The present study investigates the effect of collaborative, individual and e-collaborative methods on the development and sustained development of lexical and grammatical accuracy among Iranian EFL learners.
4.1. The effect of individual writing on lexical and grammatical accuracy.
MANOVA was run to compare the individual group’s means on the pretests, posttests and delayed posttests of lexical and grammatical accuracy. The MANOVA was followed by post-hoc comparison tests to compare the means across three time intervals of pretest to posttest and delayed posttest. The results of the MANOVA (F (5, 38) = 16.06, p = .000, partial η2 = .679 representing a large effect size) (Table 2) indicated that there were significant differences between the individual group’s means on the pretests, posttests and delayed posttests of lexical and grammatical accuracy.
Table 2
Multivariate Tests; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (Individual Writing Group)
Effect
|
Value
|
F
|
Hypothesis df
|
Error df
|
Sig.
|
Partial Eta Squared
|
Pillai's Trace
|
.679
|
16.069
|
5
|
38
|
.000
|
.679
|
Wilks' Lambda
|
.321
|
16.069
|
5
|
38
|
.000
|
.679
|
Hotelling's Trace
|
2.114
|
16.069
|
5
|
38
|
.000
|
.679
|
Roy's Largest Root
|
2.114
|
16.069
|
5
|
38
|
.000
|
.679
|
As displayed in Table 3, the individual group showed an increase in their means from pretest of grammatical accuracy (M = 36.76) to posttest (M = 49.34); however, their mean score decreased to 47.25 on the delayed posttest. The same pattern can be seen on the lexical accuracy. The means on the pretest, posttest and delayed posttest were 40.95, 47.09 and 45.07 respectively.
Table 3
Descriptive Statistics; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (Individual Writing Group)
Tests
|
Mean
|
Std. Error
|
95% Confidence Interval
|
Lower Bound
|
Upper Bound
|
Pre-grammatical
|
36.767
|
.955
|
34.840
|
38.695
|
Pre-lexical
|
40.953
|
1.079
|
38.775
|
43.132
|
Post-grammatical
|
49.349
|
1.118
|
47.092
|
51.605
|
Post-lexical
|
47.093
|
1.056
|
44.962
|
49.224
|
Delayed-grammatical
|
47.256
|
1.076
|
45.085
|
49.427
|
Delayed-lexical
|
45.070
|
.987
|
43.078
|
47.061
|
Based on the results displayed in Table 4 it can be concluded that a) the individual group had a significant improvement in their mean on grammatical accuracy from pretest (M = 36.76) to posttest (M = 49.34) (MD = -12.58, p = .000). They also sustained their significant difference from pretest to delayed posttest (M = 47.25) (MD = -10.48, p = .000), b) the individual group had a significant improvement in their mean on lexical accuracy from pretest (M = 40.95) to posttest (M = 47.09) (MD = -6.14, p = .000). They also sustained their significant difference from pretest to delayed posttest (M = 45.07) (MD = -4.11, p = .005). Therefore, it can be concluded that there was a statistically significant difference in lexical and grammatical accuracy of individual writing from pretest to posttest measures, and the difference was sustained.
Table 4
Pairwise Comparisons; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (Individual Writing Group)
(I) Tests
|
(J) Tests
|
Mean
Difference (I-J)
|
Std. Error
|
Sig.
|
95% Confidence
Interval for Difference
|
Lower
Bound
|
Upper
Bound
|
Pre-grammatical
|
Post-grammatical
|
-12.581*
|
1.443
|
.000
|
-15.493
|
-9.669
|
Delayed-grammatical
|
-10.488*
|
1.434
|
.000
|
-13.382
|
-7.595
|
Pre-Lexical
|
Post-Lexical
|
-6.140*
|
1.506
|
.000
|
-9.179
|
-3.100
|
Delayed-Lexical
|
-4.116*
|
1.397
|
.005
|
-6.935
|
-1.298
|
*. The mean difference is significant at the .05 level.
|
4. 2 The effect of collaborative writing on lexical and grammatical accuracy.
