Language-Driven CLIL: Developing Written Production at the Secondary School Level

This research study analyzes the effect the implementation of language-driven CLIL has on senior learners from Manuel J. Calle High School in Cuenca, Ecuador in relation to the development of written production in terms of Syntax, Content, Communicative Achievement, Organization, and Language compared to a non-language-driven CLIL classroom. There were 40 participants in the experimental group, and 38 participants in the control group. Learners from the experimental group received a condensed 35-hour intervention using CLIL. This study features an exploratory, mixed-method, and quasi-experimental research design. To collect qualitative data, an open-ended questionnaire was administered to explore the subjects learners preferred to study in a language-driven CLIL classroom. To collect quantitative data, a Pre and Post-Test based on the writing section of Cambridge Objective Primary English Test was administered. The data was analyzed through the Independent T-Test and Paired-T-Test to determine if there was a statistically significant difference present between the language-driven CLIL classroom and the non-language-driven CLIL classroom. The data was calculated through the Statistical Package for Social Sciences (SPSS). A survey was administered to collect data on learners’ perceptions about CLIL and then analyzed statistically. Results indicated that learners preferred to study History, Biology, and Spanish Language and Literature. Results also demonstrated that the experimental group also demonstrated improvement in all the examined parameters when compared to the control group. However, when results from both groups are compared, there is only a statistical improvement in Organization and Syntax.

Gené-Gil, Juan-Garau, and Salazar-Noguera (2015) left inconclusive questions to confirm or reject the effect of language driven CLIL on written production in other contexts. That is why this study aims to analyze how the implementation of the language-driven CLIL helps senior learners from Manuel J. Calle High School develop the writing production of texts in comparison to a non-language driven CLIL classroom.

Research Problem
Writing is a difficult task to acquire even for native speakers (Klimova, 2012) since many elements have to be developed simultaneously (Javed, Xiao, & Nazli, 2017;Muluneh, 2018). Learners from Manuel J. Calle High School have difficulties giving a valid argument, organizing ideas, using grammar and vocabulary properly, and using correct punctuation. Research studies on language-driven CLIL show positive results and indicate it could be adapted to any contexts (Cenoz, Genesee, & Gorter, 2013;Šulistová, 2013). Thus, the current study addresses two research questions: To what degree does the implementation of language-driven CLIL impact the development of written production in terms of Syntax, Content, Communicative Achievement, Organization, and Language in comparison to a non-language-driven CLIL classroom? What are learners' perceptions towards language-driven CLIL when writing paragraphs?

Methodology
This project features an exploratory, mixed-method, and quasi-experimental research design. The researcher first collected qualitative data and then collected quantitative data to explain quantitative results (Mertler, 2017). It also presents a mixed methods research design because of the integrated elements of qualitative and quantitative research approaches to provide breadth and depth of understanding of the research problem (Schoonenboom & Johnson, 2017). Qualitative research helped the investigator explore and understand learners' preferences on content subjects and topics. Quantitative research, on the other hand, tested if there was a statistically significant improvement after the intervention between the control and the experimental group (Creswell, 2014). Finally, this study shows a quasi-experimental research design since participants in both groups were selected without random assignment. Both groups took the Pre-Test and the Post-Test, but only the experimental group received the treatment (Creswell, 2014).

Participants
This study took place at Manuel J. Calle High School, a public school located in Cuenca, Ecuador. Participants were students from the Third Year of Baccalaureate: Classes A and B. There were 20 female learners, which represents 53%, and 18 male learners, which represents 47% in the control group. Meanwhile, there were 9 female learners, representing 21 % and 31 male learners, representing 79% in the experimental group. There was a small gender gap in the control group with more males than females, while the gender gap in the experimental group was large, with males unevenly outnumbering females. Learners were aged 16 to 19 years old. The control group presented a lower average age of 17 in comparison to the experimental group, whose average age was 18.

Qualitative Data Collection
At the beginning of the second semester of the 2018-2019 school year, qualitative data was collected through an open-ended questionnaire to determine learner's preferences about the content subject and topics they found interesting. The content subjects were taken from the Ecuadorian Curriculum (2016). The open-ended questionnaire was elaborated in Spanish and was piloted and validated with 76 students during the first semester of the 2018-2019 school year.

