UNCOVERING METALANGUAGE IN GRAMMAR EXAM AND ITS IMPLICATION ON STUDENTS’ COGNITION

As language scholars, have you heard the word ‘metalanguage’? and how about the effect of using metalanguage? The current study aimed to investigate the categories of metalanguage in grammar exam. A total of 17 master students in TEFL were voluntarily recruited as the subjects. Their exams were analyzed in detail to collect the data. The researchers conducted a qualitative inquiry called content analysis. The result exposed that the students produced two categories of metalanguage namely technical and non-technical metalanguage. Based on its frequency, technical metalanguage was produced much more frequently than non-technical metalanguage. Both categories of metalanguage affected students’ cognition. However, technical metalanguage ‘drained’ their cognition a lot more while they were not aware of it.


Introduction
The term 'metalanguage' may not be sound familiar even among language scholars. This term shares some familiar characteristics with 'grammar terminology' or 'metalinguistic terminology'. Indeed, in recent decades, a number of experts have already proposed the definition of these three terms. However, as expected, some definitions seem to overlap each other's.
It is quite interesting how the vague notion of those three terms does not make any teachers and lecturers neglect to use it (Alderson & Hudson, 2013;Berry, 1997Berry, , 2014Borg, 1999;Doherty & Perner, 1998;Hu, 2011a;Reder, Marec-Breton, Gombert, & Demont, 2013;Venuti, 2015). Grammar, as a subject course, is the ideal place where this phenomenon can be observed thoroughly. Whether it is during the teachinglearning process or during the tests, metalanguage is produced and used frequently. Some of those tests are Language Awareness test (1999a), Grammaticality Judgment Test (Ellis, 2004), and the Rule Verbalization Task (Hu, 2011a). Hence, those circumstances imply the importance of metalanguage for learning and testing purposes. It can be said that the use of metalanguage in grammar classes is inevitable. Unfortunately, the studies or the references on metalanguage were rather insufficient. Many of language scholars may not recognize the potential or the effect of metalanguage.
However, the use of metalanguage is still debatable among scholars. On the one side, metalanguage is needed by learners while they conduct the explicit discussion of the structural and the functional features of highly complex structures even in a Communicative Language Teaching (CLT) classroom (Hu, 2011b). On the other side, either producing or learning metalanguage adds an extra cognitive burden for learners (Mohamed, 2012). Hence, the decision whether the use or not use metalanguage in a classroom is the teachers' responsibility even though it will quite difficult to avoid it particularly in grammar class.
Some scholars (Basturkmen, Loewen, & Ellis, 2002;Berry, 2010) argued that metalanguage is more appropriate for advanced learners. Here, the advanced learners refer to the L2 students who have gained the 'understanding' to analyze and reflect some aspect of grammatical rules (Schmidt, 1990). Indeed, these learners are different from other learners who only have the ability to 'notice' or 'know' the grammaticality of sentences. In brief, the levels (beginner-intermediate-advance) are classified based on their knowledge.
At least, there were two pioneers' studies that uncovered some notable findings of metalanguage. First, Basturkmen, et al., (2002) investigated the use of metalanguage by the teachers and the students in focus on form classroom. One of their questions was what kind of metalinguistic terms do the students and the teachers use. Their study involved 12 hours of observations in a private English-language school, New Zealand. In total, there were 24 teenagers with mixed nationalities and 2 teachers. Then, the data was analyzed through focus on form episodes (FFEs). The results showed that the most common terms were primarily of a non-technical nature such as mean, word, name, question, say, etc. Precisely, it consisted of 188 terms.
Second, three years later, Fortune (2005) conducted a similar study. He compared the employments of metalanguage between advanced and intermediate L2 learners. Those 56 learners had mixed L1 (Italia, Arabic, Spanish, Korean, Japanese, Georgian, etc.). The data were gained from 4 meetings of the students' group interactions in Dictogloss (a form-focused collaborative writing task). Furthermore, he used Language Related Episodes (LREs) as the analytical framework. As noted, LREs sometimes called as FFEs. For the results, in general, there were 100 technical and 240 non-technical terms which were used by the students. In a comparison, advanced learners employed technical metalinguistic terms more frequently than the intermediate leaners.
The present study aims to identify the categories of metalanguage in a grammar exam done by TEFL students. Theoretically, this study may become one of the pioneers which investigate metalanguage in Indonesia. Practically, for the lecturers and teachers, this study may provide a new perspective related to the students' cognition and how it works. Indeed, language teachers and lecturers have gained various experiences in teaching and examining the students' cognition. However, there is a possibility that some of them do not realize how the students' cognition work. From this refection, the teachers and the lecturers are able to formulate the better learning for their students. On the other side, for the students, this study gives an opportunity to reflect insight their cognition. Understanding the cognition is not only about strength but also weaknesses. It is quite possible that there are many students who cannot reconstruct their knowledge since they do not aware of their weaknesses. By involving in this study, the students are expected to reconstruct the cognition based on their weaknesses.

