Luận án A cognitive study of negative transfer of english prepositions made by Vietnamese learners

: strongly agree, agree, not sure, disagree, and strongly disagree; (iv) Ratio scale data which indicated equal distances between two adjacent values, but not the same as interval values, for we can count for a gradual increase of learner's proficiency, e.g., the final score of their tests. 3.2.1.6. Reliability and validity In order to ensure that the investigation was reliable, the author purposefully set up three repeated, randomly selected questions from the questionnaire to ensure that students have been well-chosen and have paid close attention throughout the investigation. Any responses from participants that differed between the two phases of the inquiry were removed from consideration for inclusion in the final results because they were considered inconsistent. In addition, the author paid special 68 attention to the validity of the measurements in this study. It was a question of whether the method used by the author for this thesis was appropriate for testing the hypotheses. The author aimed to have complete control over our inquiry in order to attain reliability and validity. All of the questions were divided into small groups so that they could be pretested by a small group of people and then corrected before being used in questionnaires. 3.2.1.7. Hypothesis Testing From the three research questions in Chapter 1, we put them out into eight intuitively understandable hypotheses in Section 1.4. Each of the hypotheses was treated carefully to present outcomes in the next Chapters 4 and 5 – Results and discussions. To achieve these goals, we tested a nut hypothesis (H0), which normally suppose there were no differences between groups or no relationship between variables. Howell (2010) claimed that people could never prove something to be true, but people prove something to be false (p. 98). We did have a reason to do that, even though many of us might not really believe that those ideas were true until we could prove that they were wrong.Therefore, when we proved the hypothesis was false, it did not mean the opposite was proved to be true. All we could learn was that the null hypothesis was rejected. In this case, If we rejected the null hypothesis, it meant that we rejected the idea or notion that there was no difference between observed groups or no relationship between variables. The mechanism of null hypotheses was extremely simple: Ha: The declaration of the null hypotheses numbered a. H0: If the difference between the groups is 0, there is no difference between groups or no relationship between variables in the hypothesis. Ha: If the difference between groups is not 0, there is a difference between groups or a relationship between variables in the hypothesis. 69 3.2.1.7. Statistics report Statistic report is an essential section for the finding and discussion section in this study. Researchers need to know clearly about the null hypothesis to understand the p-value. Every experiment will either have an effect or result in a difference between groups, or it will have neither. The p-value that is less than 0.05 traditionally indicates the results are statistically significant, also including two values of 0.051 and 0.049. P is always italicized and (sometimes capitalized). Typically, researchers do not use the number ―0‖ before the decimal point (.) for p- value in statistical results, e.g., p < .005 because it cannot equal 1. According to Larson Hall, each statistical test has a specific mathematical symbol associated with it (2016, p. 46). In this thesis, these statistics are calculated and resulted in numbers in the following tests: - A chi-square test has a chi (χ); - A T-test has a t; - A correlation has a Pearson‘s γ; - and an ANOVA has an F. These kinds of tests are also known as the statistical test of group differences. These kinds of tests examine the membership in a particular group that affects the outcome of the research. We also look at questions in order to investigate how strongly one variable influences another under the terms of correlation, partial correlation, and regression.  Chi-square test In analyzing the relationship between two variables, the researcher applies the chi-square test of independence, which is named after the Greek letter (chi) for the calculation of the variance in the investigation. The purpose of this test is the comparison of the observation of outcomes (values) with the expectation for the 70 survey. For the demonstration of this, the research will follow the data collection procedure drawn from the research on ―Language transfer and discourse universals in the Indian English article use‖ (Sharma, 2005). The Chi-Square test is suitably applied to test relationships between two categorical variables from the survey. If its null hypothesis (H=0) happens, there is no relationship exists between the categorical variables in the investigated population. In other words; they are independent with each other. In this research, we use to test the following hypotheses: Hypothesis 3 (H2): L1 transfer negatively influences the use of Prepositions in English by Vietnamese learners of English. Hypothesis 4 (H4): Vietnamese learners of English tend to use simple English prepositions rather than complex English prepositions. Hypothesis 5 (H5): There are no significant differences among the low, intermediate, and advanced levels of Vietnamese learners of English in terms of negative transfer affecting a specific preposition usage. Hypothesis 6 (H6): There are no significant differences between Writing and Speaking skills for using English prepositions in terms of negative transfer. In order to know whether these variables in the hypotheses numbered 3, 4, 5, and 6 are independent or not. The researcher needs to compute the chi-square statistic with the help of the SPSS software for analysis. First, the researcher will find the difference between the observed values (outcomes) and expected ones. Then, the researcher squares the difference before dividing this difference by the expected values. This calculation can be expressed as follows: (where fo = the observed values and fe = the expected frequency) 71 The important things to do with the chi-square test are to determine the p- value and distribution‘s degree of freedom (abbreviated as df.) to know which values are meaningful in the survey.  