3. Results
3.1 Test A
Table 12
Descriptive Statistical Data on Test A
Test n M SD SE
95% CI
Min Max
Min Max
Pretest 1 91 5.81 2.832 .297 5.22 6.40 1 10
Pretest 2 91 8.41 2.352 .247 7.92 8.90 2 10
Posttest 1
(Immediate) 91 8.80 2.192 .230 8.35 9.26 0 10
Posttest 2
(Delayed) 91 8.18 2.731 .286 7.61 8.74 0 10
Total 364 7.80 2.787 .146 7.51 8.09 0 10
Note. CI = confidence interval.
A one-way ANOVA was conducted to investigate significant effects among the tests,
and the results demonstrate main effects among them [F (3, 360) = 25.667, p < .01, η²
= .18]. The effect size was calculated by the EXCEL worksheet retrieved from Mizumoto and Takeuchi (2008) and the result demonstrates that it was large. The result of post-hoc Games-Howell comparisons on each of the four tests demonstrates significant differences among the mean scores on pretest 1 and pretest 2 (p < .01, d = 1.00), posttest 1 (p < .01, d = 1.18), and posttest 2 (p < .01, d = 0.85). Notably, there were no significant differences among each of the mean scores on pretest 2, posttest 1, and posttest 2. Figure 9 presents the sequence of the progression of the participants’ learning. The mean scores on pretest 2 (8.41), posttest 1 (8.80), and posttest 2 (8.18) were greater than the mean score on pretest 1 (5.81). Although the mean score on pretest 2 was statistically at the same level as on posttest 1 (the immediate
posttest), the highest mean score on posttest 1 slightly decreased, to the same level as that on posttest 2 (the delayed posttest).
Figure 9. Mean Scores on Test A
A histogram of Test A (Figure 10) clearly indicates that the number of participants who scored 10 points on pretest 2 (51%), posttest 1 (63%), and pretest 2 (52%) was greater than that on pretest 1 (16%). The number of participants who scored 10 points on all four tests was 12 (13%), that of those who scored 10 points from pretest 2 to posttest 2 was 25 (27%), and that of those who scored 10 points in posttest 1 and posttest 2 was 1 (1%). If we regard participants who scored 9 points on those tests as those who learned sufficiently, the
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
pre-test 1 pre-test 2 post-test 1 post-test 2
Mean Scores
Tests
Mean Scores on Test A
number of those who had learned without treatment was 20 (22%) out of 91 (100%). The number of those who learned after treatment was 35 (38%) out of 71 (78%) (the number of the participants who had not learned yet). In other words, approximately 50% of the participants who had not learned postmodification by themselves before the treatments learned this grammar point sufficiently after the treatments.
Figure 10. Histogram of Test A
Table 13 presents the mean score on each item in Test A. The highest mean score was 0.75 (No. 4), and the lowest mean score was 0.44 (No. 9) in pretest 1. The mean score on No. 4 was 0.89 in posttest 2, and the mean score on No. 9 was 0.74 in posttest 2. The lower
0 10 20 30 40 50 60 70 80 90 100
0 1 2 3 4 5 6 7 8 9 10
Frequency (%)
Scores
Histogram of Test A
pre-test 1 pre-test 2 post-test 1 post-test 2
mean scores in pretest 1 increased in posttest 2 as a result. Notably, all the mean scores in Test A increased significantly in pretest 2 after Activities 1, 2, and 3 were administered.
Table 13
Response Rates of Correct Answers on the Items in Each Test
Test 1 2 3 4 5 6 7 8 9 10
Pretest 1 0.58 0.52 0.59 0.75 0.49 0.73 0.66 0.48 0.44 0.57 Pretest 2 0.88 0.91 0.85 0.87 0.81 0.89 0.82 0.79 0.75 0.84 Posttest 1 0.88 0.90 0.87 0.93 0.82 0.90 0.81 0.95 0.85 0.89 Posttest 2 0.84 0.90 0.87 0.89 0.80 0.75 0.88 0.78 0.74 0.74
Table 14 presents the response rates for each choice in pretest 1. Approximately 20%–30% of the participants chose a particular distractor on each item. There were three distractors in each item. The first type had distractors A, B, and C in items No. 1 and No. 6.
