive strategies. These findings depart from those of previous research. Based on previous research, (Chien, 2010; Ridhuan & Abdullah, 2009), higher level learners usually make more use of metacognitive strategies while lower level ones use of these strategies is limited and not as effective. The findings are, however, in line with those of Baker and Boonkit (2004), who found no significant difference between higher level and lower level learners in how frequently they use strategies.

Unlike the difference between writing scores of higher level and lower level participants, which could not be explained by their use of metacognitive strategies, better composition scores of bilinguals could be accounted for by their more frequent use of metacognitive strategies. In the light of the findings of the current study, we recommend teachers to help learners improve their writing ability by using writing metacognitive strategies more effectively. In addition to metacognitive strategies, other factors such as learners’ cognitive, affective, and social strategic behavior might also have contributed to differences in composition scores. Despite the emphasis on the importance of metacognitive strategies (Anderson, 2002), there’s a consensus among researchers (Garner 1994; O’Neill 1992; O’Neill & Todaro 1991) that learners benefit more when they use metacognitive strategies along with other strategies (say, cognitive strategies). Hence, it might be more advisable to teach metacognitive strategies in parallel with other writing strategies.

Although different studies have been done in the area of L3, considering different combinations variables (e.g. background language, proficiency level, age, etc.) and also contradictory results, there’s a need for more studies, especially more comprehensive ones in which different types of strategies are examined together. Research has shown that examining strategies separately might not reveal a difference between two groups while a combination of the same strategies and considering them as one variable might indicate some differences (Afsharrad & Sadeghi Benis, 2017).

Conclusion

The large number of bilingual people in Iran and the variety of languages they speak has attracted considerable attention and researchers have investigated bilingualism and its consequences from different points of view. However, the role played by bilingualism in L3 writing has not been addressed adequately yet. In this study differences between bilinguals and monolinguals in terms of writing metacognitive strategies and writing performance was examined. Bilinguals reported more use of metacognitive strategies and had better writing ability. However, participants with higher academic level and lower academic level were not different in their metacognitive strategic behavior, although higher level participants had significantly better composition scores than lower level ones.

The findings of this study add to the line of research supporting benefits of bilingualism. Previous research revealed better performance of bilinguals in learning strategies and metacognitive reading strategies. Based on the findings of this study it can be concluded that bilingualism develops bilinguals’ metacognitive strategy use and improves writing performance in L3. This implicates that metacognitive strategies may play a mediating role in improving L3 writing skill. In addition, no difference between higher and lower academic level learners’ use of writing metacognitive strategies may be an implication of inappropriate strategy training at Iranian universities. Based on the findings of this study, instructors are recommended to teach writing metacognitive strategies as one of the ways to help learners with their writing ability.

This study had some limitations, none of which jeopardize the validity of the study. First, the participants’ gender was not taken into account in data analysis since the researchers could not convince enough male students to take part in the study. Therefore, the findings should not be used exclusively for males or females. Second, it was not possible to examine the interaction effect of language and academic level on writing performance. Although Kruskall-Wallis is a widely used test, it does not examine the interaction effect of independent variables. Both of these limitations can be addressed by future research. A replication of this study in which gender is considered as an independent variable can shed light on how gender may affect learners’ use of writing metacognitive strategies and writing performance. Moreover, further research with normally-distributed data enables us to understand the interaction effect of academic level, linguality, and gender on metacognitive strategy use and writing performance. In this study the difference between the writing scores of higher and lower level participants could not be accounted for by their use of writing metacognitive strategies. Future research can also explore other factors (such as cognitive, social, and affective strategies) as possible sources of difference between the two groups’ writing skill.