Relationship Between Writing Skill and Oral Performance of Taiwanese Students

English is the most important foreign language in Taiwan because of globalization. Hence, receiving a satisfactory grade in English class is required to all college students before they graduate. In learning English, oral performance and knowing how to write articles and reports are used to evaluate the student’s proficiency ability. To understand this issue, this study investigated the relationship between writing skills and oral performance in a sample of 32 students who were randomly selected from a population 140 students in 2 classes. Students’ age and gender were used as independent variables while oral performance and writing skills were used as dependent variables. Regression analysis was used to measure relationship between students’ oral performance and writing scores. The regression results show that better writing skills significantly improve oral performance as the age of a student increases. Further, students coming from areas other than northern Taiwan present a negative relationship between their writing skills and oral performances. The results of this study present useful findings that inform Taiwanese teachers of English about the importance of improving writing skills on oral performance.


Introduction
Taiwan benefits significantly from globalization such as international trade, knowledge distribution and foreign direct investment. As a result of this situation learning English has increased as the government has paid special attention since English is the language that is used in almost all transactions and international communications. In addition to economic transactions, development of new techniques and exchanges of agricultural commodities are usually firstly released in English version. Therefore for Taiwanese to catch up new information, it is very important to learn and use English. For this reason, universities in Taiwan require students to obtain satisfactory English grade before they graduate. In other words, equipping desired English level is one requirement to all Taiwanese college students. In routine works such as preparing a report or a business document, writing is the sole skill that may be involved. However, as international trade is intensified in Taiwan, it is very common to communicate in English with business partners such as giving a speech in a formal conference or having a conversation in meal time. Therefore only being able to write documents is not enough to survive in modern society. Thus, equipping desired level of English in both writing and oral ability would benefits people in their current job search or future professional career all over the word. Indeed, improving students' English ability become much more important since Taiwan is inevitably affected by globalization as most of its economic, social, and educational activities are linked to the United States., In order to not fall behind, students and graduates who want to pursue their career internationally or advanced education have understood the necessity and importance of learning English [1][2]. It is reasonable to see why the Taiwanese government mandates students to take at least one full year of English class. However, this does not mean that all Taiwanese college students can write or speak English fluently; instead, this policy is set to ensure that people who has a bachelor's (or higher) degree is capable of communicating in English to foreigners and understands the latest news and information.
Taiwan is a small island that does not even have enough natural resource to satisfy domestic consumption; most of commercial and agricultural products rely on international trade. For this reason, being able to communicate with foreign companies if of importance and English is the most important tool to meet the requirement.
The study aim of this study was to explore the factors that influence Taiwanese students' writing and oral skills in English language.. To investigate this question, this study developed a set of multivariate regression models to evaluate the potential factors that influence students' oral and writing performances and discuss how writing skills may be related to oral performance. The results can be useful to the faculty staffs who are involved in teaching English because they will focus on important factors that will enhance students' English performance and stimulate students' spirit in English study. Moreover, the results can also benefit personnel who are involved in recruiting students who are proficient in English language.