Theoretical Underpinning Metalanguage and Its Categories
Metalanguage is language which is used to analyze or describe language (Johnson & Johnson, 1998). Indeed, the word 'language' contains broad aspects such as words, phrases, sentences, structures, etc. However, for the purpose of the present study, the notion of metalanguage was limited as any words used to explain grammatical rules (Ellis, 2004). Basturkmen et al., (2002) distinguished metalanguage into two categories: technical and non-technical. Metalanguage technical terms are items likely to be found in a grammar book or linguistic reference and are more likely to be used by a limited section of the population, such as language teachers or linguists (verb, adverb, past tense, etc.). In contrast, non-technical terms are words that have common usage within a specific context. However, the distinction between technical and non-technical terms was not always straightforward (Basturkmen et al., 2002). As noted, metalanguage does not include evaluative comments on the language use, such as 'good' or 'that's right'.

Terminology and Its Types
Berry (2010) defined that terminology is a collection of terms and it is the meaning of the word that language teachers and learners are familiar with. In addition, terminology is be the most obvious manifestation of metalanguage (Berry, 2004).
There are three types of terminology based on its characteristics i.e. transparent, opaque, and iconic term (Berry, 2008). First, transparency is where the meaning of the term indicates what its referent is about . Typically, the clue is given by the meaning of the grammatical term ('the past tense' refers to the event which is occurred in the past). Second, opaqueness is where there is no obvious relationship between the term and its referent; learners have no clue from the term as to what it is about . The most common terms in English, those word classes, are opaque such as noun, verb, adjective, etc. The main disadvantage is the learning load they impose on learners. Third, iconicity is the simplification between transparency and opaqueness . For instance, a teacher can say "you should use -ing here" or "you should use an -ing form". However, such iconic terms are limited in their application. For, instance, a teacher cannot use iconicity to express 'noun'. Comparing those three types, iconicity terminology is the most useful one. In addition, there is a sub-class of iconic terms which is called Eponymous. For instance, when speakers or writers refer to used to (as the grammatical item) by saying 'used to' (as the terminological item); it refers itself. The number of purely iconic terms in English is quite limited, but there are many mixed terms where an iconic element is combined with another such as 'third-person', '5W+1H question word', 'to infinitive', etc.

Tests and Metalanguage
In several studies (Ellis, 2005;Ercetin & Alptekin, 2013;Mirzaei, Rahimi, & Shakerian, 2011), the Untimed Grammaticality Judgments Test (UGJT or Untimed GJT) were mostly used as the instruments that requires metalanguage. As the examples, these following UGJT were constructed by Mirzaei, et al. (2011) Meanwhile, the UGJT should be completed by giving the correct form of ungrammatical sentence and explaining the rule. Indeed, the rule contained metalanguage production.
Instruction: 1) Underline the grammatically incorrect word(s) in the mini dialogue, 2) Write its correct form, and 3) State the grammatical rule that has been broken A. What do you usually do on Fridays? B. I often goes to the cinema Correct form: go Rule: the verb must agree with the subject. 'I' is the first person singular subject, but 'goes' agree with a 3 rd person singular subject.

Figure 3 Untimed GJT
Since the focus of the present study was investigating metalanguage, the researchers provided two other examples of the test. First, measuring and examining explicit knowledge in written form by completing several tasks such as metalanguage recognition, correction, production, and explanation or it can be generalized as the test of grammatical rules. The original test was designed by Andrews (1999b) then adapted by Tsang (2011).

Task 1 Metalanguage production
What grammatical terms would you use to describe the item underlined in each of the sentences? WRITE your description in the SPACE provided. NOTE: For each item provide a full description. For example: 1. He is funniest clown in the circus superlative adjective

Task 3 Grammatical error correction and explanation
This section consists of fifteen English sentences, each of which contains a grammatical mistake. For each sentence: 1. Rewrite the faulty part of the sentence correctly. (there is only one part that is wrong) DO NOT rewrite the whole sentence. Underneath each sentence, explain the error. For example: 1. I often goes to the cinema. Correct version: go Explanation: the verb must agree with the subject.