T-test The T-test is a test for group differences, and even when researchers may test two populations that researchers might think they are the same, there exists a variation in somewhat may be significant. Studies, which have been used independent sample T-test in their research questions, vary in many fields of linguistics and TESOL such as checking for English speaking learners of Japanese for pronunciation training course via computer (Hirata, 2010), using formulaic sequences in improving L2 English fluency (Boers, Stengers, & Demecheleer, 2006), investigating benefits of intensive explicit French grammar instruction for English-speaking students (Macaro & Masterman, 2006), and investigating the role of gender factor in foreign language learning of Polish students (Iwaniec, 2015). The T-test determines if the differences between groups are small enough to attribute them to the random variation in scores that would happen each time we take a new sample of the same population, or whether the differences are large enough that the two groups can be said to belong to two different populations. (Larson-Hall, 2016, p. 46) In the paper, the researcher sets up a survey for three groups of participants according to their academic lifetime. The researcher employs the two main kinds of T-tests to the different scores between groups in the survey: the Independent Sample T-test and the Paired-sample Test (also called matched sample T-test). The researcher assigned students randomly to one of two treatments, and later at the end, the scores will be tested for the mean of each of the groups based on the independent sample T-test. When the means are compared and interpreted, the answers to the question that there is a difference between groups to form the total of data results. 72  ANOVA Test The statistical methods employed in this present paper is descriptive statistics. The collected data present the portions of language transfer and intralingual errors by Vietnamese learners of English. To achieve this purpose, one- way ANOVA was employed in the data processing procedure. One-way ANOVA is a test of group differences of three or more groups in using English prepositions. The results of one-way analysis of variance demonstrates how the scores of these groups differ statistically. The one-way ANOVA applies to test one variable with three or more levels or groups to see the significant differences among the mean scores of the different groups (Urdan, 2016). In this thesis, we used one-way ANOVA for both research questions for the three levels of the variables in using English preposition by Vietnamese learners: preposition stranding, pied piping, and null prepositions. Besides it, One-way ANOVA is employed to test the categorical levels of age (teenager, young, and adult), qualifications (secondary school, high school, undergraduate, and graduates), L2 proficiency (low proficiency, intermediate and advanced level). Therefore, one-way ANOVA is used more often than two-way or three-way ANOVA because two-way or three-way ANOVA normally requires at least two independent variables with multiple levels. The questions designed in this research are examined with one independent variable that is the Mean Scores of their performance in using English prepositions. 3.2.2. Empirical design 3.2.2.1 Participants The participants were recruited from undergraduate students at the Ho Chi Minh City University of Food Industry, Ho Chi Minh City on a voluntary basis. There were two groups of students: the experimental and the control group. The initial number of participants in each group was the same at 100 students each. The first group was the control group, who attended a 15-period course on using English 73 Prepositions without a Facebook chatbot. The second group attended ten periods of the class but had the help of a Facebook chatbot. The same teacher taught all students. The students were randomly selected ranked from the low to intermediate based on their TOEIC scores and divided into two nearly equivalent levels. The students enthusiastically did all of their assignments during the experiment. 3.2.2.2 Materials We use the same materials with the survey research design (see 3.2.1.2). A chatbot which was designed to help students learn the material was also built for experimental groups with main contents sourced from the coursebook as mentioned above English Pronouns and Prepositions. This chatbot based on Chatfuel templates in Facebook Messenger. The chatbot is linked to the Facebook page at https://www.facebook.com/englishprepositionuasages/ . 3.2.2.3 Procedures  The control group All students were required to attend three class meetings of five periods each. They were taught ten lessons from unit 13 to unit 22 of English Pronouns and Prepositions (Swick, 2011). Students were supplied with the materials, and teachers delivered the lectures in strict adherence to the course procedures. Then, the students were tested on what they had learned in the same way as the experimental group.  The experimental group Step 1: All students were introduced to the chatbot and given time to do its placement test. Step 2: Students were asked to write the kind of prepositions they had got incorrect during the tests and then given suggestions on which units of the coursebook they should attend the lectures. 74 Step 3: Students in the experimental group could choose the classes on English Pronouns and Prepositions to attend based on chatbot topic suggestions. Step 4: Students were tested and surveyed with the students in the control group. 