CHAPTER FOUR RESEARCH DESIGN
4.2 RESEARCH METHODOLOGY
4.3.1 Phase One: The survey
The aim of the survey was to determine the views of lecturers of MFL at the selected institutions on the provision of MFL according to the aspects represented by the respective sections of the questionnaire.
4.3.1.1 Sampling
a) China
As mentioned (see 1.1.4), more than 400 universities and colleges offer various Chinese programmes to foreign learners. Three main categories can be distinguished: first class universities, base universities and non-base universities (see 3.5.2 & 3.5.3). The first class
universities are large, prestigious institutions, have the longest history, the largest scope, and the most qualified teaching faculties in the area of teaching Chinese language and culture to foreigners. Comprehensive universities are the base of Mandarin teaching as a foreign language in China and offer Mandarin to a certain scope and level, have qualified teaching staff and advanced facilities. Both of these kinds of universities offer courses for both degree and non-degree Mandarin education for foreigners and have large and stable student numbers every year. Non-comprehensive universities have a clear specialty such as technique, economics, foreign languages and so on; they only offer Mandarin training to a relatively small number of students and do not offer degrees.
Purposeful sampling and convenience sampling were used to select three universities from this huge range to be sites of inquiry and sources of suitable participants for both phases of the study. Purposeful sampling is a procedure where the researcher identifies key participants: persons who have some specific knowledge about the topic being investigated (Lodico et al 2006:140). Convenience sampling is the least desirable sampling method and it includes samples that are convenient (Lodico et al 2006:141). Beijing Language and Culture University (BLCU) represents the first class universities in TCFL (see 3.5.1). It is the only international university in China whose main task is teaching the Chinese language and culture to foreign students. BLCU has the longest history, the largest scope, and the most qualified teaching faculties of its kind in the area of teaching the Chinese language and culture to foreigners in China (see 3.5.1). Nankai University (NU) represents a base university approved by the Ministry of Education in China (see 3.5.2). NU became one of only a few Chinese higher educational institutes to start admitting international students in 1954. Tianjin University of Science and Technology (TUST) is a newcomer to Mandarin as FL teaching and is representative of the non-base universities (see 3.5.3). Its foreign student enrolment is not very big or stable. I have lived in Tianjin and have established professional relationships with colleagues at BLCU and TUST; this facilitated the gathering of the data.
A non-probabilistic convenience sampling procedure (McMillan & Schumacher 2006) was used to select respondents for the questionnaire from the selected universities in both countries. At the time of the survey (June 2012), there were just under 100 academic staff teaching MFL in BLCU (BLCU 2011), 51 academic teaching staff at NU (see 3.5.2) and 7 at TUST (Yang 2011). The respondents for the questionnaire were made up as follows: at BLCU, 30 teaching
staff were selected with the help of a gatekeeper and 28 completed the questionnaire. NU employs a total of 51 teaching staff for TCFL and 11 of these were present on the day of data collection.
They all completed the questionnaire. At TUST, all 7 teaching staff completed the questionnaire.
b) South Africa
The researcher targeted all four universities offering MFL in South Africa: Unisa (see 3.8.1), Stellenbosch University (SU) (see 3.8.2), Rhodes University (RU) (see 3.8.3) and the University of Cape Town (UCT) (see 3.8.4). Compared to China, the MFL total number of teaching staff is very small so all of them were selected as participants. The total of 8 lecturers were selected as survey respondents, however, only 6 completed the questionnaires. These were distributed as follows: Unisa has 3 lecturers of which 2 are permanent positions and 1 is a secondment by the Hanban for a 2-year term. Two of the 3 completed the questionnaire; one member of staff declined to complete it. SU has 3 staff members, 2 of which are from Hanban of China seconded for 2-year contracts. One staff member has a 5-year contract position. As the new contractor had only arrived in 2012, she was not yet qualified to complete the questionnaire. RU and UCT each have only 1 professor seconded by Hanban of China; both completed the questionnaire.
4.3.1.2 Phase One: Data collection by survey
The survey is a research method for gathering information. The various approaches to survey research have the same purpose: gathering opinions, beliefs, or perceptions about a current issue from a large group of people (Lodico et al 2006:157; Johnson 2008:94). A questionnaire is relatively economical, asks the same questions of all subjects and can ensure anonymity (McMillan & Schumacher 2006:194). The use of questionnaires at the start of a project can often be very useful because it helps the researcher to collect a range of information with relative ease, which can then be followed up as necessary (Koshy 2005:87; Lodico et al 2006:159).
Questionnaires enable a researcher to collect background and baseline information quite easily. They can help the researcher to gather a reasonable amount of data in a short time, can provide information which can be followed up on and can provide a format which makes it easy to represent information. In addition, a questionnaire is suitable for collecting initial information on attitudes and perceptions (Koshy 2005:89). Online questionnaires have an added advantage of reaching a relatively large sample group, and they can be spread over different geographical locations. Questionnaires can shorten the distance and time needed to collect data.
For these reasons, a survey questionnaire was considered the most efficient means of data collection in Phase One of this study. It was used to obtain an overview of the teaching practices used in teaching MFL from the lecturers in the selected universities in China and in South Africa. The questionnaire was self-designed for the study, pilot tested and then refined (Lodico et al 2006:168-169). Three academic staff were chosen for the pilot test: 2 from China (1 from BLCU and TUST respectively) and 1 from South Africa (Unisa). Pilot respondents were requested, and given the opportunity, to highlight any perceived lack of clarity and problematic questions or shortcomings in the questionnaire. The critique was analysed with expert peer assistance using the promoter of the study and an expert statistician, and changes were made to the final survey questionnaire where it was deemed necessary.