(DO NOT write: Change 'goes' to 'go') Figure 4 Language Awareness Test
Second, Hu (2011a) administered a rule verbalization task which covers the selection of six target language into two groups: the articles (a/an, the, and Ø) and three tenses (simple present, past, and present perfect). In addition, this kind of test can be used to verify the explicit knowledge when it combined with Andrews' (1999b) test.
Instruction: Explain why the underlined structures are used 1) Could you please shut the door? 2) A leopard is a very dangerous animal 3) If he were here, he would be able to help us a lot 4) Who has broken the window? Figure 5 Verbal Elicitation Test

Methodology
This study applied qualitative approach within post-positivist paradigm to inquire the use and the cause of phenomenon from multiple perspectives (Creswell, 2007). The subjects were the students at Advanced Grammar Class year 2017, TEFL Graduate Program, Universitas Sebelas Maret (UNS).
Documents referred to the results of the students' Advanced Grammar Exam. This exam entitled TOEFL model examination contains 40 questions which are divided into section A and B. In section A, the students should answer 15 questions of multiple choices which take form of incomplete sentence. Section B contains 25 questions where the students should identify unaccepted word or phrase in a sentence and mention the grammatical rules or errors (e.g. parallel structure, subject-verb agreement) for each sentence. This exam should be finished in 90 minutes. For the purpose of this study, the researchers only used Section B as the data because it contained the students' metalanguage. In total, there were 17 exam papers which were collected from 17 students.
Conducting qualitative inquiry involves a lot of cooperation between the researchers, the subjects, and also the sources. Regarding this issue, the researchers believed it was important to consider the ethics in conducting this study. To gain the students' trust and build a good relationship, the researchers informed the purposes and the benefits of being the subjects of this study. In addition, to keep their privacy, all names which were substituted by some initial numbers.
Validating data in a qualitative study cannot be taken for granted because it is relative; relies on the context of the study (Ary, Jacobs, Sorensen, & Razavieh, 2010). As the replacement, commonly, qualitative studies use the term 'credibility' which concern the trustworthiness of the findings. To enhance the credibility, this study applied triangulation of theory and rater-check.
The results of students' exam were collected by the students to the researchers. The researchers continued to read the raw data for several times to familiarize and comprehend the meaning in each description which was mentioned by the students. Meanwhile, notes or memos were also given during these processes. To begin the analysis, the table of metalanguage was used. This table was adapted from Basturkmen, et al. (2002) and Berry's (2010). It analyzed the categories of metalanguage and the types of terminology which was found in the students' answers in Advanced Grammar Exam. In addition, the frequency of metalanguage and terminology was calculated to find out the dominant use of particular categories and types. Both researchers, who played as raters as well, analyzed the data independently and then combined the results. The score of the final result agreement between researcher 1 and researcher 2 was high (91%).

Findings and Discussion The Categories of Metalanguage Produced by The Students
After analyzing the students' exam, the researchers found that the students used technical and non-technical categories of metalanguage in their answers. All of the students' answers were identified as metalanguage because metalanguage could take a form of single word, phrase, or even clause. Thus, each of the students' answers on the exam' items were counted as one metalanguage/ term.
Initially, based on the lecturer's instruction, the Advanced Grammar Exam should be completed by using technical metalanguage as the answers. For that reason, all of the answers should be the technical metalanguage. In fact, almost the entire students' answers were identified as the technical metalanguage. However, unfortunately, there were some students who used non-technical metalanguage as their answers. As noted, the use of phrase problem with… in the students' answers was the lecturer's suggestion; not an obligation, to address the error.
Technical metalanguage All of the three types (opaque, transparent, and iconic) were found in the students' exam. Based on its frequency, the opaque type and the transparent type were the most dominant types which were used by the students. Meanwhile, there were only a few terms which were identified as the iconic type.
The following descriptions presented the detail explanations of each type. Indeed, some parts of evidence i.e. the students' answers and examples of the exam's items were provided. Each of the examples (i.e. opaque, transparent, and iconic) was printed in Italic.

