Language skills
CHAPTER 3. Data analysis and results
73 respondents provided valid data in terms of their first and second languages. All participants reported Hungarian as their first language (language learnt from birth) and
82
English as their second language. For some reason 36,98 % (27) of the participants have not answered the question referring to the time spent in the country of their mother tongue or in a second language environment. As the response rate of the control group was low, comparing the two groups based on this question does not yield adequate results, hence only descriptive statistics can be given. According to the valid (46) answers, participants spent about 158 days (nearly 5 months) in the second language environment, while 4709 days (nearly 13 years) in the first language context. While the mean value for the first language covers balanced data, the average number of days spent in the target language environment is made up of extreme values. Many of the participants have not been abroad yet (minimum: 0,00), that is, some outliers (maximum: 2555) increased the mean value. The total numbers (sums) of days spent in the first and second language context reveal a remarkable difference. In case of the first language, it is 29,66 times the number of the days spent in the second language (216615-7302).
Participants’ estimation on the presence of the listed languages in their everyday lives revealed a remarkable difference (nearly 43%) between the presence of their first and second languages. Participants reported that they would be more likely to choose their first language ( 70,92 percent) over their second language ( 27,27 percent). In this case there were no extreme values, 0 percent minimum or 100 percent maximum, for either language, so the data set were compact in each case with balanced standard deviation (SDHun =14,025, SDEng=13,035) and variance (σ2 Hun=196,707, σ2
Eng=169,913).
3.1.4. Control group – L1 and L2 use
The following section reveals control group participants’ language preferences in different contexts. In case of reading or listening to something of their interest, Hungarian ( =64,6301 percent) and English ( =33,3288) are chosen at different rates, pointing to the fact that the second language is preferred half as much as their mother tongue. In this case there were some extreme values, 0 percent minimum or 100 percent maximum, however standard deviation (SDHun =27,47752, SDEng=26,59054) and variance (σ2Hun=755,014, σ2Eng=707,057) seem to be balanced.
In case of chatting with someone in the first or the second language the Hungarian language would be chosen at a higher rate ( =70,8630 percent) than the English
83
language ( =27,0822 percent). The values of variance (σ2Hun=536,648, σ2Eng=473,910), that of standard deviation (SDHun=23,16566, SDEng=21,76947) and the sums (Hun=5173,00, Eng=1977,00) reveal that some respondents produced extreme values, however it can be finally concluded that control group members seem to consciously choose the first language over the target language for activities like chatting.
In the following sections the results of the applied methods are analyzed and reported in case of the estimated level of language skills and activities being done in the first language. Regarding the self-estimated level of first language skills, the result of the Friedman-test (X2(4)=25701, p=,000) reveals that there is no overall statistically significant difference between the mean ranks, that is, participants are on the same opinion. Control participants considered writing (2,28) as the most and accent (3,29) as the least problematic area in their first language.
The necessary conditions for Principal Components Analysis are met with regard to the adequate KMO value (0,793) and Bartlett’s Test of Sphericity (p=0,000) meaning that there is correlation between the variables, hence the model is adequate for the analysis and it should yield reliable factors. Considering that, all the items had an upper value than 0,25, none of them were extracted confirming the strong correlation among all five items. Results reveal that while using their first language, control group learners regarded speaking, listening comprehension, reading, writing and accent as dominant and important skills, among which accent was considered as the least problematic.
Regarding the everyday activities being done in their first language, the result of the Friedman-test (X2(12)=73,971, p=,000) reveals no overall statistically significant difference between the mean ranks. The data show that first language use is the least characterized by writing blogs and texts (3,29), the use of skype (3,45), or playing online games with others (5,50). In contrast, the most frequent activities are internet use (8,62), watching films (8,74) and family chat (9,33)
The condition for Principal Components Analysis, adequate KMO value and significance level are both met. KMO (0,76) and the Bartlett’s Test of Sphericity (p=0,002) reveal that the data are adequate for the PCA analysis. Considering that, online games with others (h2=0,003), writing blogs and text (h2=0,174), internet use (h2=0,96) and listening to music (h2=0,005) showed the weakest correlation with all other variables, they were extracted one by one from the list of variables, then the two
84
tests were run again. This time the KMO value (0,661) and the result of the Bartlett’s Test (p=0,001), again, justified the adequacy of the variables for the analysis. Results show that while using their first language, control group members regarded watching films, chatting, learning, family chat, skype use, info search, reading and watching TV as the most dominant activities in their everyday lives in their mother tongue.
