grammatical structure.Finally, students were asked to write on the same topic provided for the explicit group, but the feedback was not given directly. The researcher wrote the number of errors the learners had made using the target structures. The learners were then encouraged to compare their own use of the grammar in the paragraph they wrote to its use in the bolded text in order to correct their errors. Needless to state that the application of these implicit techniques differed in CALL-only, face-to-face and blended groups as it was the case with the explicit manner of the instruction. For the implicit CALL-only instruction of grammar, Viber application was used as the medium. If extra information or additional files were required the links were provided via Viber.
Other additional activities were applied in the course of treatment as well: In the implicit instruction of grammar, the emphasis was more on the message rather than the abstract rule while for the explicit groups it was the other way round. In the implicit instruction of grammar, little or no grammar meta-language is used. In the explicit mode, some metalinguistic questions were asked about sentences and participants were supposed to identify rules in the given sentences.
For implicit grammar instruction, before introducing the grammar point, teachers built schema and showed this grammar in use rather than explaining it. In the implicit groups, teachers used students’ lives, current events and the like which were interesting to learners. Exemplification was a crucial point for implicit grammar instruction since learners were supposed to be provided with rich context to deduct rules based on lots of examples. In the explicit grammar instruction, grammatical points were explained in details and students were required to practice them in the exercises, whereas in the implicit instruction, teachers asked lots of questions related to using the intended grammatical points while they avoided giving answers unless it was necessary and no students could answer or figure it out. In the implicit grammar instruction, students were provided rich context to discover rules themselves rather than being provided with ready-made solutions. In other words, implicit grammar instruction was considered a discovery learning and suggested discovery results in long term acquisition. In this regard, students could grasp the intended grammatical points in the form of self-correction or experiencing failure in the communication of the message and consequent correction to overcome the problem. Since schema building and comprehensibility of the grammatical input were important in the implicit instruction, learners were involved in negotiating ideas about real information, hobbies, likes and dislikes, experiences, food preferences, sports they played, the real needs they had (shopping, going to the doctor, finding a job) and other real and relevant life experiences.
Activities such as word scramble, diagrams with words, cooperative writing, editing activities, cloze exercises, error correction, surveys of all kinds about the natural world, out of class interviews and observations, sensory description writing assignments, writing/speaking about nature, role-plays, info gap games, class surveys, group games, group writing tasks (posters, FAQs, op-ed letters), targeted creative writing, individual projects (family tree, auto-biography), and journaling were among other activities practiced in the implicit grammar instruction.
Post-treatment
Having practiced ten sessions of treatment, the participants in all groups took the teacher-made test of grammatical accuracy as the posttest. The researcher intended to measure learners' likely improvement as a result of treatment by comparing their pretest and posttest scores.
4. Results
After the data collected through the administration of the instruments, the data were analyzed using SPSS software version 24 and measures ofindependent t-test,one-way ANOVA, two-way
Vol. 6, Issue 9, December 2016 Page 101
ANOVA, and post-hoc Scheffe’s tests. As all these measures have a main assumption; i.e. normality. The normality of the present data was probed through the Kolmogorov-Smirnov (KS) and skewness and kurtosis ratios and in both cases the normality of the data was confirmed. NELSON General Language Proficiency TestThe NELSON test was administered to 153 cases. Based on the mean (M = 57.65) plus and minus one standard deviation (SD = 14.35), 120 cases were selected to participate in this study. The KR- 21 reliability index for the NELSON test was .89 (Table 1).
Table1: Descriptive Statistics; NELSON Test
N Mean Std. Deviation Variance KR-21
NELSON 153 57.65 14.357 206.111 .89
Pretest of Grammatical Accuracy
A two-way ANOVA was run to compare the explicit and implicit CALL only, face-to-face and blended groups’ means on the pretest of grammatical accuracy in order to prove that they were homogenous in terms of their grammatical accuracy prior to the main study. Before discussing the two-way ANOVA results it should be mentioned that the groups enjoyed homogenous variances on the pretest of grammatical accuracy. As displayed in Table 2 the Levene’s F-value of 1.77 was not significant (P > .05).
Table 2: Levene's Test of Equality of Error Variances
F df1 df2 Sig.
1.777 5 114 .123
As displayed in Table 3 the CALL only (M = 27.50, SE = 1.10), face-o-face (M = 26, SE = 1.10) and blended methods (M = 26.55, SE = 1.10) showed almost the same means on pretest of grammatical accuracy.
Table 3: Descriptive Statistics; Pretest of Grammatical Accuracy by Groups
Group Mean Std. Error
95% Confidence Interval
Lower Bound Upper Bound
CALL Only 27.500 1.106 25.309 29.691
Face-to-Face 26.000 1.106 23.809 28.191
Blended 26.550 1.106 24.359 28.741
Based on the results displayed in Table 4 (F (2, 114) = .47, P > .05; η2 = .008 representing a weak
effect size) it can be concluded that there was not any significant difference between CALL only, face-to-face and blended groups’ means on the pretest of grammatical accuracy. Thus, it was claimed that they were homogenous in terms of their grammatical accuracy prior to the main study.
Table 4: Tests of Between-Subjects Effects, Pretest of Grammatical Accuracy by Groups by Types of Instruction
Source Type III Sum of Squares Df Mean Square F Sig. Partial Squared Eta
Group 46.067 2 23.033 .471 .626 .008
Instruction 73.633 1 73.633 1.505 .222 .013
Group * Instruction 26.867 2 13.433 .275 .760 .005
Error 5577.400 114 48.925
Total 91164.000 120