Analyzing the Influence of Task-Based Teaching Method on Students’ Language Proficiency Enhancement in College English Teaching Based on Big Data

The application of task-based teaching method in the teaching of English majors in colleges and universities refers to the learning task-oriented, so that students can have a deeper understanding of English language knowledge while completing the learning task, and enhance their ability to learn and apply. Under the application of task-based teaching method, students dare to express their ideas boldly by utilizing the tasks, strengthen their application of language knowledge, and provide students with more opportunities for communication and interaction under the environment of language tasks, which are of great significance in promoting students’ acquisition of language knowledge and enhancing their ability to use English knowledge.

1)

Objectives

The application of task teaching method in English teaching in colleges and universities will formulate targeted, personalized and stage-specific learning objectives according to the actual learning situation, and take the learning tasks as the focus of English learning, design and arrange learning tasks that match the needs of the discipline, and ensure that students use English to express themselves as much as possible in the whole process of learning. Task-based teaching objectives are targeted and can transform specific classroom interactions into executable learning tasks, which are more operational.

In the application of task-based teaching methodology, design is a very important link, in which it is very important to ensure the structure of task design. In general, the application of task-based teaching method mainly contains the following aspects, and the task design should also be implemented in combination with these aspects.

First of all, clear teaching objectives. The most important thing of task-based teaching method is to practice students’ practical ability, so the goal is to improve students’ English communicative ability and the use of language knowledge, which includes not only the basic listening, reading and writing skills and mastery of grammatical knowledge, but also the use of knowledge of the English language, discourse analysis, language measurement ability and sociolinguistic competence and so on.

Second, input. Teachers need to apply the corresponding reference materials when assigning learning tasks, which is called input. For example, teachers use magazines, songs, movies, textbooks and so on as the source of designing tasks, all of which belong to the content of input and the scope of instructional design.

Finally, the task environment. In the process of students’ execution of tasks, tasks and roles need to be assigned, and ways of individual completion, group collaboration and so on need to be formulated, which all belong to the task environment.

The application of task-based pedagogy has a certain phase type, which is divided into three different phases: before, during and after the task. The pre-task stage is the preparation stage, which includes clarifying the purpose, making plans and so on. The task implementation stage is to carry out the implementation of the task to complete the corresponding learning objectives. The latter stage is to consolidate the content of the task through the corresponding organizational forms in order to promote the improvement of students’ learning efficiency. First of all, before the task begins, it is necessary to clarify the fundamental purpose of this task and weigh the students’ English learning level, their interest in learning, and their overall cognitive situation according to the actual situation of the students, so as to ensure the operability and authenticity of this activity. Afterwards, students are guided to search for and find the relevant information they need to use in the process of the task, and they can maintain a high level of enthusiasm for learning in the process of the task. Secondly, in the task implementation stage, teachers should play the role of their own teaching guide, guide and organize students, and supervise the completion of the students. Teachers in the design of the task should be the task implementation of the knowledge needed to prepare for the organization of the processing, and informed students in advance, so that they can do a good job of knowledge reserves, you can look for relevant information, which facilitates the completion of the task, but also conducive to enhance the students’ self-confidence in learning, and to promote the growth of the students.

Finally, after the completion of the task, the teacher should make a pertinent evaluation of the students’ performance, in which the good performance of the students to give affirmation and encouragement, for the implementation of the task in the process of the problems of the students are guided accordingly and commented on to encourage them to make corrections. At the same time, teachers should also guide students to have a comprehensive understanding of the task and summarize, and guide students to consolidate and review the key knowledge involved in the task, so as to promote the overall development of students.

In the application of task-based teaching method, it is necessary to practically start from the actual situation of the students, to combine the cognitive level and previous experience of the students, and not to consider the problem purely from the teacher’s point of view, so that the tasks designed can be more targeted. Therefore, in the process of task design, it is necessary to take the students’ communicative ability and English knowledge application ability as the basis, and establish the task on the basis of the students’ known experience, to ensure that all the students can actively participate in the task practice, and to ensure the authenticity of language communication.

Univariate linear regression mainly means that there is only 1 independent variable in the regression model, and the functional equation of univariate linear regression can be noted as h(θ) = θ₀ + θ₁x, where θ₀, θ₁ is the parameter that needs to be determined after the linear regression model has been trained, the θ₀ parameter stands for the meaning of the intercept, and the θ₁ parameter stands for the meaning of the slope of the fitted image. In order to solve for the θ₀, θ₁ parameter, the cost function needs to be kept as small as possible, and in this section, the true value is noted as y, and h_θ(x) is the predicted value, and the error is continuously reduced using the idea of least squares, and the loss function is shown below. Solving the one-variable linear regression equation can be solved using the gradient descent method, as the name suggests, the gradient descent method is in the process of constantly changing the θ₀ and θ₁ parameters, the loss function J(θ₀, θ₁) can reach the global minimum or local minimum, the core idea of the gradient descent method is the use of the loss function of the θ₀ and θ₁ constantly seek the partial derivatives until the seek the partial derivatives of 0, you can get the formula (4) and formula (5), respectively. (3)J(θ0,θ1)=∑i=1n(hθ(x(i)−y(i))2 (4)∂J(θ0,θ1)∂θ0=2∑i=1n(y(i)−θ0−θ1x(i))=0 (5)∂J(θ0,θ1)∂θ.=2∑i=1n(y(i)−θ0−θ1x(i))x(i)=0