Qualitative Data Analysis
Responses from the open-ended questionnaire were manually transferred into an excel spreadsheet. Then, they were classified into categories until saturation was obtained. Finally, inductive and descriptive analysis were done to determine learners' preferences about content subjects. Once the top three content subjects were obtained, they were used to plan the intervention.

Qualitative Data Results
Findings revealed that both groups (32% in the control group and 50% in the experimental group) preferred History. Second was Biology with 29% in the control group and 26% in the experimental group. In third place was Spanish Literature with 11% in the control group and 16% in the experimental group. Learners also stated that learning those subjects in English would give them the opportunity to improve content and technical vocabulary. Sub-topics and themes for planning language-driven CLIL lessons were selected by learners. The World Wars (History), the Human Body (Biology), and Decapitated Era (Spanish Literature) were the topics that learners expressed that they would like to study. Regarding methodological strategies, 55% participants in the control group and 61% of participants in the experimental group stated they preferred group-work activities.

Intervention
A language-driven CLIL Unit was elaborated based on the qualitative results and on theories by Met (1999), Coyle, Hood and Marsh (2010), Olsson (2010), Bentley (2010), and Kusmayadi and Suryana (2017). It was essential to divide the unit into single lessons considering Content-Compatible Language and Content-Obligatory language objectives (Bentley, 2010;Banegas, 2012).Seven lesson-plans were created, and each one of them took 5 class periods. As learners in the experimental group were at an A1 level according to the Common European Framework of Reference for Languages (CEFR), the writing tasks started from writing simple isolated phrases and sentences to writing texts at the end of the semester. On the other hand, a unit plan for the control group was created following the Communicative Approach since it is an approach promoted by the Ministry of Education in Ecuador (Ecuadorian EFL Curriculum, 2016). Thus, the intervention took place from April 22 nd to June 11 th , 2019. There were 35 hours of intervention. Similarly, the researcher worked with the Communicative method in the control group during the same period of time.

Quantitative Data Collection
Quantitative data was collected through the use of two instruments. The first was the Cambridge Objective Primary English Test (PET exam), and the second was a survey. Both were collected with the participants' consent and administered by the researcher. The PET exam was employed with two aims. First, the PET exam helped determine participants' general English proficiency before and after the 35-hours of intervention. The proficiency test provided data to be able to analyze if language-driven CLIL helped learners move from one level to another in general terms. Secondly, the written section of the PET exam was used as the Pre and Post-Test in order to analyze how language-driven CLIL influenced the development of written production. Thus, the writing skill section was scored based on the PET writing rubric, which had five parameters: Syntax, Content, Communication Achievement, Organization, and Language. Each of these parameters had a grading scale in which 5 was the highest score.

Data Analysis: PET Test and Written Production
The results of listening, reading, and speaking of the PET exam were transferred to an excel spread sheet. The writing results were also processed in an excel spreadsheet, but were organized according to the writing PET rubric. To analyze the obtained results of the Pre-Test and the Post-Test, Llinares and Whittaker's (2007) and Olsson's (2010) criteria was followed. Such criteria suggested the use of the T-Test in order to compare results between a language-Driven CLIL classroom with non-language-driven CLIL classroom. With this in mind, the independent T-Test and Paired T-Test were used as a hypothesis testing tool that allowed the researcher to determine if there was a significant difference between the means of the experimental and control group. To calculate these means, the Statistical Package for the Social Sciences (SPSS) was used. After that, a descriptive analysis of the main measures of central tendency as well as the distribution of variables was done. The variation in the written production from both groups was described through inferential statistical analysis. This analysis, along with the analysis of the students' perceptions, converged to triangulate the validation of the research questions.

The Survey
In order to find out students' perceptions on language-driven CLIL, a survey was elaborated and taken anonymously at the end of the intervention. The survey was elaborated following Ikeda's (2013) study. The survey had two sections. The first section had 5 closed-ended questions and rested on Coyle's 4 Cs. For each question, learners were given a five-point Likert scale (5=strongly agree, 4=agree, 3=neither agree nor disagree, 2=disagree, 1=strongly disagree). The second section of the questionnaire had one open-ended question, which asked learners to write freely about how they considered language-driven CLIL helped them write in English. The survey was elaborated in the learner's native language, and was piloted in similar classes to have it validated.