The MANOVA followed by post-hoc comparison tests were run to compare the collaborative group’s means on the pretests, posttests and delayed posttests of lexical and grammatical accuracy in order to probe the second null-hypothesis. The results of the MANOVA (F (5, 38) = 17.64, p = .000, partial η2 = .699 representing a large effect size) (Table 5) indicated that there were significant differences between the collaborative group’s means on the pretests, posttests and delayed posttests of lexical and grammatical accuracy.
Table 5
Multivariate Tests; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (Collaborative Writing Group)
Effect
|
Value
|
F
|
Hypothesis df
|
Error df
|
Sig.
|
Partial Eta Squared
|
Pillai's Trace
|
.699
|
17.642
|
5
|
38
|
.000
|
.699
|
Wilks' Lambda
|
.301
|
17.642
|
5
|
38
|
.000
|
.699
|
Hotelling's Trace
|
2.321
|
17.642
|
5
|
38
|
.000
|
.699
|
Roy's Largest Root
|
2.321
|
17.642
|
5
|
38
|
.000
|
.699
|
As displayed in Table 6, the collaborative group showed an increase in their means from pretest of grammatical accuracy (M = 34.58) to posttest (M = 50.14); however, their mean score decreased to 48.97 on the delayed posttest. The same pattern can be seen on the lexical accuracy. The means on the pretest, posttest and delayed posttest were 40.83, 48 and 47.07 respectively.
Table 6
Descriptive Statistics; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (Collaborative Writing Group)
Tests
|
Mean
|
Std. Error
|
95% Confidence Interval
|
Lower Bound
|
Upper Bound
|
Pre-grammatical
|
34.581
|
.887
|
32.790
|
36.372
|
Pre-lexical
|
40.837
|
1.065
|
38.687
|
42.987
|
Post-grammatical
|
50.140
|
1.307
|
47.503
|
52.776
|
Post-lexical
|
48.000
|
1.250
|
45.477
|
50.523
|
Delayed-grammatical
|
48.977
|
1.285
|
46.383
|
51.570
|
Delayed-lexical
|
47.070
|
1.235
|
44.578
|
49.561
|
Based on the results displayed in Table 7 it can be concluded that a) the collaborative group had a significant improvement in their mean on grammatical accuracy from pretest (M = 34.58) to posttest (M = 50.14) (MD = -15.55, p = .000). They also sustained their significant difference from pretest to delayed posttest (M = 48.97) (MD = -14.39, p = .000), b) the collaborative group had a significant improvement in their mean on lexical accuracy from pretest (M = 40.83) to posttest (M = 48) (MD = -7.16, p = .000). They also sustained their significant difference from pretest to delayed posttest (M = 47.07) (MD = -6.23, p = .000). It can be concluded that there was a statistically significant difference in lexical and grammatical accuracy of collaborative writing from pretest to posttest measures, and the difference was sustained.
Table 7
Pairwise Comparisons; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (Collaborative Writing Group)
(I) Tests
|
(J) Tests
|
Mean
Difference (I-J)
|
Std. Error
|
Sig.
|
95% Confidence
Interval for Difference
|
Lower
Bound
|
Upper
Bound
|
Pre-grammatical
|
Post-grammatical
|
-15.558*
|
1.646
|
.000
|
-18.880
|
-12.236
|
Delayed-grammatical
|
-14.395*
|
1.745
|
.000
|
-17.918
|
-10.873
|
Pre-Lexical
|
Post-Lexical
|
-7.163*
|
1.616
|
.000
|
-10.425
|
-3.901
|
Delayed-Lexical
|
-6.233*
|
1.429
|
.000
|
-9.116
|
-3.349
|
*. The mean difference is significant at the .05 level.
|
4.3 The effect of e-collaborative group on lexical and grammatical accuracy.
The MANOVA followed by post-hoc comparison tests were run to compare the e-collaborative group’s means on the pretests, posttests and delayed posttests of lexical and grammatical accuracy in order to probe the third null-hypothesis. The results of the MANOVA (F (5, 38) = 8.64, p = .000, partial η2 = .532 representing a large effect size) (Table 8) indicated that there were significant differences between the e-collaborative group’s means on the pretests, posttests and delayed posttests of lexical and grammatical accuracy.