In pretest 1, approximately 25% of the participants chose distractor B, which presented a different meaning in a different construction, namely, a prepositional phrase used as an adverbial phrase. The second type had distractors B, C, and D in items No. 5 and No. 10. In pretest 1, approximately 25% of the participants chose distractor D, in which the nouns in the postmodified noun phrase of the correct answer were switched. The third type had distractors A, D, and E in item Nos. 2, 3, 4, 7, 8, and 9. In pretest 1, the most commonly chosen distractor was D, except for No. 8 and No. 9. In pretest 1, the results on items No. 8 and No. 9
showed that the number of participants who chose distractors was equally divided (Table 14).
Table 14
Response Rates in Pretest 1
Choices 1 2 3 4 5 6 7 8 9 10
Correct 0.58 0.52 0.59 0.75 0.49 0.73 0.66 0.48 0.44 0.57 Distractor A 0.07 0.04 0.08 0.05 0.01 0.04 0.18 0.20
Distractor B 0.25 0.14 0.23 0.08
Distractor C 0.10 0.09 0.03 0.12
Distractor D 0.33 0.22 0.16 0.27 0.27 0.10 0.19 0.23
Distractor E 0.10 0.11 0.03 0.01 0.23 0.18
Notably, the number of participants who chose distractors gradually decreased in pretest 2 and posttest 1. By contrast, the number of participants who chose correct answers increased in pretest 2 and posttest 1 (Tables 15 and 16).
Table 15
Response Rates in Pretest 2
Choices 1 2 3 4 5 6 7 8 9 10
Correct 0.88 0.91 0.85 0.87 0.81 0.89 0.82 0.79 0.75 0.84 Distractor A 0.03 0.03 0.03 0.03 0.04 0.03 0.05 0.14
Distractor B 0.08 0.04 0.07 0.01
Distractor C 0.01 0.01 0.00 0.01
Distractor D 0.05 0.10 0.08 0.13 0.14 0.10 0.10 0.14
Distractor E 0.00 0.02 0.02 0.00 0.05 0.01
Table 16
Response Rates in Posttest 1
Choices 1 2 3 4 5 6 7 8 9 10
Correct 0.88 0.90 0.87 0.93 0.82 0.90 0.81 0.95 0.85 0.89 Distractor A 0.02 0.03 0.03 0.00 0.00 0.09 0.01 0.08
Distractor B 0.09 0.05 0.09 0.04
Distractor C 0.01 0.02 0.01 0.00
Distractor D 0.07 0.08 0.05 0.10 0.07 0.04 0.02 0.07
Distractor E 0.00 0.01 0.01 0.03 0.00 0.05
In posttest 2 (the delayed posttest, after implementing Task Activity), the mean score
on each item decreased but was significantly different from that in pretest 1 (p < .01, d =
0.85). As presented in Table 17, the number of participants who chose distractor B or D slightly increased.
Table 17
Response Rates in Posttest 2
Choices 1 2 3 4 5 6 7 8 9 10
Correct 0.84 0.90 0.87 0.89 0.80 0.75 0.88 0.78 0.74 0.74 Distractor A 0.02 0.00 0.03 0.00 0.03 0.00 0.05 0.11
Distractor B 0.09 0.04 0.22 0.04
Distractor C 0.05 0.00 0.00 0.05
Distractor D 0.08 0.09 0.09 0.15 0.10 0.10 0.11 0.16
Distractor E 0.02 0.01 0.02 0.02 0.07 0.04
3.2 Test B
To clarify whether the participants could correctly produce postmodified noun phrases by means of rearranging given words, statistical analyses and error analysis were performed. Table 18 presents descriptive statistical data on the Test B scores. The scores are categorized for each of the four tests (pretest 1, pretest 2, posttest 1, and posttest 2) to demonstrate the differences in the mean scores among them.
