4.2. The relationships between the participants’ formulaic performance and
4.3.1. The relationship between the participants’ FS errors and their learning
The participants’ FS errors were examined in relation to their English learning background, perceptions of FSs, and their perceptions of their own acquisition and use of FSs. The Eta correlation ratio (ƞ) was calculated for all the variables making up the participants’ English learning background, perceptions of FSs, and perceptions of their acquisition and use of FSs. These include Starting Age (V1), Learning Context (V2), Sources of Exposure (V3), Importance Awareness (V4), Benefit Awareness (V5), Noticing (V6), Active Study (V7),
81
Study Techniques (V8), Active Use in Writing (V9), Usage Strategies (V10), and Learning Difficulties (V11). The test was conducted at the confidence level of α = .01 for N = 41. According to the SAGE Research Methods, the Eta correlation ratio could be interpreted using Pearson’s correlation coefficient scale (Scott-Jones, 2019) which is summarized in Table 4.2.
Table 4.2. Eta correlation ratio values and interpretations Eta values Interpretation
0.00 No correlation
0.01-0.19 No or negligible correlation 0.2-0.39 Weak correlation 0.4-0.69 Medium correlation
0.7-1.0 Strong correlation
Source: SAGE Research Methods (Scott-Jones, 2019, p.6) It is worth mentioning that the correlation indicated by the Eta correlation ratio is nonlinear in nature. Thus, the Eta correlation ratio does not suggest the direction of the correlation. However, based on the data, it is possible to make an informed estimation as to which level of the independent variable has the most impact on the dependent variable by calculating and comparing the average values of each level of the independent variable (Scott-Jones, 2019). Since the impact of the variables with a weak correlation to the participants’ FS errors would likely be insignificant, estimations were made for only correlations with medium strength.
Additionally, the explained variance was also calculated for correlations with weak to medium strength to show how much of the variability in the quantity of FS errors could be explained by the learning factor variables. The explained variance value is the squared correlation ratio, ƞ2 (Snijders & Bosker, 2011, p.99).
Table 4.3 presents the Eta values calculated for participants’ FS errors and the variables mentioned above (refer to Appendix F for the actual SPSS output tables).
82
Table 4.3. Eta values of FS errors and learning factor variables Variable Learning factors Eta value (ƞ)
V1 Starting Age 0.141
V2 Learning Context 0.467
V3 Sources of Exposure 0.291 V4 Importance Awareness Not available
V5 Benefit Awareness 0.231
V6 Noticing 0.017
V7 Active Study 0.138
V8 Study Techniques 0.437
V9 Active Use in Writing 0.579
V10 Usage Strategies 0.066
V11 Learning Difficulties 0.269
First, several learning factors were found to have no correlation or only a negligible correlation with the participants’ FS errors. Those variables include Starting Age (ƞ = 0.141), Noticing (ƞ = 0.017), Active Study (ƞ = 0.138), and Usage Strategies (ƞ = 0.066). Another variable that became irrelevant to the participants’ formulaic performance in the test was Importance Awareness. When responding to the question regarding Importance Awareness in the questionnaire, all of the participants agreed that FSs played an important role in learning and using English. Since all responses were the same, the variable Importance Awareness became a constant, thus, it was not possible to calculate a correlation ratio between FS errors and this variable.
In terms of the starting age, despite starting their learning at slightly different ages (before 5 years old, between 5 and 10 years old, or after 10 years old), the participants all started learning during their childhood due to English being a compulsory subject in public schools. The drastic variation in FS acquisition and performance as reported by Schmidt (1983) and Yorio (1989) would probably be more prominent in samples with a wider age range. As for the other variables mentioned above, there was not enough information to explain why they did not have an influence on the number of FS errors made by the participants in the test.