Data Analysis of the Survey
The responses from the survey were immediately transferred into an excel spreadsheet. Each rank on the Likert scale was tabulated independently. The responses from the survey were analyzed using the relative frequency on learners' perceptions about language-driven CLIL. To check reliability of the five-items, descriptive statistical analysis with percentages was done through the SPSS. elt.ccsenet.org English Language Teaching Vol. 13, No. 8;2020 13. Results

Analysis of the Written Production Results-Experimental Group
To analyze the results of the written production in terms of Syntax, Content, Communication Achievement, Organization, and Language of the experimental group, the mean, median, and mode of the results of the Pre-Test and Post-Test were calculated. The writing section of the Cambridge Objective Primary English Test had three parts. Part 1 evaluates Syntax by asking learners to complete the sentence so that it had the same meaning as the example. Part 2 and 3 asked learners to write descriptive and narrative texts. In Part 1, the experimental group improved their average performance from 40% in the Pre-Test to 60% in the Post-Test. In Part 2, the results show that students improved from 60% to 80% in terms of Communication, Organization, and Language after the intervention with the Soft CLIL method. Meanwhile, in terms of Content, the average performance was maintained at 60% in the Pre-and Post-Test. Prior to intervention, the results of the group's performance in the second section had a symmetric distribution. After the intervention, only "Organization" became negatively skewed, in which the majority of students obtained 60%, which represents a lower than average performance. Regarding the evaluation of the third section, the four parameters have an average performance of 60% with a symmetric distribution before the intervention. In terms of Content, Communication, and Organization, the performance improved to 80% after the intervention. The average performance was maintained in terms of Language. In this section, the distribution of the results obtained in Content, Communication, and Language had a negative skew. Thus, most students had a 20% lower performance than the average group performance (80%) as shown in the previous table.

Written Production: T-Test Paired Sample Analysis
In order to reinforce the descriptive results that show improvement in the written production in the experimental group, a two-tailed hypothesis test was carried out through the Paired T-Test (Paired-Student). This helped determine if the differences between the average scores in the Pre-Test and the Post-Test were statistically significant. The following tables indicate the results of the parametric analysis of the paired samples in order to contrast the hypothesis and to determine if there is a significant difference between the final averages of the experimental group before and after the intervention.   Vol. 13, No. 8;2020 us to the conclusion that with an error of 0.000 in all cases, there is a significant difference between the average scores of the evaluated parameters in the Pre-Test and the Post-Test. In short, through the use of Language Driven CLIL methodology, the development of the written production in English in the experimental group improved .
There is a difference of 0.58 points between the average performance of Part 1 obtained in the Pre-Test by the experimental group (2.25) with the average performance obtained in the Post-Test (2.83). This reveals a statistically significant improvement in the development of written production through Language-Driven CLIL.
The average performance in terms of Content in Part 2 is 2.98 in the Pre-Test, and the average in the Post-Test is 3.40 in the experimental group. There is a difference of 0.42 points, which shows a statistically significant improvement in the development of written production through Language-Driven CLIL. The average performance in terms of Content in Part 3 is 2.78 in the Pre-Test and 3.80 in the Post-Test. The difference of 1.02 points also demonstrates a statistically significant improvement. This means that the reader was informed on the topics that writers described in their paragraphs and that the content was relevant and appropriate.
The average performance in regards to Communication in Part 2 is 2.80 in the Pre-Test and in the Post-Test it is 3.50. There is a difference of 0.70 points, a statistically significant improvement. In regards to Communication in Part 3, the average performance is 2.58 in the Pre-Test and in the Post-Test it is 3.50, which is also a statistically significant improvement. Students improved in the use of conventions of the communicative task to express direct ideas.
Concerning "Organization" in Part 2 in the Pre-Test, the average performance is 2.85, and the average performance obtained in the Post-Test is 3.58. There is an improvement of 0.73 points, which is statistically significant. Concerning, "Organization" in Part 3, in the Pre-Test the result is 2.75, and the average performance in the Post-Test is 3.68. There is a difference of 0.93, which shows a statistically significant improvement. This result indicates that learners in the experimental group developed more connected, consistent texts through the use of linking words and cohesive devices than in the initial stages.
In regards to Language in Part 2, the average performance in the Pre-Test is 2.80, and the average performance in the Post-Test is 3.60. There is a difference of 0.80 points, which is a statistically significant improvement. In regards to Language in Part 3, the average performance in the Pre-Test is 2.90, and the average performance is 3.50 in the Post-Test. There is a difference of 0.60, which is a statistically significant improvement. Students in the experimental group showed good use of everyday vocabulary and used complex grammatical forms in their writings by the end of the intervention.