Table 18
Descriptive Statistical Data on Test B
Test n M SD SE
95% CI
Min Max
Min Max
Pretest 1 91 5.52 2.994 .314 4.89 6.14 0 8
Pretest 2 91 6.66 2.600 .273 6.12 7.20 0 8
Posttest 1 91 7.14 2.148 .225 6.70 7.59 0 8
Posttest 2 91 6.93 2.220 .233 6.47 7.40 0 8
Total 364 6.56 2.581 .135 6.30 6.83 0 8
Note. CI = confidence interval.
A one-way ANOVA was conducted to investigate whether there were significant
effects on Test B, and the results demonstrate main effects [F (3, 360) = 7.580, p < .01, η²
= .06]. Based on Mizumoto and Takeuchi (2008), the effect size was small. The result of post-hoc Games-Howell comparisons on each of the four tests shows significant differences among the mean score on pretest 1 and pretest 2 (p < .01, d = 0.41), posttest 1 (p < .01, d = 0.62), and posttest 2 (p < .01, d = 0.54). The results on Test B were similar to those on Test A, with no significant differences between the mean scores on pretest 2, posttest 1, and posttest 2.
Figure 11 presents the sequence of the progression of the participants’ learning. The mean scores on pretest 2 (6.44), posttest 1 (7.08), and posttest 2 (6.80) were higher than that on pretest 1 (5.25). Although the mean score on pretest 2 was statistically at the same level as on posttest 1 (the immediate posttest), the highest mean score on posttest 1 slightly decreased on
posttest 2 (the delayed posttest).
Figure 11. Mean Scores on Test B
A histogram of Test B (Figure 12) clearly shows that the number of participants who scored 8 points (full points) on pretest 2 (69%), posttest 1 (80%), and pretest 2 (73%) was higher than that on pretest 1 (44%). The number of participants who scored full points on all four tests was 38 (42%), that of those who scored full points from pretest 2 to posttest 2 was 7 (18%), and that of those who scored full points on posttest 1 and posttest 2 was 3 (3%). If we regard participants who scored 7 points on those tests as those who had learned sufficiently,
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
pre-test 1 pre-test 2 post-test 1 post-test 2
MeanScores
Test
Mean Scores on Test B
the total number of those who had learned without treatment was 50 (55%). The number of those who learned after treatment was 17 (19%) out of 41 (45%) (the number of participants who had not learned yet).
Figure 12. Histogram of Test B
Table 19 presents the mean scores on each item in Test B. The highest mean score was 0.77, and the lowest mean score was 0.62 on pretest 1. The highest mean score on pretest 1 was higher than that on posttest 2. Notably, the lowest mean score on pretest 1 is not lower than that on the items in posttest 2. All the mean scores on pretest 1 increased after pretest 2.
0 10 20 30 40 50 60 70 80 90 100
0 1 2 3 4 5 6 7 8
Frequency (%)
Scores
Histogram of Test B
pre-test 1 pre-test 2 post-test 1 post-test 2
Table 19
Response Rates of Correct Answers on Each Test
Test 1 2 3 4 5 6 7 8
Pretest 1 0.62 0.64 0.69 0.73 0.67 0.77 0.66 0.75 Pretest 2 0.85 0.84 0.86 0.86 0.75 0.86 0.79 0.87 Posttest 1 0.90 0.90 0.89 0.92 0.88 0.86 0.88 0.91 Posttest 2 0.87 0.91 0.89 0.90 0.82 0.87 0.85 0.82
Table 20 presents the response rates for each correct and incorrect answer in each item on pretest 1. From 23% to 38% of the participants did not place the correct nouns before and after the prepositions corresponding to the meanings. Most of the participants did not put the preposition before the two nouns (e.g., under the cat the box) in pretest 1 (see the section marked “Others” in Table 20). On pretest 2, from 13% to 25% of the participants made the same errors as on pretest 1, but no participants placed the prepositions incorrectly (Table 21).
On posttest 1, from 8% to 14% of the participants made the same errors as on pretests 1 and 2 (Table 22). On posttest 2, from 9% to 18% of the participants made the same errors as on the pretests and posttest 1 (Table 23).