Second, there is a weak correlation between the participants’ FS errors and their sources of exposure to English (ƞ = 0.291), awareness of the benefits of FSs
83
(ƞ = 0.231), and the learning difficulties they experienced with FSs (ƞ = 0.269). For the variable Sources of Exposure, the explained variance, ƞ2 = 0.084, indicates that only around 8% of the difference in FS errors could be attributed to sources of exposure to English. As reported in the questionnaire responses, besides their formal English classes, the participants got exposed to English from a variety of sources. However, these sources did not seem to have a major influence on the FS errors they made. Meanwhile, the explained variance of the variable Benefit Awareness is ƞ2 = 0.053, which means only around 5% of the difference in the participants’ FS errors could be attributed to Benefit Awareness. Even though the participants had positive perceptions of FSs – they realized the benefits of FSs – such perceptions did not significantly affect the number of FS errors they made.
The same could be said about the variable Learning Difficulties. The explained variance, ƞ2 = 0.072, indicates that only around 7.2% of the difference in the participants’ FS errors could be attributed to their learning difficulties.
Third, a medium correlation was found between the participants’ FS errors and Learning Contexts (ƞ = 0.467), Study Techniques (ƞ = 0.437), and Active Use in Writing (ƞ = 0.579). The explained variance for the variable Learning Contexts, ƞ2 = 0.22, indicates that around 22% of the variance in the number of FS errors the participants made could be explained by the differences in their learning contexts.
Table 4.4 is the SPSS output table showing the actual numbers of FS errors committed in each level of the variable Learning Context. There are five levels in this variable, including (1) studying at school, (2) at English centers, (3) at school and English centers and study groups, (4) at school and self-study, (5) at school and self-study and at English centers. The error count column shows the number of FS errors while the column for each level shows the number of participants making the corresponding number of errors. For example, in the group of participants who only studied at school, there was one participant who made 6 errors, one who made 9 errors, two who made 13 errors, one who made 15 errors, one who made 19 errors, and one who made 22 errors. Therefore, the average number of errors made by the participants in this group was calculated to be 13.85, avg = 13.85.
84
Table 4.4. FS errors by levels of Learning Context
Crosstab Count
2. Learning Context Total
School Center School Center Group
School
&
Self-Study
School
& Self Study
& Center
Error Count 5 0 0 0 1 0 1
6 1 0 1 1 0 3
7 0 0 0 1 0 1
8 0 0 0 4 2 6
9 1 0 1 1 2 5
10 0 0 0 1 2 3
11 0 0 2 3 0 5
12 0 0 0 2 2 4
13 2 0 0 1 1 4
14 0 1 0 1 0 2
15 1 0 0 1 0 2
16 0 0 0 0 1 1
18 0 1 0 1 0 2
19 1 0 0 0 0 1
22 1 0 0 0 0 1
Total 7 2 4 18 10 41
Average (avg) 13.85 16.00 9.25 10.33 10.70
In particular, the participants who reported only studying English in English centers were found to make the most FS errors with an average of 16 errors for each participant (avg = 16.00), while those who studied at school and self-studied at home made fewer FS errors with an average of 10.33 errors (avg = 10.33), followed by those who studied at school and at English centers as well as self- studied at home with an average of 10.7 errors (avg = 10.70). Participants who studied at school, in English centers, and in study groups made the fewest FS errors with an average of 9.25 errors (avg = 9.25). It appears that those who studied English in different types of settings (formal, semi-formal, and informal) seem to have better formulaic performance in the test than those who study in one type of learning context. It is possible that their own motivation and learning autonomy might have also played a role as the participants who claimed to self-study at home in addition to formal learning seems to have good formulaic performance as well.
85
When it comes to Study Techniques (ƞ = 0.437), the explained variance is ƞ2 = 0.19, indicating that around 19% of the difference in the participants’ FS errors could be attributed to their study techniques.
Table 4.5 is the SPSS output table showing the actual numbers of errors in each level of the variable Study Techniques. Similar to Table 4.4 presented earlier, Table 4.5 displays the six levels of the variable Study Techniques with the corresponding number of FS errors. For example, in the group of participants who reported using only dictionaries to study FSs, there was one participant who made 6 errors, one with 8 errors, one with 9 errors, two with 11 errors, one with 15 errors, and one with 16 errors. The average number of errors committed by the participants in this group was calculated to be 10.58, avg = 10.58.