Table 8
Multivariate Tests; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (E-collaborative Writing Group)
Effect
|
Value
|
F
|
Hypothesis df
|
Error df
|
Sig.
|
Partial Eta Squared
|
Pillai's Trace
|
.532
|
8.644
|
5
|
38
|
.000
|
.532
|
Wilks' Lambda
|
.468
|
8.644
|
5
|
38
|
.000
|
.532
|
Hotelling's Trace
|
1.137
|
8.644
|
5
|
38
|
.000
|
.532
|
Roy's Largest Root
|
1.137
|
8.644
|
5
|
38
|
.000
|
.532
|
As displayed in Table 9, the e-collaborative group showed an increase in their means from pretest of grammatical accuracy (M = 36.74) to posttest (M = 45.83); however, their mean score decreased to 43.80 on the delayed posttest. The same pattern can be seen on the lexical accuracy. The means on the pretest, posttest and delayed posttest were 41.27, 43.95 and 41.88 respectively.
Table 9
Descriptive Statistics; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (E-collaborative Writing Group)
Tests
|
Mean
|
Std. Error
|
95% Confidence Interval
|
Lower Bound
|
Upper Bound
|
Pre-grammatical
|
36.744
|
.883
|
34.962
|
38.526
|
Pre-lexical
|
41.279
|
1.068
|
39.124
|
43.434
|
Post-grammatical
|
45.837
|
1.434
|
42.944
|
48.730
|
Post-lexical
|
43.953
|
1.391
|
41.146
|
46.761
|
Delayed-grammatical
|
43.814
|
1.356
|
41.077
|
46.551
|
Delayed-lexical
|
41.884
|
1.320
|
39.221
|
44.547
|
Based on the results displayed in Table 10 it can be concluded that a) the e-collaborative group had a significant improvement in their mean on grammatical accuracy from pretest (M = 36.74) to posttest (M = 45.83) (MD = -9.09, p = .000). They also sustained their significant difference from pretest to delayed posttest (M = 43.81) (MD = -7.07, p = .000), b) the e-collaborative group did not have any significant improvement in their mean on lexical accuracy from pretest (M = 41.27) to posttest (M = 43.95) (MD = -2.67, p = .175). No significant difference was sustained from pretest to delayed posttest (M = 41.88) (MD = − .605, p = .747). There was a statistically significant difference in grammatical accuracy of e-collaborative writing from pretest to posttest measures, and the difference was sustained; however they did not show any significant improvement not sustenance in their means on lexical accuracy.
Table 10
Pairwise Comparisons; Pretests, Posttests and Delayed Posttests of Lexical and Grammatical Accuracy (E-collaborative Writing Group)
(I) Tests
|
(J) Tests
|
Mean
Difference (I-J)
|
Std. Error
|
Sig.
|
95% Confidence
Interval for Difference
|
Lower
Bound
|
Upper
Bound
|
Pre-grammatical
|
Post-grammatical
|
-9.093*
|
1.738
|
.000
|
-12.601
|
-5.585
|
Delayed-grammatical
|
-7.070*
|
1.705
|
.000
|
-10.511
|
-3.628
|
Pre-Lexical
|
Post-Lexical
|
-2.674
|
1.938
|
.175
|
-6.586
|
1.237
|
Delayed-Lexical
|
− .605
|
1.860
|
.747
|
-4.359
|
3.150
|
*. The mean difference is significant at the .05 level.
|
4.4 The effect of modality on accuracy.
The fourth research question was broken into two parts to compare the three groups’ means on the posttest and delayed posttest of grammatical accuracy first; and then on the posttest and delayed posttest of lexical accuracy.
4.4.1 The effect of writing modality on grammatical accuracy.
The MANOVA was run to compare the three groups’ means on the pretest, posttest and delayed posttest of grammatical accuracy. Before discussing the results it should be noted that the assumption of homogeneity of variance and covariance matrices were met. As displayed in Table 11, the probabilities associated with the pretest, posttest and delayed posttest of grammatical accuracy were not significant (p > .05).