Literature Review
Learning and improving English have been studied widely and intensively from many aspects like engaging (reading), producing (writing), and talking about texts across the contents and levels of education through a variety of strategies [2][3][4]. Some studies are based on defining the characteristics of English learning strategies and autonomous learning [3,5] while some, through needs analysis and the investigation of freshmen's English, the paper believes that it is time for college English teaching to shift gradually to English as a second language program (ESP) [6][7]. For example, Zhang [5] discusses the relation of college English teaching with General English teaching, and the issues of textbooks and teachers involved. He verifies this idea through an empirical research on the impact of English language learning strategies based on autonomous learning. Another study [8] explored and suggested that a central strategy may improve the literacy education of marginalized students. In addition, Fairclough [9] showed that an intimate relationship exists in critical language awareness and development of personal language ability and practices of many people. This implies that skills of learning a language may potentially influence the sophisticated use of the language. However, one problem associated with this study is that it did not tell whether improvement of the efficiency of the use of a specific language can be made via practice.
Moreover, Hanushek [10] showed that using quality as a primary variable to represent school and teacher is crude because the classification of the quality is rough, which results in a core problem to the analysis. For this reason, it will be difficult to indicate how such characteristics are related to specific teachers and alter the results. Another study [11] accommodates the advantages of the mobile learning to present a personalized intelligent m-learning system (PIMS) by recommending English news articles to learners based on the non-English learners' reading abilities that are evaluated by the proposed fuzzy item response theory (FIRT). A previous study [12] indicated that the level of degree alone neither classifies the colleges of differing quality nor does conveys any useful information about major of students, certification requirements fulfilled in college, or subsequent professional development. Under such circumstance, it will be necessary to quantify how scores can be transformed through the dedication from teachers and students.
This study focused on education-related variables that investigated the core questions raised by previous studies associated with input and output transformation (i.e. learning and scores in this case). Ehrenberg and Brewer [13] believe that performance of students is positively related to the quality of colleges where the teachers are coming while Monk and King [14] point out that reparation of a teacher in certain subjects such as mathematics and science may not result in a positive impact on student achievement. In addition, Yao [15] shows that how to inspirit is much more important. Based on their results, this study accommodates several important factors not only potentially influencing students' writing efforts in terms of their term scores, but also their oral performances.

Methodology
The study sets up a group of multivariate linear regression models to examine the relationship between a dependent variable, Y ij where subscripts are defined as the score of j th class for the i th student, and advanced writing, business writing, and oral class are defined from j = 1 to j=3, respectively. To evaluate the relationship among classes, five predictor variables X 1 (age), X 2 (gender), X 3 (height), X 4 (weight), X 5 (dummy variable of living address) are also used.
The study is designed as below: the author firstly estimates the relationship among dependent and independent variables in the first session (as will be shown in model I), and then estimates whether the performance in the business writing class is influenced by that in the advanced writing class, which is expressed in model II. Basically, the reduced form of a multivariate linear regression model is expressed as equation (1).
Thus, the model I can be expressed as where equation (2) describes the basic form of the multivariate regression model, e i is a random fluctuation (or error) in Y i1 and thus E(e i ) in equation (1) is equal to 0.
This equation is used to analyze the students' performance in their advanced writing class. In this case the response variable Y i1 (performance in advanced writing class) is predicted from 5 predictor variables X 1 , X 2 , …,X 5 and the relationship between Y i1 and X 1 ,X 2 , …,X 5 is linear in the parameters β 0 ,β 1 ,β 2 ,...,β 5 . To see how student's performance of business writing class in the second semester is related to that of the advanced writing class, the model I is further modified to model II With such modification, equation (3) shows how student's performance in the business writing class in the next session can be predicted by their advanced writing classes in the first session. In addition, to understand how oral performance is related to students' writing skills, the study adds one additional, dependent variable (Y i3 =Oral) to examine how this factor is explained by the writing performance in the first session. The model III is thus shown as below: With equation (4), the relationship between oral performance and advanced writing class may be explored by classifying the students' characteristics and their writing skills.

Variable Description
The data contains 32 effective observations randomly drawn from 2 classes. The population is 140 students in 2 classes and the probability of a student to be selected in the analysis is assumed to follow a normal distribution. To avoid data selection bias, the students are randomly picked from writing classes by the computer program. In addition, because complete data set of certain students is not available, only 16 students are randomly picked from each class.
The student's performance in writing classes is used as dependent variable in equation (2) and (3) to investigate their score fluctuations to various characteristics. To examine the relationship between oral performance and writing scores, the dependent variable of oral class is set in equation (4). The independent variables are selected based on their potential influences on students' test performance. X 1 is age, which potentially reflects the mental maturity of a student, implying a potential difference of capability of organizing and structuring an essay. X 2 is gender, describing the natural difference among students. X 3 is height, showing the genetic difference in nature. X 4 is weight, which potentially describes the nutrients they receive. X 5 is dummy variable for residential address, all of which potentially affect student's writing performance either directly or indirectly. With such formulation of regression models, the study can evaluate and measure the difference of their writing and oral performance in terms of term scores.