3.2.2.4 Variables and measurement This empirical research examined the relationship between the learning environment and language transfer. The author employed the descriptive method for describing all the linguistic and nonlinguistic factors, the correlation between the variables under investigation, and the interviews with the participants. Each of the variables had five possible values or outcomes in this experiment, ranked from one to five points. The respondents could only choose one of the options that they believed best reflected their performance or situation. For the test results, the author used a 10-point scale to grade the students‘ performance. The empirical research data were then analyzed and interpreted to determine how well participants in the two groups performed. 3.2.3. Corpus design 3.2.3.1. Preparatory Courses for VSTEP Corpus (PCVC) The author compiled a corpus of English writing and speaking assignments from students enrolled in Preparatory Courses for VSTEP (PCVC) at Da Nang University of Foreign Language Studies. The VPCC was constructed from practice assignments which were submitted online, including 190 written papers and speaking audio files. A text corpus application, Sketch Engine, was used to analyse the data, and the results were uploaded and saved at the URL ske.li/jtl. Table 3.4.VSTEP Preparatory Courses Corpus (PCVC) 75 COUNTS info Tokens 122,227 Words 100,796 Sentences 5,216 Documents 190 3.2.3.2. British National Corpus (BNC) To conduct the study, the author used the British national Corpus (BNC) as the comparison data source. The BNC is a 100-million-word collection of samples of British English written and spoken in the late twentieth century. The BNC is divided into two parts: the written component (90 percent), which includes newspapers, academic books, letters, essays, etc., and the spoken part, which is smaller (remaining 10 percent, e.g., informal conversations, radio shows, etc.). Furthermore, the BNC topics are essentially identical or nearly the same as the topics in VPCC since these topics are quite frequent in language education and English literature courses. It would be appropriate to compare the uses of English prepositions between two corpora. The BNC contains 112,345,722 tokens, 96,134,547 words, 6,052,190 phrases, 1,514,906 paragraphs, and 4,054 documents in total. Table 3.5. British National Corpus (BNC) COUNTS info Tokens 112,345,722 Words 96,134,547 Sentences 6,052,190 Paragraphs 1,514,906 Documents 4,054 76 3.2.3.3. Text collection and Processing The author gathered and processed students' assignment texts for storage in Sketch Engine at https://ske.li/nrn for the PCVC. This is where the author‘s real corpus is being stored and processed. For the purpose of examining how semantic features were employed differently in the two corpora, the author compared preposition use in the BNC to specific preposition usage in the author's corpus. Figure 3.1. A sample of the first 39 concordances of preposition „IN‟ in the PCVC The author used a web-based tool program called Sketch Engine, since it already had a large number of corpora of various languages. The author employed the most basic types of analysis, such as creating frequency lists, keyword lists, and investigating concordance lines. These approaches helped the author to investigate more sophisticated and interesting language patterns of English prepositions in both corpora. The results were organized topically in Chapters 4, 5, and 6. 77 3.3. Summary The author has discussed the rationale for using both qualitative and quantitative methodologies in this research. All the population, materials, and procedures have been described to obtain the research data. The questionnaire was employed as a tool to gather data in the first study design. There were 25 multiple-choice questions in the first exam. The deletion of English prepositions was used to assess their overall semantic accuracy and capability to fill in the blanks appropriately in this test. In the first section, correct responses were coded with the digit "0" and wrong answers were coded with the digit "1" in order to assess the frequency of correct and incorrect answers to the questions in the first section. After that, the participants were asked to describe where an item was located or how it was spatially related to the other things shown in the photos in the second section. There was a total of five photos given in the questionnaire. In the second study design, the author linked Facebook auto messenger with conversational activities to teach English prepositions, allowing learners to connect automatically at any time and from any location. Preposition Chatbot has shown to be a very valuable tool in encouraging students to participate in the course, particularly in practice activities using the mobile-based application that was accessible via social media networks. When the students had to manage to accomplish the tasks set to them by ITC apps, they quickly became used to a new manner of learning, particularly when a chatbot was used in the course. This empirical investigation assisted the author in determining how the learning environment influenced student performance and reduced language transfer when Vietnamese was not included or limited during their learning process. In the third research design, a corpus-based method was employed to assess the frequency of English preposition usage from a conceptual transfer perspective. The author built up the VSTEP Corpus Preparatory Courses (PCVC). This corpus 78 was used to analyze the frequency of prepositions in their writing essays and oral presentations. The concordance output was compared to the British National Corpus (BNC) as benchmark data to establish the frequency of English preposition usages and error categories. The findings indicated that there was a negative correlation between prepositional senses and their Vietnamese collocations. The method assisted the author in defining negative conceptual transfer with solid evidence. The author chose to employ mixed methods in this research to get a firm conclusion on a complicated phenomenon, language transfer. The findings were given quantitatively and qualitatively in the most often seen erroneous usage and an analysis of the evidence of mother tongue influence in these inappropriate uses of English prepositions. 