Table 11
Levene's Test of Equality of Error Variances; Grammatical Accuracy
|
F
|
df1
|
df2
|
Sig.
|
Pretest
|
.018
|
2
|
126
|
.982
|
Posttest
|
1.273
|
2
|
126
|
.283
|
Delayed Posttest
|
1.073
|
2
|
126
|
.345
|
The assumption of homogeneity of covariance matrices; i.e. correlations between any two dependent variables should be roughly equal, was also met (Box’s M = 7.36, p = .850) (Table 12).
Table 12
Box’s Test of Equality of Covariance Matrices
Box's M
|
7.368
|
F
|
.593
|
df1
|
12
|
df2
|
76937.538
|
Sig.
|
.850
|
The results of the MANOVA (Table 12) indicated that a)There were not any significant differences between the three groups’ means on the pretest of grammatical accuracy (F (2, 126) = 1.90, p = .153, partial η2 = .029 representing a weak effect size). Thus, it can be claimed that they were homogenous in terms of their knowledge on grammatical accuracy prior to the main study, b) there were significant differences between the three groups’ means on the posttest of grammatical accuracy (F (2, 126) = 3.13, p = .047, partial η2 = .047 representing an almost moderate effect size), c)There were significant differences between the three groups’ means on the delayed posttest of grammatical accuracy (F (2, 126) = 4.46, p = .013, partial η2 = .066 representing a moderate effect size).
Table 13
Tests of Between-Subjects Effects; Pretest, Posttest and Delayed Posttest of Grammatical Accuracy by Groups
Source
|
Dependent Variable
|
Type III Sum of Squares
|
df
|
Mean Square
|
F
|
Sig.
|
Partial Eta Squared
|
Group
|
Pretest
|
135.550
|
2
|
67.775
|
1.907
|
.153
|
.029
|
Posttest
|
451.023
|
2
|
225.512
|
3.139
|
.047
|
.047
|
Delayed Posttest
|
594.295
|
2
|
297.147
|
4.460
|
.013
|
.066
|
Error
|
Pretest
|
4478.326
|
126
|
35.542
|
|
|
|
Posttest
|
9052.791
|
126
|
71.848
|
|
|
|
Delayed Posttest
|
8395.674
|
126
|
66.632
|
|
|
|
Total
|
Pretest
|
172086.00
|
129
|
|
|
|
|
Posttest
|
312217.00
|
129
|
|
|
|
|
Delayed Posttest
|
290110.00
|
129
|
|
|
|
|
Table 14 displays the means on the three groups on the pretest, posttest and delayed posttest of grammatical accuracy.
Table 14
Descriptive Statistics; Pretest, Posttest and Delayed Posttest of grammatical Accuracy by Groups
Dependent Variable
|
group
|
Mean
|
Std. Error
|
95% Confidence Interval
|
Lower Bound
|
Upper Bound
|
Pretest
|
collaborative
|
34.581
|
.909
|
32.782
|
36.381
|
individual
|
36.767
|
.909
|
34.968
|
38.567
|
e-collaborative
|
36.744
|
.909
|
34.945
|
38.543
|
Posttest
|
collaborative
|
50.140
|
1.293
|
47.581
|
52.698
|
individual
|
49.349
|
1.293
|
46.791
|
51.907
|
e-collaborative
|
45.837
|
1.293
|
43.279
|
48.395
|
Delayed Posttest
|
collaborative
|
48.977
|
1.245
|
46.513
|
51.440
|
individual
|
47.256
|
1.245
|
44.792
|
49.719
|
e-collaborative
|
43.814
|
1.245
|
41.350
|
46.277
|
The results of the post-hoc comparison tests (Table 15) indicated that a) there was not any significant difference between the collaborative (M = 50.14) and individual (M = 49.34) groups’ means on the posttest of grammatical accuracy (MD = .791, p = .666), b) there was not any significant difference between the individual (M = 49.34) and e-collaborative (M = 45.83) groups’ means on the posttest of grammatical accuracy (MD = 3.51, p = .057), b) the collaborative group (M = 50.14) significantly outperformed the e-collaborative group (M = 45.83) on the posttest of grammatical accuracy (MD = 4.30, p = .020), d) there was not any significant difference between the collaborative (M = 48.97) and individual (M = 47.25) groups’ means on the delayed posttest of grammatical accuracy (MD = 1.72, p = .303), e) There was not any significant difference between the individual (M = 47.25) and e-collaborative (M = 43.81) groups’ means on the delayed posttest of grammatical accuracy (MD = 3.44, p = .053), f) the collaborative group (M = 48.97) significantly outperformed the e-collaborative group (M = 43.81) on the delayed posttest of grammatical accuracy (MD = 5.16, p = .004).