Table 20
Response Rates of Correct and Incorrect Answers on Pretest 1
Choices 1 2 3 4 5 6 7 8
Correct 0.62 0.64 0.69 0.73 0.67 0.77 0.66 0.75 Incorrect 0.38 0.35 0.31 0.26 0.33 0.23 0.33 0.25
Others 0.00 0.01 0.00 0.01 0.00 0.00 0.01 0.00
Table 21
Response Rates of Correct and Incorrect Answers on Pretest 2
Choices 1 2 3 4 5 6 7 8
Correct 0.85 0.84 0.86 0.86 0.75 0.86 0.79 0.87 Incorrect 0.15 0.16 0.14 0.14 0.25 0.14 0.21 0.13 Others 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Table 22
Response Rates of Correct and Incorrect Answers on Posttest 1
Choices 1 2 3 4 5 6 7 8
Correct 0.9 0.9 0.89 0.92 0.88 0.86 0.88 0.91 Incorrect 0.09 0.1 0.11 0.08 0.12 0.14 0.12 0.09
Others 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Table 23
Response Rates of Correct and Incorrect Answers on Posttest 2
Choices 1 2 3 4 5 6 7 8
Correct 0.87 0.91 0.89 0.9 0.82 0.87 0.85 0.82 Incorrect 0.13 0.09 0.11 0.1 0.18 0.13 0.15 0.18 Others 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
As Tables 21 and 22 show, the number of participants who produced the postmodified noun phrases correctly gradually increased on pretest 2 and posttest 1. On posttest 2 (the delayed posttest), the mean score on each item slightly decreased but was significantly different from that on pretest 1 (p < .01, d = 0.54).
3.3 Test C
To clarify whether understanding the construction of a postmodified noun phrase can help lower secondary school students grasp the construction of sentences with postmodification accurately, statistical analyses and error analysis were performed. Table 24 presents the descriptive statistical data on the Test C scores. The scores are categorized for each of the four tests (pretest 1, pretest 2, posttest 1, and posttest 2) to demonstrate the differences in the mean scores among them (Figure 13).
Table 24
Descriptive Statistical Data on Test C
Test n M SD SE
95% CI
Min Max
Min Max
Pretest 1 91 3.86 2.589 .271 3.32 4.40 0 8
Pretest 2 91 4.10 2.595 .272 3.56 4.64 0 8
Posttest 1 91 4.24 2.579 .270 3.70 4.78 0 8
Posttest 2 91 4.63 2.739 .287 4.06 5.20 0 8
Total 364 4.63 2.630 .138 3.93 4.48 0 8
Note. CI = confidence interval.
A one-way ANOVA was conducted to investigate whether there were significant effects among the tests, and the results demonstrate that there were none. [F (3, 360) = 1.368, p
= .252, η² = .01]. The effect size was small (Mizumoto and Takeuchi, 2008).
Figure 13. Mean Scores on Test C
A histogram of Test C (Figure 14) clearly indicates that the number of participants who scored full points is small, but notably it increases on each test.
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
pre-test 1 pre-test 2 post-test 1 post-test 2
Mean Score
Test
Mean Scores on Test C
Figure 14. Histogram of Test C
Table 25 presents the mean score on each item in Test C. The highest mean score is 0.76, and the lowest mean score is 0.30. The highest mean score on pretest 1 was on No. 7 (The ice cream in the cup is yours.); this became the highest (0.90) on pretest 2, but became increasingly lower on posttest 1 (0.85) and posttest 2 (0.77). The second highest mean score (0.72) was on No. 5 (People enjoy jogging in the park.), but this sentence did not present postmodification. The mean score decreased on pretest 2 (0.63) but became increasingly
0 10 20 30 40 50 60 70 80 90 100
0 1 2 3 4 5 6 7 8
Frequency (%)
Scores
Histogram of Test C
pre-test 1 pre-test 2 post-test 1 post-test 2
higher on posttest 1 (0.80) and posttest 2 (0.85). The lowest mean score (0.30) on pretest 1 was on No. 4 (Books in that library are popular.); this gradually increased from pretest 2 to posttest 2, but that increase does not represent sufficient improvement.