Writing Level of the Experimental Group
The four sections of the Cambridge Objective Primary Test had a maximum score of 185 points, which is equivalent to a C1 on the Common European Framework for Reference, and the maximum score in the Writing section is 45 points. Thus, having a reference of the maximum levels of the PET Exam as well as the writing section, the following table was created in order to determine the experimental group's writing level. The table reveals that before the intervention, the experimental group had an average writing score of 24 points, which is equivalent to an A1 level (99 points). In the Post-Test, the average writing score increased to 31 points, which puts learners at an A2 level (127 points).

Learners' Perceptions about the Language-Driven CLIL
This section describes the experimental group's perceptions regarding the impact of the Language-Driven CLIL on the development of written production. The results were obtained from the survey, which had 5 closed-ended questions in the form of statements, evaluated on a Likert-scale and one open-ended question: 1. Content learned through the use of Language-Driven CLIL was relevant and easy to understand. 2. Language-Driven CLIL helped produce texts using the conventions of writing (spelling and punctuation) to communicate direct ideas.
3. Language-Driven CLIL helped develop critical and creative thinking through organizing the text in a coherent and cohesive way.
4. Language-Driven CLIL facilitated the appropriate use of grammar and vocabulary to transmit knowledge of the different subjects studied in class.
5. I consider that the Language-Driven CLIL method influenced the development of written English in a practical and efficient way.
6. How do you consider the Language-Driven CLIL method influenced the scale you selected in the previous sentence?
Results for the first statement show that 48% (sum of 18% totally agree and 30% in agreement) of students agreed with the statement; 25% of learners said that Language-Driven CLIL did not make a difference to them; meanwhile, 28% (sum of 15% disagree and 13% totally disagree) of students disagreed with the statement. More than half of the students in the experimental group (58%) agreed on the second statement. On the other hand, 18% of students neither agreed nor disagreed; meanwhile, 25% of learners said that Language-Driven CLIL did not have any advantage when it came to writing. Results show that 53% of learners agreed with the third statement; 28% of learners neither agreed nor disagreed, but 20% of students said that they disagreed with the statement. Findings reveal that 60% of the students agreed with the fourth statement; 25% of the students expressed neutral opinions, and 15% disagreed with the statement. Results in regards to the fifth statement are described below: Figure 1. Language-Driven CLIL influenced in a practical and efficient way to develop the writing skill The figure reveals that 65% (sum of 35% totally agree and 30% in agreement) of learners agreed that the Language-Driven CLIL method influenced the development of the writing skill in a practical and efficient way; 20% of the learners had a neutral opinion, but 15% of the students disagreed with the statement. Concerning question number 6, learners who agreed with this statement said that the Language-Driven CLIL method allowed learners to produce written texts because they were provided with writing examples, they analyzed how written texts are organized, and arranged words, sentences and phrases to communicate ideas and opinions according to different situations. In contrast, the learners who stated that Language-Driven CLIL did not influence them in an efficient and practical way said this because content from the curricular subjects was hard to understand. They said that new words learned in class were difficult to memorize and it was therefore difficult to use them properly.