Regression Results and Discussion
The regression result is shown in table 1. The R-square of regression set (1) to (2) is 0.321and 0.604 for model I and model II respectively, showing that the data descript the models is just fine. However, if we use cross components to examine the relationship of model II as shown in regression set (3), the goodness of fit increases to 0.715, implying the model II is improved. The variable writing1 used in the regression sets is the advanced writing performance, m1 is the cross variable male* writing1, and n1,mm1 and s1 are dummy location*writing1. The small R-square in regression set (1) can be explained by that some important but unknown factors may be ignored in the model and by that the sample size is not sufficiently large. However, when business writing performance is taken into account, the model is improved. In addition, the results show that only gender is significantly related to students' advanced writing performance, and male students perform worse than female students in this class. This finding is less interesting because, in general, it is commonly recognized that women have better performance in skills such as paragraph organization, logic presentation, and essay structure. In addition, previous studies also indicated that male students often are good in math and science that is highly related to logic presentation. Up to this point, the less interesting results bring an interesting question: why better logic presentation of man does not result in a good writing skill? One possible explanation for this question may be that the organization and structure may outweigh the importance of logic presentation in writing. In other words, this result implies that male students may not organize what they are writing as good as female students. However, this variable is significant in the 90% confidence level and as more data come in, the result may be altered. Now let's move to how advanced writing class affects the business writing class. The regression results show that the students do well in advanced writing class can also perform well in business class. It is explained as that students perform well in the advanced writing class may already embed a desired ability in writing and thus, they can also get good grades in the net writing class. In addition, living region does have a significant influence on student's business writing performance. Specifically, the result shows that students coming from northern Taiwan perform better than those coming from other regions. This is because northern Taiwan is the most developed region in terms of economic condition, educational resource, number of university, average salary, etc., and therefore, students coming from this region may understand more about economy and business, which help them to write business-related articles. The result in regression set (3) is based on the cross term regression model, in which shows that male student whose performance in advanced writing class is positively influencing his business writing scores. This is shown in the m1, representing the male* advanced writing performance. The same condition of the small R-square in regression set (4) can also be explained by that some important but unknown factors may be ignored in the model and by that the sample size is not sufficiently large. This is because the sample size from various regression sets is the same and the less the variables we use, the less fitness we can observe. However, when the relationship among independent variables is jointly analyzed, fitness of the model is improved. The result indicates that students coming from central and southern Taiwan do worse in oral skills. The may be due to that the underdevelopment of those regions, compared to Taipei, and therefore, students can write well but do poor in oral communication. Combining (4) and (5), it is also clear that as the age of a student increases, they can improve their communication skills significantly. From regression set (6), the result also indicates that students' writing performance and oral skill are negatively related, indicating that students may either have no access to better education in terms of lower quality of teachers or simply do not have time to learn both, especially for female students.

Conclusion
This study develops a set of multivariate regression model to evaluate the potential influences of important factors on oral and writing performance and discuss how writing skill may be related to their oral performance. The Study has shown that equipping a desired level of English is much more important in modern society. As Taiwan is highly relied on the import and export, proficiency in English has been continuously been given important attention in the education system. The results imply that the education level in Taiwan is not uniformly distributed and more resources (in terms of quality of school and teachers) are allocation in northern Taiwan. Students coming from central and southern Taiwan have less access to better education resource and have less ability to improve both writing and oral skills at the same time. Although the oral skill is shown to be improved as age increases, it is not a good signal for education system in the long-run. Therefore the government has to understand such issues and try to allocate education resources more efficiently. Further research would highlight appropriate strategies of improving English language teaching in Taiwan especially in areas where the teaching of English language is less developed. For example, the government may setup some subsidy policies in the forms of tax deduction or direct compensation to encourage teachers to move to schools in southern Taiwan. The government expenditure may increase in the short-run, may the efforts may eventually enhance the education quality in the long-run. However, the impacts should be analyzed both qualitatively and quantitatively so the influences from such policies can be determined.