Table 25
Response Rates on Each Item in Each Test
Test 1 2 3 4 5 6 7 8
Pretest 1 0.50 0.33 0.37 0.30 0.72 0.51 0.76 0.32 Pretest 2 0.56 0.39 0.40 0.34 0.63 0.50 0.90 0.34 Posttest 1 0.51 0.45 0.39 0.35 0.80 0.47 0.85 0.42 Posttest 2 0.56 0.44 0.47 0.41 0.85 0.58 0.77 0.51
The highest mean score on posttest 2 was on No. 7. The sentence consisted of four words or phrases, but the mean score on No. 8, which also consisted of four words or phrases, was relatively low. The mean scores on No. 3 and No. 6, in which the sentences consisted of six words or phrases, were relatively low, but the positions of the postmodified noun phrase were different (subject vs. object). The score on No. 3 (0.47) was slightly lower than that on No. 6 (0.58). The preposition used for No. 2 was the same as that in No. 6, and the number of words or phrases in No. 2 was five, but the mean score was slightly lower than that on No. 6.
The position of the postmodified noun phrase in No. 2 was a subject.
The errors were analyzed in light of how the participants rearranged the given words or phrases to form complete sentences. A major error in No. 1 was “The students speak English from France.” The prepositional phrase was used as an adverbial phrase. This error is of type 1. Approximately 30% of the participants provided this response in pretest 1, 27% in pretest 2 and posttest 1, and 21% in posttest 2.
The second major error was “France from the students speak English.” This error is of type 2. Tables 26 to 31 show the response rates on each item from pretest 1 to posttest 2.
The participants who made type 2 errors did not place the given nouns in grammatical positions in the postmodified noun phrase in relation to the meanings. The rate of type 2 errors was lower than of type 1.
Table 26
Response Rates on Frequent Answers to No. 1 from Pretest 1 to Posttest 2
pretest 1 pretest 2 posttest 1 posttest 2 Correct 0.50 0.56 0.51 0.56 Error type 1 0.30 0.27 0.27 0.21 Error type 2 0.06 0.03 0.03 0.03
Table 27
Response Rates on Frequent Answers to No. 2 from Pretest 1 to Posttest 2
pretest 1 pretest 2 posttest 1
posttest 2 Correct 0.33 0.39 0.45 0.44 Error type 1 0.17 0.11 0.09 0.11 Error type 2 0.12 0.10 0.10 0.03
Table 28
Response Rates on Frequent Answers to No. 3 from Pretest 1 to Posttest 2
pretest 1 pretest 2 posttest 1
posttest 2 Correct 0.37 0.40 0.39 0.47 Error type 1 0.12 0.09 0.10 0.04 Error type 2 0.04 0.02 0.05 0.02
Table 29
Response Rates on Frequent Answers to No. 4 from Pretest 1 to Posttest 2
pretest 1 pretest 2 posttest 1
posttest 2 Correct 0.30 0.34 0.35 0.41 Error type 1 0.15 0.16 0.22 0.22 Error type 2 0.18 0.23 0.10 0.07
Table 30
Response Rates on Frequent Answers to No. 6 from Pretest 1 to Posttest 2
pretest 1 pretest 2 posttest 1 posttest 2 Correct 0.51 0.50 0.47 0.58 Error type 1 0.08 0.05 0.10 0.09 Error type 2 0.03 0.05 0.05 0.05
Table 31
Response Rates on Frequent Major Answers to No. 8 from Pretest 1 to Posttest 2 pretest 1 pretest 2 posttest 1
posttest 2 Correct 0.32 0.34 0.42 0.51 Error type 1 0.32 0.45 0.28 0.17 Error type 2 *** *** *** ***
Note. *** Nouns could not be switched in this item because the given phrase was “in the park”
3.4. Test D
As mentioned in Section 2.3.3, this test was administered to compare data with those in surveys in 2012, 2013, and 2015 in Okumura (2017), which had no treatments for acquisition of postmodification. Tables 32 and 33 show the results on the test.
Table 32
Descriptive Statistical Data Categorized into Grammar Items
Administ- rative Year n
Present Participial Phrase
Past Participial Phrase
Prepositional Phrase
M SD M SD M SD
2012 90 2.48 1.31 2.56 1.22 2.54 1.43
2013 104 2.68 1.33 2.38 1.30 2.40 1.36
2015 99 2.81 1.21 2.59 1.19 2.45 1.44
2016 101* 2.91 1.28 2.66 1.31 2.93 1.30
Note. *The number of the participants are different from that of the present study because the researcher was not able to exclude the participants who were absent from the language activities and the pretests and the posttests8.