Opaque type
In the students' answers, the opaque type was found as the mostly used terminology than the others (i.e. transparent, and iconic). According to its frequency, the researchers found that there were 209 opaque terms.
In a closer look, there were four patterns of opaque terms which were commonly used by the students. First, the use of opaque term (only): noun, verb, pronoun, article, adjective, subjunctive, and appositive. Second, the use of opaque term + revision form: should be noun "capability", should be adverb "broadly", should be in simple present "sneezes", should be in past perfect "there had been", noun-should be "formation", needs noun not adjective, past participle (drank-drunk), and omit "who"appositive. Third, the use of problem (with/in) + opaque term: problem with article, problem with noun, problem with omitted article, problem with verb, problem with determiner, problem with pronoun, problem with adjective, problem with adjective clause, problem with subject verb disagreement, problem with the meaning of the verb problem in thingnoun, and problem in positioning adjective. Fourth, the use of opaque term + common words: appropriate verb, verb form, noun phrase, ambiguous adverb, definite article, general noun, to infinitive, uncountable noun, parallel verb, double adverb, double verb, subject-verb agreement, verb agreement, incorrect article, omitted article, negative inversion, possessive pronoun, transitive verb, definite noun, thing-noun, relative pronoun, bare infinitive, infinitive verb, to infinitive, compound noun, double subject, noun use, adjective use, adjective clause connector, modifier in noun phrase, noun form in no phrase, incorrect order of noun phrase, repetition of subject is not necessary, plural singular noun, choice of words/ noun, adverb of manner, should be adjective, agreement of modal in conditional sentence, form of verb in meaning, basic form of adjective, pronoun agreement with the reference, modifier of countable noun, relative pronoun to relate object, and subject verb agreement in inversion.
Indeed, some terms such as subject, noun, verb, adjective, subjunctive, and article were the most obvious form of the opaque terms. If one of these terms was combined with another common word, commonly, it would be classified into opaque type. For example, agreement was a common word which was used by any person. In the contrary, subject and verb were particular terms which were frequently used in a grammar class. Therefore, subject verb agreement, as the combination of those three, was the opaque type.
The opaque terms above were used to answer most of the exam's items. Below, Figure 6 presented some of the items which were answered by using the opaque terms.

s Items Answered by Opaque Terms
In her exam, most of the Student 10's answers were identified as the opaque type. Some of the terms were presented below.

Part B 21
A (Noun phrase)

A (Subject verb agreement) Figure 7 The Student 10's Answers
Briefly, within the opaque type was not only consisted of opaque terms. Sometimes, in the students' answers, the opaque terms were combined with another word in order to form a set of grammatical rule/ terminology to explain the error. As mentioned before, the total of opaque terms was quite many. Moreover, more than half of technical metalanguage consisted of opaque terms Transparent type Although the transparent type was less dominant than the opaque type, the number of the terms' frequency between these two types was not significantly different. 184 transparent terms were identified in the students' answers.
Similar to the opaque patterns, the use of transparent term was distinguished into four common patterns. First, the use of transparent term (only): comparative, connector, conjunction, present, comparison, preposition, and plural. Second, the use of transparent term + revision form: parallel construction-should be "sneezes", comparative-should be 'prettier', preposition-"effect" matches with "on", preposition-"despite" is not followed by "of", comparative-should be "as quick as", should be in simple past "lived", comparison-should be "as fast as", missing of preposition-it should be "in sleeping", preposition-it should be 'by', problem with to be-it must be 'were', past future/will-would, and parallel structure/to hunt-hunting. Third, the use of problem with + transparent term: problem with connector, problem with comparison, problem with to be, problem with conditional type, problem with preposition, problem with the form, problem with the meaning, and problem with usage/ like. Fourth, the use of transparent term + common words: parallel structure, conditional sentence, appropriate preposition, double adverb of time, if clause type 2, conditional tense type 2, object of preposition, form of comparative, past tense, parallel structure agreement, misused of preposition, passive voice, double negative expression, comparative degree, relative clause, distinguish past/ present meaning, active sentence, simple present tense, compound sentence, incorrect comparative form, incorrect preposition, and prepositional use.
Parallel, comparative, past, and present were some of the transparent terms which were frequently used in the students' answers. Similar to the combination of terms in the opaque type, the comprehensive grammatical rules in the transparent type were mostly followed by another term. For instance, the grammatical rule of (problem with) parallel structure of a verb was related to the sentence/ clause which contained an error on its verb. Indeed, verb was an opaque term. However, the terms parallel and structure were transparent terms. As a result, parallel structure of a verb was classified into the transparent type.
Here, Figure 8 presented some of the items which were answered by using the transparent terms above.

s Items Answered by Transparent Terms
One of the students who frequently used the transparent terms as the answers was Student 06. As the example, 4 of his 16 transparent terms were mentioned in Figure  11.