Table 15
Post-Hoc Comparisons; Posttest and Delayed Posttest of grammatical Accuracy by Groups
Dependent Variable
|
(I) group
|
(J) group
|
Mean Difference (I-J)
|
Std. Error
|
Sig.
|
95% Confidence Interval for Difference
|
Lower Bound
|
Upper Bound
|
Posttest
|
Collaborative
|
Individual
|
.791
|
1.828
|
.666
|
-2.827
|
4.408
|
e-collaborative
|
4.302*
|
1.828
|
.020
|
.685
|
7.920
|
Individual
|
e-collaborative
|
3.512
|
1.828
|
.057
|
− .106
|
7.129
|
Delayed
Posttest
|
Collaborative
|
Individual
|
1.721
|
1.760
|
.330
|
-1.763
|
5.205
|
e-collaborative
|
5.163*
|
1.760
|
.004
|
1.679
|
8.647
|
Individual
|
e-collaborative
|
3.442
|
1.760
|
.053
|
− .042
|
6.926
|
*. The mean difference is significant at the .05 level.
|
4.4.2 The effect of writing modality on lexical accuracy.
The MANOVA was run to compare the three groups’ means on the pretest, posttest and delayed posttest of lexical accuracy. Before discussing the results it should be noted that the assumption of homogeneity of variance and covariance matrices were met. As displayed in Table 16, the probabilities associated with the pretest, posttest and delayed posttest of lexical accuracy, were not significant (p > .05).
Table 16
Levene's Test of Equality of Error Variances; Lexical Accuracy
|
F
|
df1
|
df2
|
Sig.
|
Pretest
|
.000
|
2
|
126
|
1.000
|
Posttest
|
2.801
|
2
|
126
|
.065
|
Delayed Posttest
|
1.656
|
2
|
126
|
.195
|
The assumption of homogeneity of covariance matrices; i.e. correlations between any two dependent variables should be roughly equal, was also met (Box’s M = 12.48, p = .441) (Table 16).
Table 17
Box’s Test of Equality of Covariance Matrices
Box's M
|
12.487
|
F
|
1.005
|
df1
|
12
|
df2
|
76937.538
|
Sig.
|
.441
|
The results of the MANOVA (Table 17) indicated that a) there were not any significant differences between the three groups’ means on the pretest of lexical accuracy (F (2, 126) = .046, p = .955, partial η2 = .001 representing a weak effect size). Thus it can be claimed that they were homogenous in terms of their knowledge on lexical accuracy prior to the main study, b) there were not significant differences between the three groups’ means on the posttest of lexical accuracy (F (2, 126) = 2.93, p = .057, partial η2 = .044 representing an almost moderate effect size); although the results should be interpreted cautiously because the effect size enjoyed an almost moderate effect size and the results of the post-hoc comparisons (Table 18 below) indicated significant difference between the collaborative and e-collaborative groups on the posttest of lexical accuracy, c) there were significant differences between the three groups’ means on the delayed posttest of lexical accuracy (F (2, 126) = 4.84, p = .009, partial η2 = .071 representing a moderate effect size).