Gradient descent method in solving linear regression equations need to carry out continuous iteration, each iteration process needs to repeat a process that needs to traverse all the data in the training set. As you can imagine, assuming that the amount of data in the training set is huge up to hundreds of thousands or even millions, the complexity of the gradient descent method will become very high, and at the same time, the convergence speed of the gradient descent method will become very slow with the increase in the amount of data.

In the process of solving the linear regression equation, a parameter, the learning rate, is used, which represents the rate at which the points in the linear regression equation approach the minimum point. A good gradient descent method has high requirements for the learning rate, which must be moderate, neither too small nor too large, too small a learning rate will lead to a very slow rate of approaching the minimum point, in which case many iterations will be required to achieve the purpose of the solution, when the learning rate is too large, it will appear that the linear regression equation is constantly repeated in the region near the minimum point, and the convergence of the purpose can not be achieved. Usually the learning rate in the gradient descent method will generally take the value of 0.1, 0.03, 0.001, 0.003, 0.0005, etc., in the process of solving the linear regression equation, the learning rate needs to be combined with the actual, choose an optimal value for the selection of generation.

Unlike univariate linear regression, there are more than one independent variables in the multiple linear regression model, and more effort is needed in solving the parameters in the multiple linear regression model, and the equation of the multiple linear regression model can be written as h(θ) = θ₀ + θ₁x₁ + θ₂x₂ + θ₃x₃ + ⋯ + θ_nx_n, where n represents the number of independent variables in the linear regression model. For convenience, the multiple linear regression model can be expressed in the form of a matrix, X→θ→=Y→ , where θ→,X→,Y→ is specifically expressed as in Equation (6), modeled after the cost function given by the univariate linear regression equation, the cost function of the multiple linear regression model can be notated as Equation (7), and after that the parameter matrix is biased by the least-squares matrix method until convergence, and the final parameter matrix is determined, and the derivatives are given in the form of Equation (8), but it is worth noting that if the inverse matrix of the least squares X→TX matrix does not exist, it can be replaced by the gradient descent method for solving. Compared with the gradient descent method, the least squares method does not use the learning rate in the process of solving linear regression equations, which also means that the least squares method does not require multiple iterations, and in the case where the amount of features is not very large, the least squares method is not a good method for solving linear regression equations. (6)θ→=[ w1 w2 w3 ⋮ wn],Y=[ y(1) y(2) y(3) ⋮ y(n)],X→=[ 1 (x1)(1) ⋯ (xn)(1) ⋮ ⋮ ⋱ ⋮ 1 (x1)(n) ⋯ (xn)(n)] (7)J(θ→)=∑i=1n(hθ(x→)(i)−y(i))2 (8)∂J(θ→)∂θ→=2X→T(X→θ→−Y)=0

Descriptive statistical analysis focuses on the statistical description and comparison of the distribution of the variables, and the preliminary statistical results of the variables involved in this study are shown in Table 2.

By analyzing the data, the following conclusions can be drawn: 1)

Among the college students of comprehensive universities surveyed in this study, their language proficiency is in the upper-middle level as a whole, especially the level of conversational language and spoken English, the mean value of which reaches 4.33 and 4.13 respectively, which indicates that they are good at using spoken English to talk with people in daily life and have a high level of speaking and listening to spoken English, and the mean value of reading ability is 3.27, and the data of the survey sample shows that most of the The data from the survey samples show that most college students were able to keep reading several times a month or several times a week in the past 12 months, and the reading volume was at least 4 books, which indicates that college students’ overall reading ability is good;

Before conducting stepwise multiple regression analysis, it is necessary to test the correlation between the relationship between the explanatory variables of this study, language proficiency (conversational language, English speaking level, reading proficiency), and the explanatory variables, influencing factors (parental qualifications, learning resources, self-efficacy, learning anxiety, health status, language attitudes), and regression analysis to find the correlation between the variables is only meaningful if there is a significant correlation between them only if there is a significant correlation between the variables, the regression analysis to find the specific form of their correlation will be meaningful. The results of the Pearson correlation analysis are shown in Table 3.