86
Table 4.5. FS errors by levels of Study Techniques
Crosstab
Count
8. Study Techniques
Total Dictionaries Google
Dictionaries + Google
Dictionaries + Google + Take Note
Google +
Take Note Take Note
Error Count 5 0 0 0 0 0 1 1
6 1 2 0 0 0 0 3
7 0 1 0 0 0 0 1
8 1 1 2 1 0 0 5
9 1 0 0 1 0 0 2
10 0 3 0 0 0 0 3
11 2 0 0 1 0 0 3
12 0 1 1 0 2 0 4
13 0 3 1 0 0 0 4
14 0 1 1 0 0 0 2
15 1 0 0 0 1 0 2
16 1 0 0 0 0 0 1
18 0 0 1 0 0 1 2
19 0 0 1 0 0 0 1
22 0 0 1 0 0 0 1
Total 7 12 8 3 3 2 35
Average (avg) 10.58 10.16 14.25 9.33 13 11.5
Interestingly, the participants who reported using both dictionaries and Google to look up meanings of FSs made the most FS errors with an average of 14.25 errors per participant (avg = 14.25), more than those who only used either one (avg = 10.58 for dictionaries and avg = 10.16 for Google). The participants who made the fewest FS errors are those who reported using both dictionaries and Google to look up FS meanings, then writing down the meanings to review later with an average of 9.33 errors per participant (avg = 9.33). Taking notes of FSs for
87
review might have helped them retain FSs better than those who only looked up their meanings. Indeed, the act of taking notes has been proven to be beneficial as an encoding process that aids information retention (Knight & McKelvie, 1986;
Sevenants, Daniởls, Janssens, & Schaeken, 2012; Wu et al., 2021). Some believe that handwritten notes are better for memorizing information (Mueller &
Oppenheimer, 2014). Nevertheless, growing evidence suggests that even digital notes taken on computers are helpful (Bui, Myerson, & Hale, 2013; Kalnikaité &
Whittaker, 2008). The general consensus is that to reach the optimal memory retention effect, notes should be concise and well-organized (Friedman, 2014;
Kalnikaitŏ & Whittaker, 2007).
With regard to the variable Active Use in Writing (ƞ = 0.579), the explained variance, ƞ2 = 0.335, indicates that around 33.5% of the difference in the participants’ FS errors could be attributed to their active efforts (or the lack thereof) to use FSs in writing.
Table 4.6 is the SPSS output table showing the number of FS errors in each of the two levels of the variable Active Use in Writing. The left column presents the actual number of errors while the next two columns show the number of participants. For example, in the group who answered “No” to the question of whether they would actively make an effort to incorporate FSs in their writing, there were a total of seven participants. Among those, there was one who made 12 errors, one with 13 errors, two with 15 errors, one with 16 errors, one with 19 errors, and one with 22 errors. Thus, the average number of errors committed by the participants in this group was calculated to be 14.57, avg = 14.57.
88
Table 4.6. FS errors by levels of Active Use in Writing
Crosstab Count
9. Active Use
Total
Yes No
Error Count 5 1 0 1
6 3 0 3
7 1 0 1
8 6 0 6
9 5 0 5
10 3 0 3
11 5 0 5
12 3 1 4
13 3 1 4
14 2 0 2
15 0 2 2
16 0 1 1
18 2 0 2
19 0 1 1
22 0 1 1
Total 34 7 41
Average (avg) 10.20 14.57
On average, the participants who stated that they would actively make an effort to incorporate FSs in their written production made fewer FS errors (avg = 10.20) than those who would not do so (avg = 14.57). Regardless of the strategies employed, actively making efforts to use FSs during written production seems to have a positive influence on the participants’ formulaic performance during the test.
So far, this section has discussed the potential relationships between the participants’ FS errors and major learning factors. Table 4.7 summarizes the results of the correlation test conducted on the participants’ FS errors and the variables constituting their English learning background, perceptions of FSs, and perceptions of their acquisition and use of FSs.
89