3 Methodology
3.5 The sample
An important step in sociolinguistic research consists of defining the sample to be used. Sociolinguistic studies are often characterised by much smaller samples of participants than other research projects (Milroy & Gordon, 2006: 28). Labov (1966: 180-1) argues that statistical representativeness is not a key factor in sociolinguistic research as language use is more homogeneous than other phenomena. Consequently, judgement samples are often employed; the researcher establishes criteria, based upon relevant social variables, which informants have to fulfil in order to be eligible for the study. The choice of variables in a sociolinguistic study is influenced by certain conditioning factors.
Llamas (2007: 14) argues that the researcher must attempt to include as many relevant social variables as possible while assuring a manageable sample size.
Separate samples are employed for the large-scale study of language attitudes and the small-scale ethnographic study of language behaviour.
3.5.1 Sample for the attitudinal study
All the data collection was carried out in secondary schools in Luxembourg. For the attitudinal section of the study, six secondary schools were contacted prior to the start of the fieldwork and access to students and teachers was negotiated. As noted earlier, the Luxembourg education system is largely split between ‘Lycée Classique’ and ‘Lycée Technique’. Classical secondary education aims to prepare students for further studies at university-level. Technical secondary schools, on the other hand, mainly serve as an initiation into various professions but do not prevent students from enrolling into higher education (Berg & Weis, 2005: 20).
The sample for the language attitude study includes students and teachers from both classical and technical schools. The classical schools are characterised by a
first three years of secondary education. This transition phase represents an ideal point in time to investigate attitudes towards the various languages employed in Luxembourg’s schools and the multilingual behaviour among students and teachers. Consequently, the classical school sample consists of students who are enrolled in their last year of German-medium education and students in their first year of French-medium education at the time of data collection. The technical schools are also characterised by changes in the medium of instruction policies after the first three years. However, due to the more vocational nature of this type of school, students are split into numerous different educational sections. In order to keep the sample at a manageable size, only students in their last year of German-medium education were included. This step was taken due to the impossibility of incorporating students from all sections after the changes of language of instruction policies. The sample was also stratified according to the geographical location of the participating schools. Informants were recruited from classical and technical schools located in the north, centre and south of Luxembourg. While the northern schools sampled for this study are within close proximity to the German border, the southern schools are located on the French border (Map 3.1). The central schools are all located in the capital, Luxembourg-city. In addition, both the south of Luxembourg and the region surrounding Luxembourg-city are more densely populated and more extensively industrialised than the traditionally rural areas of the north.
Map 3.1: Areas of the six secondary schools included in the sample of the attitudinal study.
A major concern of the research consisted of creating minimal disturbance in the daily school routine. Consequently, access was negotiated to different classes in the various schools where large numbers of students could be asked to fill in the attitude questionnaire simultaneously. The sampling technique is characterised by elements of random sampling. No exact numbers for the various gender and ethnic categories were established prior to the administration of the questionnaire.
The author arranged with the headmaster of every school to gain access to two classes of students (20-25 students per class on average) of every type (classical pre-medium of instruction change, classical post-medium of instruction change, technical pre-medium of instruction change). This approach did not seek to assemble a sample entirely balanced according to the ethnicity and gender of the informants. Table 3.1 provides an overview of the distribution of the students sampled for the quantitative investigation of language attitudes. For ease of representation, only three major ethnic groups (Luxembourgish, Luxembourgish dual, Portuguese) are represented separately in this table. Students originating from other ethnic backgrounds have been included in the category labelled
‘Other’. Further ethnic groups are represented by Italians (14), students from Ex-Yugoslavia (8), Germans (4) and students from France and French-speaking Belgium (8). Students were assigned to different ethnic categories based on responses they provided to open-ended questions regarding their nationality. Due to the high number of Luxembourgish students who claimed to have two nationalities a separate category labelled ‘Luxembourgish dual’ was employed for the categorisation of informants. These students typically have one Luxembourgish parent and one parent from a different ethnic background. As previously acknowledged (2.2.3), the categorisation of students into separate ethnic groups may constitute an oversimplification due to the possibility that students’ perceptions of their own ethnic identities may be more fluid than the categories established by the researcher. However, a fairly rigid categorisation of students’ ethnic backgrounds was necessary in order to account for ethnic diversity among the sampled population during the statistical analysis of the data.