The questionnaire comprised 7 sections: Section A dealt with biographical data (8 items); Section B dealt with institutional factors (19 items); Section C dealt with lecturers‟ perceptions of MFL students (23 items); Section D dealt with teaching methods (23 items); Section E dealt with curriculum (17 items) and Section G dealt with degree purposes for honours and postgraduate students (8 items); Section F was an open question. Sections A to G comprised closed items, in which subjects choose between predetermined responses (Koshy 2005:87; Johnson 2008:94-95; McMillan & Schumacher 2006:197). The questionnaire was translated into Chinese by the researcher.
The questionnaire was distributed in both English and Chinese to all respondents (see Appendix A). The questionnaire was accompanied by a covering letter (see Appendix A) stating that it was not necessary to indicate a name and that the anonymity of the respondent was assured. An explanation of the purpose of the research, a request for participation, an
advisory of the right of refusal or withdrawal, an assurance of confidentiality, and an explanation of the survey instrument and how to complete it were also given in the letter. The respondents could not proceed from one section of the questionnaire to the next unless they had acknowledged (by selecting an option) that they had read and understood the information and requests and were willing and able to move on to the next section. The questionnaire concluded with a note of thanks.
In total, 52 questionnaires were completed (46 in China and 6 in South Africa). The response rate was as follows: 93% return rate from BLCU and 73% from NU; 100% from TUST and 75% from the four South African universities combined. Thus the overall response rate was 87% which was considered satisfactory.
4.3.1.3 Data analysis and presentation of data
Quantitative research relies heavily on numbers in reporting results, sampling and providing estimates of instrument reliability and validity (Koshy 2005:109). Statistical analysis was used to analyse the data gathered by the survey in Phase One. Statistics are methods of organising and analysing quantitative data. Statistics is an international language that only manipulates numbers. Statistics and numbers do not interpret themselves, and the meaning of the statistic is derived from the research design (McMillan & Schumacher 2006:150). To make sense of the data, the researcher generally begins to summarise them in the form of descriptive statistics (Lodico et al 2006:243).There are two broad categories of statistical techniques: descriptive and inferential. Descriptive statistics transform a set of numbers or observations into indices that describe or characterise the data. Inferential statistics are used to make inferences or predictions about the similarity of a sample to the population from which the sample is drawn (McMillan & Schumacher 2006:150; Lodico et al 2006:243-247).
The data gathered by the questionnaire was analysed by an expert statistician using the SAS software programme (cf.1.5.2.6). The following broad analysis strategy was followed:
Frequency distributions on the biographical properties of respondents (country wise). This allowed for a description of the sampled respondents as well as the Chinese and South African subsamples.
Two-way country-wise frequency distributions on subsets of „similar-topic‟ questionnaire items for sections B-D of the questionnaire. Furthermore for each frequency table, Pearson‟s Chi-square tests and Fisher‟s exact probabilities associated with the relevant Chi-square statistic, and Cochran-Armitage trend tests were conducted. The purpose of the two-way tables was to introduce the sub-constructs and investigate whether response patterns to all questions within a subset agree or differ statistically significantly for some question statements. Apart from the general perception trend that the frequency distributions of the subsets of questions reveal, response patterns of individual questionnaire statements that differ from the general response pattern of a subset of questionnaire statements, provide more in-depth knowledge on the sub-construct.
Scale reliability tests on the subsets of responses for the sub-constructs to verify the internal consistency reliability of the sub-constructs. This determined whether sub-construct perception measures derived from the subsets of responses for a particular sub-construct are reliable measures of perception.
Analyses of variance (Anova) on the derived sub-construct perception measures to determine whether country (China/SA), degree or non-degree institutions, and pass rate of students affect perceptions regarding aspects (sub-constructs) of MFL teaching. In each analysis a sub-construct measure of perception was entered in the analysis of variance model as the dependent variable and the properties of the country, degree or non-degree institution and pass rate were entered as explanatory variables in the analysis anova model.
Throughout results, the statistical significance was indicated according to a general protocol, namely:
Significance legend:
* Significance on the 5% level of significance ** Significance on the 1% level of significance *** Significance on 0.1% level of significance
The results of the analysis are presented in Chapter 5 in the sequence in which the questionnaire was designed, that is according to sections A-G. In each section (except section G), country-wise frequency distributions per questionnaire item or subset of questionnaire
items are presented first. In some instances, a second set of comparisons based on degree or non-degree MFL course distinction is also listed. This is then followed by a summary table of results or findings and interpretation of findings.
The significance legend has not been included in each and every table. It is made up of the statistical significance protocol followed in the study. A number of frequency distribution tables have been included in the report. Some of the „less important‟ tables (for example the general frequency tables in sections B, C and D – the 16-; 20-; and 23-item tables, since their content are also presented in the sub-construct tables) is included in appendices.
4.3.1.4 Reliability and validity
Because correlational research involves the measurement of many different variables, the quality of the measures used is a major consideration. Reliability and validity are two concepts that are used to judge the quality of educational measures. Reliability refers to the consistency of measurement. Validity refers to whether the measure accurately and appropriately measures whatever it is supposed to measure. If no evidence of reliability or validity is provided or available, the results of the study may be suspect (Conrad & Serlin 2006:406-407; Flick 2007:5-6; Lodico et al 2006:230). In order to ensure the consistency of measurement, all data was collected, analysed and interpreted in the same way.