The results of Pearson’s correlation analysis showed that there was a significant positive correlation between conversational language and self-efficacy (r=0.27, p<0.01), health (r=0.24, p<0.01), language attitudes (r=0.36, p<0.01), parental qualifications (r=0.22, p<0.01) and learning resources (r=0.18, p<0.01) , and a significant negative correlation between and learning anxiety (r=-0.16, p<0.01).

There was a significant positive correlation between the level of spoken English and self-efficacy (r=0.39, p<0.01), health (r=0.32, p<0.01), language attitude (r=0.37, p<0.01), parental education (r=0.24, p<0.01) and learning resources (r=0.18, p<0.01), and a significant negative correlation between learning anxiety (r=-0.17 p<0.01) were significantly negatively correlated with each other.

There was a significant positive correlation between reading ability and self-efficacy (r=0.47, p<0.01), health status (r=0.26, p<0.01), language attitudes (r=0.19, p<0.01), parental qualifications (r=0.12, p<0.01), and learning resources (r=0.33, p<0.01), and a significant negative correlation with learning anxiety (r=-0.21, p<0.01). 0.01) with a significant negative correlation between them. The correlations between the variables were clear enough to proceed to a more accurate regression analysis.

The adjusted R² in the multiple regression analysis represents the share of variation in the dependent variable explained by the independent variables. The adjusted R² in the multiple regression analysis ranged from 0.02-0.13, which is a small effect size; the adjusted R² ranged from 0.13-0.26, which is a medium effect size; and the adjusted R² ranged from 0.26-1, which is a large effect size. The stepwise regression method was used, with stepwise inputs as six independent variables of parental education, learning resources, self-efficacy, learning anxiety, health status, and language attitude to form six regression models. The results obtained are shown in Table 4.

The R² values of the five models become higher and higher with the input of the independent variables, explaining more and more about the dependent variable students’ English language proficiency. Finally the adjusted R² of model six in this study is 0.564, which is a large effect size, indicating that the five independent variables in this study can explain 56.4% of the variance in students’ English language proficiency, and the remaining 43.6% may be affected by contextual variables or other factors, which need to be further explored.

Multicollinearity refers to the phenomenon of model estimation distortion or difficulty in accurately estimating the relationship between variables in a linear regression model due to precise correlation or high correlation between the dependent variables [27]. The presence of multicollinearity in multiple linear regression is mainly tested by eigenvalues, conditional indices, tolerance, and variance inflation factor (VIF). The results of regression analysis multicollinearity diagnosis are shown in Table 5.

The study shows that the eigenvalues of all independent variables are greater than 0.01, the conditional indices are less than 30, the tolerances are greater than 0.01, and the VIF values are less than 10, which indicates that there is no multicollinearity among the independent variables. All the results of the multicollinearity test in this study meet the standard requirements and are therefore suitable for regression analysis.

In regression analysis, this study used the regression standardized residual histogram and regression standardized residual scatter plot to determine whether the residuals are normally distributed and whether there is heteroskedasticity [28].

Figure 1 shows the histogram of regression standardized residuals and Figure 2 shows the scatter plot of regression standardized residuals. Figure 1 shows that the residuals in this study conform to normal distribution, indicating a good fit. From Figure 2, it can be seen that no matter how the predicted values in a specific range change, the corresponding residuals are always near the 0 level line, and the amplitude of the wave remains basically stable, with no obvious signs of heteroscedasticity. It can be determined that there is no significant difference between the standardized residuals and the standardized normal distribution, which indicates that the conditions are met and the regression model is appropriate and feasible.

Figure 1.

Regression normalization residue

Figure 2.

Regression normalization residue

Through a series of basic tests, the adjusted R² value obtained from the previous section is 0.564 as a large effect size; D-W value is near 2, there is no serial correlation; eigenvalue, conditional index, tolerance, VIF value meets the requirements of the study, and there is no multiple covariance in the independent variables; the residuals are normally distributed, and there is no heteroscedasticity, which indicates that the regression equation is scientific and reasonable. The five independent variables were put into the regression model sequentially, and the unstandardized multiple linear regression equation was obtained as:

Y=0.625+0.224*Selfefficacy+0.222*Learninganxiety+0.192*Languageattitude+0.187* Learning resources+0.139* Health status +0.108*Parent degree

The standardized multiple linear regression equation is:

Y=0.225*Selfefficacy+0.216*Learninganxiety+0.196*Languageattitude+0.177*Learning resources+0.132* Health status+0.102* Parent degree

Table 6 shows the regression coefficients and significance test of the equation. From the results of the coefficients of the standardized multiple linear regression equations, it is clear that the factors have a positive influence on students’ English language proficiency: 1)

Self-efficacy has a more significant positive effect on students’ English language proficiency, with the first place of influence. The coefficient of influence of this factor of self-efficacy on students’ English language proficiency is 0.225 and the significance coefficient is 0.004, which supports the hypothesis H2;