flexible or continuous measurement of ethnicity could not be incorporated in the analysis of the data. In total 367 students from six state secondary schools were sampled for this study. In 2007, 31,289 students were enrolled in 27 secondary schools in Luxembourg (Statec, 2008: 41).
Type of School L LD P Other
Classical 64 13 9 6
North
Technical 17 2 12 6
Classical 61 15 1 17
Centre
Technical 11 3 20 8
Classical 41 8 8 7
South
Technical 13 4 16 5
Table 3.1: Distribution of students sampled for the quantitative investigation of language attitudes.
L = Luxembourgish, LD = Luxembourgish dual nationality, P = Portuguese, Other = miscellaneous nationalities. N=367.
A sample of teachers working at the various schools was also asked to complete the attitude questionnaire. The researcher approached teachers in the staff room during breaks and free periods. All teachers took part on a voluntary basis. 43 teachers completed the questionnaire and the distribution of the teacher sample can be found in Table 3.2. The teachers’ questionnaire responses will not be analysed in the context of this thesis but will provide the basis for postdoctoral research.
Languages Other subjects
North 12 4
Centre 13 6
South 5 3
Total 30 13
Table 3.2: Distribution of sampled teachers according to the location of the school and subject specialisation. (N=43)
3.5.2 Sample for the behavioural study
The time-consuming nature of ethnographic research is at the origin of the selection of only one school for the collection of language behaviour data. Access to the school was negotiated with the headmaster prior to the start of fieldwork.
The project’s particular interest in the change of the language of instruction from German to French has resulted in the inclusion of a longitudinal aspect into the research design. The data collection is split into three separate four-week periods.
The initial data collection phase took place in May 2008 when the target group of students was at the end of their last year of German-medium education. The second data collection phase was carried out in October 2008 during the first month of French-medium education. The last data collection phase (May/June 2009) coincides with the end of the first year of French-medium education for the targeted students.
In collaboration with the headmaster a group of students, who at the time of the first data collection phase were enrolled in their last year of German-medium education, was selected. The group consists of 21 students who stay together as a class throughout the three data collection phases. These students spend their entire time at school together as a group. Classes of students frequently change in between different academic years due to administrative constraints and, therefore, particular attention was paid to gaining access to a group of students who would not be split into separate classes across the various fieldwork phases. The only group of students that were guaranteed to remain as a class between different academic years (i.e. fieldwork phases) were those who were enrolled in Latin education. A minority of students enrol in Latin classes in Luxembourg’s secondary schools as high academic achievements constitute a prerequisite for the enrolment in this type of education. Consequently, the students participating in the ethnographic study are academically stronger than average and must be viewed as overachievers in the Luxembourg education system. The sample consists of 8 girls and 13 boys. The majority of the students are Luxembourgish nationals. The other nationalities represented in the class are Serbian (2), Macedonian (1), Russian (1), Italian (1), German (1) and French (1).
Prior to the start of the first data collection phase the headmaster set up a meeting
the objectives of the research project and asked all teachers for their permission to carry out recordings in their lessons over a four-week period. Eight teachers (4 male, 4 female) gave their informed consent and access was granted to Biology, Geography, History, Religious Studies, English, Latin, German and French lessons. A similar meeting was held prior to the second data collection phase because informed consent had to be obtained from some new teachers. For the second and third phases of data collection access was granted to Geography, Biology, History, Physics, Maths, Art History, German and English classes. For History, Biology and English the teachers remain the same throughout the separate phases. For other school subjects teachers change between Phase A and Phase B.