Theoretical Framework REVIEW OF LITERATURE

Department at Universitas Sanata Dharma in 2016 academic year and had passed the Structure I until Structure V. A sample is a smaller number of observations taken from the total number making up a given population Sprinthall, Schmutte, and Sirois, 1990: 28. Furthermore, Creswell explains that sample is the subgroup of the population for generalizing about the target population 2015: 141. The chosen sample must represent the population or otherwise the conclusion can not be accepted Sugiyono, 2013: 149. Therefore, there is a systematic way to decide how many people who can be taken as the sample of a research. Here is the formula by Isaac and Michael in Sugiyono 2013, 158: Chart 3.1 Sample Formula � 2 . N. P. Q s = ------------------------------------ � 2 � − + � 2 . . s = the amount of sample λ 2 = Chi Quadrate that depends on the degree of mistakes 1; 5; and 10 N = the number of population P = the opportunity of true 0,5 Q = the opportunity of false 0,5 d = the difference between sample’s average with population’s average. It could be 0,01; 0,05 and 0,1 The population of seventh semester students who had passed Structure V is 111 students. For the first half test, the researcher gave the test and the questionnaire to 24 seventh semester students. Due to the need of taking sample as the representation, there would be 83 students of the seventh semester see appendix 1.

a. Test

A test is a method to measure someone’s ability or knowledge in a given area which has the purpose of measuring Brown, 1987: 219. The kind of test given was called achievement test, a test that was related to classroom lessons, units, or curriculum and limited to particular material Brown, 1987: 225. The test would be divided into three parts. The first part aimed to find out whether the seventh semester students had acknowledged enough the different function between make, have, get and the active form of causative. The second section aimed to find out the comprehension of passive causative form. The third section would consist of four questions where the students translated the question in Indonesian into English in causative form. The test done by the researcher contained of two kinds of test. The first test was given to measure the validity of the questions that the researcher had writen. Here, it can be said that the first test was the experimental test in order to get the result of the number of questions that the researcher can use for the real test see appendix 6. The second tes t was the actual test that is used to measure students’ comprehension about the topic see appendix 7. To be clear, the first test was the test given to 24 students, which were taken in Seminar on Language Class A on 7 November 2016. The second test was the test given to 83 students that were held in Pragmatics Classes on 23-28 November 2016 see appendix 2. The test result used for analyzing the mastery of causative that students had is the second test.

b. Questionnaire

Creswell in Sugiyono demonstrates that questionnaires are forms used in a survey design that participant in a study complete and returns to the researcher 2013: 230. The questionnaire in this research would be a closed questionnaire. The instruments would be focusing on: 1 the way the students learned and practiced Structure and 2 the comprehension of causative material especially make, have, and get. The questions would be multiple choice questions that each option had different score. Based on the experiment done in the first test, the researcher gave 10 questions in the questionnaire which question 1, 2, and 4 are proven not valid see appendix 8. Later, the researcher will only consider the rest valid questions.

2. Data Analysis

a. Reliability and Validity

The test has to through the process of validity and reliability of instrument in order to be dependable. Reliability measures the instrument in this case is the test to achieve consistent result Brown, 1987: 220. Consistency in this context means that the result will be the same from the day the test taken with the result taken next week after, or at least fulfil the same expectation. Reliability in this research would be calculated by using the split half technique where the test would be divided into two