Table 33
Descriptive Statistical Data Categorized by Year
Year n M SD SE
95% CI
Min Max
Min Max
2012 90 2.53 1.32 .080 2.37 2.68 0.33 4
2013 104 2.49 1.33 .074 2.35 2.64 0 4
2015 95 2.62 1.29 .078 2.47 2.77 0 4
2016
(present survey) 101* 2.84 1.30 .075 2.69 2.98 0.33 4
Note. CI = confidence interval.
The number of participants was also different from that of the present study, as explained in Table 32.
Figure 15. Response Rates on Correct Answers in Each Year
In these surveys from Okumura (2017), the mean scores were categorized in relation
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2012 2013 2015 2016
Rate (%)
Year
Mean Response Rates on Correct Answers in Each Year
to three grammar items (present participial phrases, past participial phrases, and prepositional phrases) and according to the three years of administration (2012, 2013, and 2015). To compare the data of the 2016 survey with those from the other surveys (Figure 15), the scores were analyzed in the same manner as a two-way ANOVA conducted to investigate whether there were significant effects among the years the surveys were administered. The results show main effects among the years [F (3, 1169) = 4.214, p < .01, η² = .01], and the result of post-hoc Games-Howell comparisons on each of the four years indicates significant differences between the mean scores in 2016 and 2012 (p < .05, d = 0.23)–2013 (p < .05, d = 0.26). No effects were observed among the three grammar items [F (2, 1169) = 1.885, p = n.s., η² = .00].
Figure 16 presents a proportion of the frequency rates of scores according to four classes, namely, A, B, C, and D, in each year. For example, class A shows the frequency rates for participants who scored 10, 11, or 12 points. Comparing the rates of class A in each year, the frequency of the participants who scored from 10 to 12 is approximately 10%–15%
higher in 2016 (this study) than that in the years in which no treatments were administered.
Figure 16. Classified Proportion of Frequency Rates of Scores
3.5 Correlation of the Results on the Tests
Research question 3 asks whether understanding the construction of postmodified noun phrases can help Japanese lower secondary school students accurately grasp the word order in sentences with postmodification. To answer the question, it was necessary to correlate the data from the tests. Table 34 presents correlations of the frequencies of total scores between Test A, Test B, and Test C in posttest 2. The results indicate Test A is strongly
related to Test B [r = .673, p < .01]; Test A is also strongly related to Test C [r = .610, p
< .01]; and Test B is moderately related to Test C [r = .511, p < .01].
49 37 32
37
23 21
30 21
18 28 23
31
11 9 18
12
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
2016 2015 2013 2012
Frequency Rate (%)
Administratve Year
Proportion of Frequency Rates of Graded Scores
A(12,11,10) B(9, 8, 7) C(6, 5, 4) D(3,2,1,0)
Table 34
Correlations between Test A, Test B, and Test C on Posttest 2 Test A Test B Test C
Test A Test B Test C
1.000 .673*
.610*
1.000
.511* 1.000 Note. *p < .01
Figure 17 shows the relationship between the scores on Test A and Test C. The participants who scored higher on Test A scored higher on Test C, and vice versa. However, even if they scored higher on Test A, they did not match those scores on Test C.
Figure 17. Scatter Diagram of theRelationship between Scores on Test A and Test C in Posttest 2
Figure 18 similarly shows the relationship between the scores on Test B and Test C.
0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8 9 10
Score on Test C
Score on Test A
Relationship between Test A and Test C
The participants who scored higher on Test B scored higher on Test C, and vice versa.
However, even if they scored higher on Test B, they did not match those scores on Test C.
Figure 18. Scatter Diagram of the Relationship between Scores on Test B and Test C in Posttest 2
Figure 19 shows the relationship between the scores on Test A and Test B. The participants who scored higher on Test A also scored higher points on Test B, and vice versa.
Comparing with the relationship between Test A and Test C, and between Test B and Test C, there were not so many participants who scored higher on each test.