Regarding the self-estimated level of second language skills, the result of the Friedman-test (X2(4)=13,422 p=,009) reveals that there is no overall statistically significant difference between the mean ranks, that is, participants are on the same opinion. They considered speaking (2,75), accent (2,81) and listening comprehension (2,83) as the most, writing (3,13) and reading (3,49) as the least problematic areas in their second language. The KMO value (0,837) and the Bartlett’s Test of Sphericity (p=0,000) justify the adequacy of variables for the factor analysis. Considering that, the level of self-estimated accent had the lowest communality value (h2=0,95), referring to the weakest correlation with all other items) it was extracted from the list of variables, then the two tests were run again. This time the KMO value (0,836) and the result of the Bartlett’s Test (p=0,000) again, justified the adequacy of the variables for the analysis. The remaining four principal components (speaking, listening comprehension, reading and writing) are all considered to be present at a remarkable level in the participants’ lives.
Regarding the activities being done in their second language, the result of the Friedman-test (X2(12)=376,262 p=,000) reveals that there is no overall statistically significant difference between the mean ranks, that is, participants are on the same opinion.
Overall, the adequate KMO value (0,733) and the result of the Bartlett’s Test (p=0,000) justify the adequacy of variables for the factor analysis.
Considering that, watching TV (h2=0,233), chatting with family (h2=0,161), learning (h2=0,055), listening to music (h2=0,210) showed the weakest correlation with all other variables, they were extracted one by one from the list of variables, then the two tests were run again. This time the KMO value (0,794) and the result of the Bartlett’s Test (p=0,000), again, justified the adequacy of the variables for the analysis. Final results reveal that while using their second language, control group members regarded reading, watching films, internet use, chatting, online games, writing texts and blogs, skype use and info search as the most dominant activities in their everyday lives. Table 11 shows CLIL and control group participants’ self-assessed strengths, weaknesses, and dominant
85
activities in their mother tongue and the second language. Table 9 and 10 show data of language presence, preference and self-assessed skills in case of Hungarian (as an L1) and English (as an L2).
Table 9: Self-assessed language parameters
Table 10: Self-assessed level of L1 and L2 skills
CLIL CONTROL
First language Hungarian Hungarian
Second language English English
Days (24 hours) spent in the first language context 4550 4709 Days (24 hours) spent in the second language context 224 158 Difference of language use between L1 and L2 18.2%
(57.86-39.66)
43.65%
(70.92-27.27) Difference of listening-reading activities between L1 and
L2
5.95%
(51.45-45.5)
31.31%
(64.63-33.32)
Difference of talking in L1 and L2 17.76%
(57.76-40)
43.78%
(70.86-27.08)
Speaking (%) Listening (%) Reading (%) Writing (%)
L1 L2 L1 L2 L1 L2 L1 L2
CLIL 80.55 71.90 84.72 76.36 83.36 77.72 77.55 70.27 Control 86.63 59.18 87.18 59.90 85.72 65.81 80.82 62.54
86
Table 11 reveals the self-assessed strengths and weaknesses in the L1 and the L2 along with the different activities they are used for:
Order of
Strengths listening, (accent) accent, (listening) Weaknesses writing, speaking, reading writing, reading, speaking
The most Rare online games, writing texts,
(Skype use)
Strengths reading, listening reading, writing Weaknesses writing, accent, speaking speaking, accent, listening
The most
Table 11: Summary table on L1 and L2 use
87 3.2. Test d2-R – test of selective attention
Data was processed by SPSS software. 142 learners produced evaluable data of whom 69 were CLIL learners and 73 were controls. There were no incomplete, unidentifiable or invaluable test sheets. Data were recorded block by block then summed for further analysis. All the 798 signs of the d2-R test are arranged in 4 blocks (three rows constitute a block that is repeated four times). Figure 11 shows how participants’ mean performance values altered during the four blocks.
Figure 11: Mean performance values of concentration
In the first block CLIL learners processed M=38 target stimuli and made M=3,36 omission errors and M=0,26 incorrect markings. In the second block they processed M=38,34 target stimuli and made M=3,75 omission errors and M=0,27 incorrect markings. In the third block they processed M=34,85 target stimuli and made M=3,23 omission errors and M=0,23 incorrect markings. Finally, in the fourth block they processed M=35,27 target stimuli and made M=4,56 omission errors and M=0,52 incorrect markings.
The control group participants processed M=38,12 target stimuli and made M=4,43 omission errors and M=0,63 incorrect markings in the first block. In the second block they processed M=38,41 target stimuli and made M=4,8 omission errors and M=0,61 incorrect markings. In the third block they processed M=36,13 target stimuli and made M=4,54 omission errors and M=0,54 incorrect markings. Finally, in the fourth block they processed M=36,68 target stimuli and made M=4,8 omission errors and M=0,64 incorrect markings.
25 30 35
Block 1 Block 2 Block 3 Block 4