79 Chapter Four NEGATIVE L1 TRANSFER IN THE USE OF ENGLISH PREPOSITIONS BY VIETNAMESE LEARNERS OF ENGLISH This chapter summarized the results of the research, which were derived from both quantitative and qualitative data. To begin, the quantitative data were supplied, and after that, the qualitative data were displayed and assessed for hypothesis testing from one to eight with the use of descriptive statistics. The null hypothesis was only true if the difference between the two languages did not account for CLI in Vietnamese learners' usage of English prepositions. This chapter was divided into eight main components, each of which was validated by multiple data sources from the three research designs. 4.1. Factors related to crosslinguistic similarity To test the first hypothesis, the author presented the data that related to language distance, typological proximity or psychotypology. This way of presenting data was comparable to the results of several well-known studies which were conducted by Kellerman (1983), Ringbom (1978a), and Henning (1978). Firstly, the author summarized the statistical difference between Vietnamese and English. The number of preposition entities in Vietnamese and English languages was vastly different. The Vietnamese prepositions were only approximately up to 22% in comparison with those in the English closed system. 80 Figure 4.1. Vietnamese and English preposition comparison According to the results of the third research design, the difference was so enormous that Vietnamese learners of English got lost in their search for appropriate English prepositions when they needed to use them in their L2 production (English). The findings indicated that participants used three English prepositions with the highest frequency among the 68 entities that served as English prepositions: 'IN' (2,357 times), 'OF' (1,721 times), and 'FOR' (1,625). Students in the preparatory course for VSTEP used ‗words‘ that functioned as English prepositions in their compositions were ‗not English prepositions‘ such as that („rằng‟, 593 times), because (‗bởi vì‟, 213 times), Although (‗mặc dù‘, 18 times), whereas (‗trong khi đó‘, one time), and forth (1 time) (see Figure 4.2). [CATEGORY NAME] [PERCENTAGE ] [CATEGORY NAME] [PERCENTAGE ] Vietnamese Prepositions English Prepositions 81 . Figure 4.2. The frequency of English prepositions in Writing assignments (Source: PCVC) (Source: Data from the PCVC corpus in the third research design) All of these mistakes were caused by the presence of L1 entities in the L2 products of Vietnamese learners of English. Vietnamese learners of English were unable to adequately express their prepositional meanings in a range of English prepositions due to a limitation in the number of Vietnamese prepositions available. As a consequence, five of the English prepositions (namely, ‗IN‟, „OF‟, „FOR‟, „TO‟, and ‗BY‟) were the most often used in their communication and preposition „IN‟ got the highest frequency. It meant that some of prepositions were highly repeated in their writing and speaking. When it came to their productive skills, it was evident that Vietnamese learners of English were using ‗L1 prepositions.' The impact of this phenomenon on Vietnamese learners weighed down their English proficiency. Secondly, in the first survey design, spatial prepositions such as „IN,‟ „ON,‟ „AT,‟ „UNDER,‟ „ONTO,‟ „BEHIND,‟ and „OVER‟ showed a strong correlation with cross- linguistic similarity in Vietnamese learners' English comprehension and production. Vietnamese speakers had the concepts of ‗trên‘ („ON‟/ „OVER‟) and ‗dưới‘ („UNDER‟/ 82 „BENEATH‟) which were quite different from its English equivalents. Participants tasked with describing Picture B came up with a sentence like "The fan is hanging under/ above/ below the ceiling or roof." There was something wrong with these responses, and their sentences were not articulated in a natural way in English, either. Actually, English speakers said that ―a fan is typically ‗fitted TO the ceiling‘ or ‗hung FROM the ceiling‘ or ‗fixed ONTO the ceiling,‘‖ but they did not specifically state that ―a fan is „UNDER‟ or „BELOW‟ or „ABOVE‟ or „AMONG‟ the roof or roofs.‖ Figure 4.3. Picture B – in the task of describing a picture When asked to describe picture B, the majority of Vietnamese participants in the first survey answered with the preposition „ABOVE‟ in their description. The reason for this was because in Vietnamese, ―cái quạt máy ở trên trần nhà‖, or ―bóng đèn ở trên trần‖ was widely used. It was obvious that the word ―trên‖ in Vietnamese created specific conceptualizations based on their embodiment recognition. The participants were not able to see the two things as belonging to the same frame of reference. It was evident that participants viewed themselves as TRs for the things they mentioned (see Table 4.1). 83 Table 4.1. The distinction between English and Vietnamese TR and LM of the „fan‟ and „ceiling‟ in Picture B Vietnamese English Landmark Their head/body The ceiling Trajector The fan The fan Using the survey results, the author discovered that the average score for the survey is really low, with a mean of 41.47/100 points. The standard devation between groups differed substantially across three groups: males (7,278), females (16,690), and "prefer not to say" (14,331). The issue of gender was to be further explored in the Chapter 6. More than 58.53 percent of the respondents were unable to identify the use of English prepositions (see Table 4.2). It was inferred that participants had different opinions on each of the items tested. The high standard meant that their scores were scattered for different English prepositions in their choices. It was concluded that when prepositions were polysemous or not the same in both languages, transfer from L1 structure to L2 structure occurred. This was especially true when English