Written Production Results-Control Group
The written production parameters of the control group follow a symmetric distribution. Therefore, the measures of central tendency focus on the average scores as seen in the table below.  Before and after the intervention, the written production in Part 1(Syntax), has an average score of 2, which represents 40% of the total score. This average score did not vary after the intervention. The written production evaluated in Part 2, in terms of Organization, maintains an average score of 60% throughout the intervention period. This means that learners created well organized and coherent texts, using a variety of linking words and cohesive devices in the Post-Test. In part 2, in terms of Content, the control group had an average performance of 80% before and after the intervention. The Content of the written production was relevant in the different tasks and allowed the target reader to be fully informed. In some cases; however, minor irrelevant details and personal pronouns were omitted in the Post-Test. After intervention, in terms of Communication and Language in Part 2, the mean average moved from 60% in the Pre-Test to 80% in the Post-Test. This means that learners used communicative tasks to hold the target reader's attention, and they used a range of everyday vocabulary and grammar forms, respectively. In Part 3, the four evaluated parameters had the average performance of 60% before and after the intervention, as seen in the above figure.

Paired Sample T-Test Analysis of Written Production
The following tables show the results of the parametric analysis for paired sample T-Test in order to test the hypothesis and determine if there is a significant difference between the final average scores of this group.  The results indicate that the null hypothesis ( ) is rejected for Part 1 and Communication in Part 3. With an error of 0,026 and 0.001 respectively, there is a significant difference in the average scores in the Pre-Test and Post-Test in all other areas. This means that the intervention with the communicative language methodology improved the development of written production in the control group . The average performance in Part 1 obtained in the Pre-Test by the Control group was 2.05, while the average performance obtained in the elt.ccsenet.org English Language Teaching Vol. 13, No. 8;2020 Post-Test was 2.35. Thus, there is a difference of 0.29 points, and indicates a statistically significant improvement. The average performance of the control group in Communication in Part 3 of the Pre-Test was 2.66, and the average performance obtained in the Post-Test was 3.18. This means that there is a difference of 0.52 points of improvement, which is statistically significant. Learners in the control group were able to use the conventions of the communicative tasks to express direct ideas. This helps us conclude that there is a 22% improvement, which is statistically significant in the control group regarding written production.

Writing Level of the Control Group
The four sections of the PET Exam had a maximum score of 185 points, which when compared to the Common European Framework for Reference is equivalent to a C1 level , and the maximum score in the Writing section is 45 points. Thus, having a reference of the maximum levels of the PET Exam as well as the writing section, the following table was created in order to determine the control group's writing level. The above table shows that learners from the control group obtained 27 points in the Pre-Test, which corresponds to an A1 Level (111 points). In the Post-Test, the writing level of the experimental group is maintained at 29 points.

Final Results in the Written Production: Control and Experimental
To determine if the written production was impacted through the language-driven CLIL model in terms of syntax, content, communicative achievement, organization, and language, scores from parameters in Part 1, Part 2 and Part 3 were put together through the Levene Test. The Levene Test for equality of variances indicates probability associated to Levene's statistic, which is higher than 0.05, equal variances are assumed for all analyzed parameters. The Paired T-Test statistic with its bilateral significance reveals that Syntax (Part )1 and Organization is lower than 0.005 in the Post-Test. This means that there is compatibility between the hypothesis of equality of the average scores of the analyzed parameters of the control and experimental group. The average performance in Syntax (Part 1) of the experimental group obtained in the Post-Test is 2.83 and is 2.34 for the control group. There is a difference of 0.49 points. This is a favorable difference for the experimental group and it is statistically significant. The average performance obtained in terms of Organization in the experimental group in the Post-Test is 3.63, and in the control group it is 3.33. There is a difference of 0.35. This difference is in favor of the experimental group and it is statistically significant. The Paired T-test statistic with its bilateral significance reveals that: Content, Communicative Achievement, and Language is higher than 0.05 in the Post-Test. This indicates that the hypothesis of equality of mean scores in the evaluated parameters in the control and experimental group is rejected. Although there is a mathematical difference in the mean scores and the experimental group has a higher score than the control group, such differences are not statistically significant, and both groups have the same writing level in the above parameters at the end of the intervention.

Skills: Experimental vs. Control Group after the Intervention
The figure below shows the percentage of variations in the average scores of the skills evaluated through the Post-Test PET exam of both groups.

Figure 2. Control and Experimental Group: Results of the skills after the intervention
The growth of the control group's average in Reading (5.6%) is higher than the experimental group by 3.1 percent. The experimental group obtained a higher variation in the average scores in writing. This group obtained 27.1%, which represents 6.1 percentage points more than the positive variation in the control group (6.1%). Listening, Speaking, and the final global score of the control group presented a higher difference of 25.1%, 3%, and 0.2% respectively in the average grades of the mentioned skills, when compared to those obtained by the experimental group.