Table 18
Tests of Between-Subjects Effects; Pretest, Posttest and Delayed Posttest of Lexical Accuracy by Groups
Source
|
Dependent Variable
|
Type III Sum of Squares
|
df
|
Mean Square
|
F
|
Sig.
|
Partial Eta Squared
|
Group
|
Pretest
|
4.512
|
2
|
2.256
|
.046
|
.955
|
.001
|
Posttest
|
387.767
|
2
|
193.884
|
2.932
|
.057
|
.044
|
Delayed Posttest
|
588.326
|
2
|
294.163
|
4.841
|
.009
|
.071
|
Error
|
Pretest
|
6212.419
|
126
|
49.305
|
|
|
|
Posttest
|
8331.535
|
126
|
66.123
|
|
|
|
Delayed Posttest
|
7656.000
|
126
|
60.762
|
|
|
|
Total
|
Pretest
|
223312.00
|
129
|
|
|
|
|
Posttest
|
285839.00
|
129
|
|
|
|
|
Delayed Posttest
|
265703.00
|
129
|
|
|
|
|
Table 19 displays the means on the three groups on the pretest, posttest and delayed posttest of lexical accuracy.
Table 19
Descriptive Statistics; Pretest, Posttest and Delayed Posttest of lexical Accuracy by Groups
Dependent Variable
|
group
|
Mean
|
Std. Error
|
95% Confidence Interval
|
Lower Bound
|
Upper Bound
|
Pretest
|
collaborative
|
40.837
|
1.071
|
38.718
|
42.956
|
individual
|
40.953
|
1.071
|
38.834
|
43.073
|
e-collaborative
|
41.279
|
1.071
|
39.160
|
43.398
|
Posttest
|
collaborative
|
48.000
|
1.240
|
45.546
|
50.454
|
individual
|
47.093
|
1.240
|
44.639
|
49.547
|
e-collaborative
|
43.953
|
1.240
|
41.499
|
46.408
|
Delayed Posttest
|
collaborative
|
47.070
|
1.189
|
44.717
|
49.422
|
individual
|
45.070
|
1.189
|
42.717
|
47.422
|
e-collaborative
|
41.884
|
1.189
|
39.531
|
44.236
|
The results of the post-hoc comparison tests (Table 20) indicated that a) there was not any significant difference between the collaborative (M = 48) and individual (M = 47.09) groups’ means on the posttest of lexical accuracy (MD = .907, p = .606), b) there was not any significant difference between the individual (M = 47.09) and e-collaborative (M = 43.95) groups’ means on the posttest of lexical accuracy (MD = 3.14, p = .076), c) The collaborative group (M = 48) significantly outperformed the e-collaborative group (M = 43.95) on the posttest of lexical accuracy (MD = 4.04, p = .023), d) there was not any significant difference between the collaborative (M = 47.07) and individual (M = 45.07) groups’ means on the delayed posttest of lexical accuracy (MD = 2, p = .236), e) there was not any significant difference between the individual (M = 45.07) and e-collaborative (M = 41.88) groups’ means on the delayed posttest of lexical accuracy (MD = 3.18, p = .060), f) the collaborative group (M = 47.07) significantly outperformed the e-collaborative group (M = 41.88) on the delayed posttest of lexical accuracy (MD = 5.18, p = .003).
Table 20
Post-Hoc Comparisons; Posttest and Delayed Posttest of lexical Accuracy by Groups
Dependent Variable
|
(I) group
|
(J) group
|
Mean Difference (I-J)
|
Std. Error
|
Sig.
|
95% Confidence Interval for Difference
|
Lower Bound
|
Upper Bound
|
Posttest
|
Collaborative
|
Individual
|
.907
|
1.754
|
.606
|
-2.564
|
4.378
|
e-collaborative
|
4.047*
|
1.754
|
.023
|
.576
|
7.517
|
Individual
|
e-collaborative
|
3.140
|
1.754
|
.076
|
− .331
|
6.610
|
Delayed
Posttest
|
Collaborative
|
Individual
|
2.000
|
1.681
|
.236
|
-1.327
|
5.327
|
e-collaborative
|
5.186*
|
1.681
|
.003
|
1.859
|
8.513
|
Individual
|
e-collaborative
|
3.186
|
1.681
|
.060
|
− .141
|
6.513
|
*. The mean difference is significant at the .05 level.
|
, posttest and delayed posttest of lexcial accuracy by groups