Part B 23
A (Conjunction)

C (Parallel structure)
29 B (Singular plural) Figure 9 The Student 06's Answers The use of transparent terms might not be as much as the opaque terms. Still, the amount of the opaque terms was quite many. Moreover, almost the half of technical metalanguage consisted of transparent terms.

Iconic type
The use of iconic type was so few because there were only 3 iconic terms which could be found: problem with -ed/-ing adjective, form of be+to+v1 and present perfect/ participle/ v3. As the definition said, these terms contained the particular parts (-ed/ing, v1, v3) of its referents; the opaque terms (adjective, present perfect/ participle).
The iconic term problem with -ed/-ing adjective above was used to address the error on accepted on item 35.

Figure 10 The Exam's Item Answered by Iconic Term
As identified in the Student 11's, answers below, the iconic term-ed/-ing was followed by an opaque term i.e. adjective. Or, the adjective was the referent for theed/-ing. Then, the researchers decided to classify it into the iconic type.

Figure 11 The Student 11's Answer
In the iconic type, the terms were always followed by another term. It was different from the opaque and the transparent type in which there was a possibility that their terms were not followed by another term.
From the explanations above (i.e. opaque, transparent, and iconic type), the researchers concluded that identifying terms into a particular type was not an easy task because most of the students' technical metalanguage contained more than one type of term. As the consequence, the results of identifying terms into a particular type might be less precise. Even so, it could be considered as a common phenomenon.
Last, the summary of the terms frequency within technical metalanguage was presented in Table 1. The different frequency between opaque type and transparent type was close. Meanwhile, compared these two types, the frequency of iconic type was much different.

Non-technical Metalanguage
From the students' answers, the researchers found 21 words and/ phrases of nontechnical metalanguage. The use of non-technical metalanguage had 2 similar patterns such as found in technical metalanguage. First, the use of non-technical (only): habitually, double, suggestion, should be omitted, and unnecessary word. Second, the use of non-technical + revision form: should be 'capability, should be 'to sleep', should be 'as quickly as', should be 'his/her', should be 'board speaking', should be 'concerned to', should be "when", "has" should be "have", "may" should be omitted, should be "whom", incomplete form of "as quickly as", there is no "the", "more" should be omitted, omit 'seldom', and "should"/ advice. In using non-technical metalanguage, the students mostly used the word should before presenting the revision form.
Furthermore, Figure 12 presented some of the items which were answered by using non-technical metalanguage.

s Items Answered by Non-Technical Metalanguage
Of all the students, Student 03 was the student who most frequently used the non-technical metalanguage than the others.
Part B

26
C (should be omitted)

Figure 13 The Student 03's Answers
Totally, there were 20 words and/ phrases of non-technical metalanguage inform of in the students' answers. By deriving the entire descriptions, distinguishing between non-technical and technical metalanguage was easier to be done than identifying between opaque, transparent, and iconic type. Indeed, the use of non-technical by the students' answers was an anomalous occasion; somehow, it enriched the analysis of metalanguage.
Analyzing categories (non-technical/ technical) and its frequencies did not always provide clear-cut results because some terms of metalanguage were slipped into a borderline category. In the end, based on the students' answers, these results found the dominant propensity of technical metalanguage which was summarized below.