Discussion
The research question in this study was to analyze how the implementation of language-driven CLIL helps learners develop written production of texts in comparison to a non-language-driven CLIL classroom in terms of Syntax, Content, Communication Achievement, Organization, and Language as well as the learners' perception towards language-driven CLIL when producing texts. According to the results, learners from the experimental and control group show a variation in the average scores in all the evaluated parameters after intervention. The language-driven CLIL classroom had higher results in comparison to the non-language-driven CLIL classroom, but not all of them were statistically significant. Only in terms of Syntax and Organization, results show a statistically significant improvement.
Concerning Syntax, the obtained results from the experimental group indicate that learners were able to rewrite sentences properly and communicate the message meaningfully in the Post-Test. Findings in this research study are similar to the ones obtained by Lahuerta's (2017) results. In that study, errors diminished substantially in terms of syntax since learners were also able to look for ways to combine words, phrases, clauses, and sentences and communicate the message meaningfully through the use of language-driven CLIL. Nonetheless, Gutierrez-Magado and Martínez-Adrian (2018) found negative results in terms of syntax-morphology, but they concluded by saying that language-driven CLIL aided in acquiring features from syntax-semantics-discourse interface, which was evident in this study, too.
Organization was another parameter that had a statistically significant improvement in writing in the language-driven CLIL classroom when compared to the non-Language-driven CLIL classroom. Findings reveal that texts were generally well-organized and coherent. Furthermore, learners used a variety of linking words and cohesive devices, such as sequencing, adding, illustrating and comparing. Learners were impacted positively in writing in terms of Organization, and these results are similar to the ones in the study conducted by Ikeda (2013). Learners in that study improved significantly in terms of organization. However, results of this study as well as Ikeda's (2013), are contradicted by Ruiz de Zarobe's (2010) findings. In that study, leaners improved significantly in terms of Content and Vocabulary, but they did not improve in terms of Organization. The researcher found that texts were not connected using linking words. Instead, learners wrote long sentences without any cohesive devices.
In regards to the other evaluated writing parameters (Content, Communicative Achievement, and Language) the results indicate that there was improvement in the language-driven CLIL classroom, but it was not statistically elt.ccsenet.org English Language Teaching Vol. 13, No. 8;2020 significant when compared to the non-language-driven CLIL classroom. Therefore, the language-driven CLIL classroom and the non-language-driven CLIL classroom maintain equal writing results on these parameters.
In Llinares and Whittaker's (2007) study, results reveal that learners' writing level improved significantly in terms of Content using language-driven CLIL. Learners were able to present a problem and give a solution, events were connected from the beginning to the end, and sequence in events was also observed. However, Gené-Gil, Juan-Garau, and Salazar-Noguera (2015) found that Content did not improve in neither language-driven CLIL nor non-language-driven CLIL classrooms. Learners' scores were low, and they showed a limited development of the main ideas. Learners did not respect e-mail conventions (title, story line, time, characters or personal opinions). In our study, on the other hand, most of the texts from the language-driven CLIL classroom were relevant to the given prompt and readers could fully understand the passage. Notwithstanding, results were not statistically significant.
About Communication, Bentley (2010) and Coyle, Hood, and Marsh (2010) stress that language-driven CLIL facilitates the learning of language by developing communicative skills. Findings in this study are similar to language-driven CLIL principles since learners were able to use conventions on the communicative task to hold the target reader's attention and communicate straightforward ideas. This is similar to what Garcia (2015) found in their research study, in which learners' texts communicated ideas meaningfully. Garcia (2015) did find significant results in that study when results were compared to a non-language-driven CLIL classroom. In our study, however, improvement is evident in the Language-Driven CLIL classroom, however, the results were not statistically significant.
In terms of Language, language-driven CLIL classroom results indicate that learners did not show a statistically significant improvement when compared to the Non-Language-Driven CLIL classroom. These results were also found by Olsson (2010). In that study, language-driven CLIL learners did not have a significant increase in the use of language in comparison to non-language-driven CLIL learners. Nonetheless, Ikeda (2013) asserted that learners were able to use a range or everyday vocabulary and complex grammatical forms in their texts through language-driven CLIL.
It is also important to refer to learners' perceptions when it comes to using language-driven CLIL. Findings revealed that most learners from the experimental group show a positive attitude toward the language-driven CLIL on all five questions. They claimed that the content of classes was relevant, they were able to express their ideas using the appropriate grammar and punctuation, they had a chance to give an opinion on the different topics covered, and they learned how to organize and link texts. These results were asserted by Ikeda (2013), who also found similar positive results. In that study, learners revealed that language-driven CLIL involved critical thinking tasks, cooperative work, knowledge increment, and vocabulary expansion. Nakanishi and Nakanishi (2016) found similar results, too. Learners had positive attitudes towards language-driven CLIL since learners improved their English Proficiency. Such results are comparable to the ones obtained in this study because learners from the language-driven CLIL classroom improved their English Proficiency Level.
The Language-Driven CLIL classroom started with an A1 level, based on the results from the Cambridge Objective Preliminary Test. After the intervention, learners moved to an A2 level. Writing and speaking results influenced the move from one level to the other. On the other hand, the non-language-driven CLIL classroom started with an A2 level in the Pre-Test, and they maintained the same level in the Post-Test. Listening is the skill in which learners obtained the lowest scores and the reason why they were unable to move to a B1 level.