Technical & non-technical metalanguage
The current finding showed that technical and non-technical categories of metalanguage were used by the students. In comparison, this finding was in line with Basturkmen, et al. (2002) and Fortune (2005) where their subjects also used both categories. Regarding these three findings, the researchers identified that the EFL context of grammar course became the main factor which contributed to this result. As it was mentioned previously, almost all of the students involved in these studies were non-native English speakers from various EFL/ESL countries (Spain, Arab, Italy, Korea, Japan, and Indonesia). This assumption proved Robuschat's (2015) statement which argued that such grammar classrooms provided a natural environment for the use of metalanguage (e.g. grammatical rules).
Regarding the dominancy, it found that technical metalanguage was much more dominant than non-technical metalanguage. This current finding did not correspond to Basturkmen, et al. (2002) and Fortune (2005) studies which found that non-technical metalanguage as the dominant category. It was quite possible that these contradictory findings were influenced by different choices of data sources. As mentioned in the previous chapter, the data of the current finding were collected from the students' answers in the grammar test. On the other hand, the other two previous findings (Basturkmen et al., 2002;Fortune, 2005) were collected from the discussions between students and teachers during the grammar course. As Creswell (2012) said, the choice of data source depends on the research questions. Moreover, Hu (2010) argued that technical metalanguage provided the explanatory precision and the efficient delimitation of the contexts especially in grammar tests. It explained why technical metalanguage is much more required that non-technical metalanguage in the grammar tests.
Despite the contradictory findings, the use of technical metalanguage was relatively high not only in the current findings but also in Fortune (2005). These findings were not surprising since all subjects were the students in advanced level of proficiency. Also, the use of grammatical terms for examination purpose is essential (Mohamed, 2012). It proved R. Ellis' (2005) statement that advanced students tended to use technical metalanguage confidently than another student (e.g. intermediate or beginner). In short, the graduate students who are also advanced learners deserve to use it.
Opaque, transparent, and iconic terminology The types of terminology were the substances of technical metalanguage. In the current finding, the students used opaque, transparent, and opaque types of terminology. This finding confirmed the notion of three types of terminology which was proposed by Berry (2008). However, so far, the researchers had not found any study which inquired these types of grammatical terminology. Possibly, it was affected by the unfamiliarity towards this notion for language scholars; especially if they were lessinterested in metalanguage. On top of that, the notion towards the types of terminology was just developed in recent years. In the other words, this notion was relatively new.
Regarding its frequency, there were two notable findings. First, the opaque type was found as the most frequently used than the others. It proved Berry's (2008) statement which said that most of the English' word classes were opaque; particularly, if it dealt with grammatical rules. Second, in contrast, the use of iconic type was so few. There were only 3 iconic terms which were identified. As Berry argued, the application of iconic type was very limited. For instance, the iconic terms such -ed and -ing forms could be applied in verbs (e.g. learn followed by -ed) but it could not be applied in nouns. As noted, originally, iconic terms were morphemes. Therefore, it was obvious if the students used the opaque terms much more frequently than the iconic terms.
During the analysis, the researchers was confronted by a dilemmatic problem in identifying the terms which were constructed of more than one type such subject verb agreement. On one side, subject and verb were obviously identified as the opaque type. On the other side, agreement was identified as transparent type. To deal with this problem, Berry (2010) identified the terms such subject verb agreement as the mixed types. However, the mixed types seemed to be an ambiguous result. Therefore, for the purpose of the current study, the researchers decided to identify it as the opaque type because it consisted of 2 opaque terms and only 1 transparent term. The distinction towards metalanguage whether some terms were identified into particular categories (i.e. technical and non-technical) and types (i.e. opaque, iconic, transparent) might not provide clear-cut result because some terms might fall somewhere between the continuum (Basturkmen et al., 2002).
Interestingly, some literature and empirical studies imply that metalanguage entails different levels of difficulty in learning; due to the abstractness vs. the novelty of grammatical rules (DeKeyser, 2003). It consists of four levels of difficulty. First, the opaque terms have a high level of difficulty because the learners have no clue from the term as to what it is about (Berry, 2008). Second, the iconic terms have a moderate level of difficulty because the number of purely iconic terms is very limited and it mostly followed by opaque terms (Berry, 2010). Third, transparent terms have a low level of difficulty. The meaning of the term indicates what its referent is about (Berry, 2008). Fourth, non-technical metalanguage has very low level of difficulty because it has common usage; not limited to grammar or language learning. Therefore, it is easier to be produced and understood than technical metalanguage (Basturkmen et al., 2002). It concludes that the more technical (opaque) the metalanguage needed to formulate a rule the more difficult that rule will be learned and used (Ellis, 2006).

Conclusion
Clearly, based on the findings, it revealed that the TEFL graduate students frequently used technical metalanguage rather than non-technical metalanguage because technical metalanguage was more precise and theoretically valid as the answers in Advanced Grammar Exam. In a closer look, their technical metalanguage had three types namely opaque, transparent, and iconic terminologies. Of these three types, opaque terms were the most used by them since most of the grammar terminologies were opaque. Although empirically the students were able to use massive metalanguage terms, they did not realize that each type of term entailed different levels of difficulty.