Conclusions
First of all, quantitative data was collected through an open-ended questionnaire in order to determine learners' preferences about content subjects and topics to be studied. Descriptive analysis was done, and the findings revealed that the selected subjects and topics had a positive impact due to the fact that learners had a statistically significant improvement in terms of Syntax, Content, Communicative Achievement, Organization, and Language when results from the Pre and-Post Test were compared.
Secondly, the writing level of the experimental and control group was determined through the PET writing rubric and analyzed through the Paired Sample T-Test (Pair student). Findings revealed that language-driven CLIL learners and non-language-driven CLIL learners had an equal writing level (A1), before the intervention. After the intervention, learners from the experimental group moved from an A1 to an A2 level. Meanwhile, the control group maintained the same A1 level in the Post-Test. Therefore, it is concluded that the language-driven CLIL method helped learners move from one level into the next.
Thirdly, in regards to the writing parameters before the intervention, both groups showed an equal average performance in terms of Syntax, Organization, and Communication. The groups differed in Language and Content. The control group had higher results than the experimental group. However, after the intervention, the experimental group obtained higher results in the written production in terms of Syntax and Organization, and an equal level in terms of Content, Communicative Achievement, and Language. Thus, it is concluded that language-driven CLIL helped learners develop written production in terms of Syntax and Organization with a statistically significant improvement.
Moreover, learners from the experimental group were asked about their perceptions towards the language-driven CLIL method when learners were producing written texts. Most learners from the experimental group agreed that the language-driven CLIL method allowed them to produce written texts because they were provided with examples, they analyzed how written texts are organized, they were able to arrange words, sentences, and phrases to communicate ideas and opinions based on different situations. These findings lead us to conclude that the learners from the experimental group had a positive attitude towards the language-driven CLIL method, and those findings correlate with the writing parameters, in which learners had a statistically significant improvement (Syntax and Organization).
Finally, in regards to the English Proficiency level, the language-driven CLIL classroom had an A1 level before the intervention, but after the intervention they moved to an A2 level. On the other hand, the control group had an A2 level in the Pre-Test, and the level was maintained in the Post-Test. Thus, we can conclude that the language-driven CLIL method did not only help learners improve the evaluated writing parameters, but it also helped improve their English Proficiency in general.
All in all, the general objective of this study was to analyze whether or not the implementation of the language-driven CLIL helped senior learners from Manuel J. Calle High School develop the written production of texts in comparison to a non-language-driven CLIL classroom. Findings in this study show that leaners did improve in all the evaluated writing parameters. However, only in terms of Syntax and Organization, the results demonstrate a statistically significant improvement when compared to non-language-driven CLIL classroom. These findings open new gaps for further research. For instance, a replication study with more hours of intervention can be done to test if language-driven CLIL helps to statistically improve written production in the other writing parameters. Furthermore, an analysis of the impact of language-driven CLIL on the speaking skill can also be investigated.