0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8
Score on Test C
Score on Test B
Relationship between Test B and Test C
Figure 19. Scatter Diagram of the Relationship between Scores on Test A and Test B in Posttest 2
0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8 9 10
Score on Test B
Score on Test A
Relationship between Test A and Test B
4. Discussion
The purpose of this study was to clarify the extent to which a battery of task-based
language activities could help Japanese lower secondary school students notice the existence
of postmodified noun phrases by comprehending the construction of such noun phrases,
namely, the order of the modificant and the modifier. Based on the analyses of the results, a
discussion of the answers to the research questions follows.
4.1 Research Question 1
The results on the tests clarify whether Task-Oriented Activities and structured
input activities helped the lower secondary school students understand the construction of
postmodified noun phrases in relation to their corresponding meanings. Test A measured the
ability to distinguish between the function of prepositional phrases as adjectives modifying
nouns and as adverbs describing places or positions. Test B measured the ability to produce postmodified noun phrases corresponding to the intended meaning by rearranging given words or phrases.
The results and the data analyses of the progress of the mean scores indicate that the
participants comprehended the function of prepositional phrases used to postmodify nouns
and the word order in postmodified noun phrases in relation to the intended meanings. In
other words, implementing Task-Oriented Activities and structured input activities clearly
helped the participants understand the relation between the head noun that was modified
and the prepositional phrase that modified it, by comparing the same construction with a
different word order and meaning (e.g., “The yellow card on the blue card.” vs. “The blue card
on the yellow card’”). In short, the participants grasped the form-meaning mappings of postmodified noun phrases.
The results on posttest 2 of Test A show their knowledge was retained, because the mean score did not decrease in the 6 weeks subsequent to the administration of posttest 1.
Clearly, the effectiveness of a combination of productive language activities (Task-Oriented Activities) and receptive language activities (structured input activities) is observed in the results on pretest 2. Furthermore, from the results on posttest 1 and posttest 2 after implementing the Task Activity, the following assertion is warranted: the effectiveness of this combination of activities clearly helped the participants learn about postmodified noun phrases.
The results on pretest 1 and pretest 2 (Histogram of Test A), show 22% of the participants scored 9 or 10 points on each test, so they may already have perceived the existence of postmodified noun phrases and distinguished their usage. However, another 49%
of the participants clearly had not previously learned about postmodified noun phrases (38%
overall), but they did so after the treatments.
Regarding mean scores on different items, the results demonstrate that the items with the highest and lowest scores in Test A were No. 4 and No. 9. These had the same illustrations of ice cream in a cup and ice cream in a cone but different arrows that pointed to one or the other: the arrow in No. 4 pointed to ice cream in a cup, and the arrow in No. 9 pointed to ice cream in a cone. Apparently, some participants did not comprehend the researcher’s intention after viewing the illustration for No. 9. According to our error analysis, one-third of the participants chose one of the three distractors. Finally, however, because all the mean scores increased in posttest 2, it is clear that the participants had learned the meanings of the illustrations by performing Activity 3.
The error analysis in Table 17 indicates two reasons participants chose distractors B and D. Those participants who chose distractor B regarded the prepositional phrases as adverbial phrases (e.g., The ice cream is yours in the cone.). Such participants need to learn that postmodifying prepositional phrases are usually placed at the ends of sentences.
Certainly this is the pattern observed in the textbooks approved by MEXT are at the end of the sentences. Those participants who chose distractor D did not distinguish between the word order of the nouns in the postmodified noun phrase and the intended meaning (i.e., “the
ice cream in the cone” vs. “the cone in the ice cream”). The number of such participants decreased in posttest 1, but finally, approximately 10% to 20% of them did not learn the relationship between the word order and the intended meaning of the postmodified noun phrase (Table 17).
Test B measured whether Japanese lower secondary school students could produce postmodified noun phrases corresponding to particular meanings by rearranging given words or phrases. The result on Test B and the data analysis demonstrate that the participants distinguished between one postmodified noun phrase (e.g., the notebook on the book) and another (e.g., the book on the notebook) whose nouns were switched in the construction, and that they produced postmodified noun phrases corresponding to the meanings. They cognitively distinguished the difference between the meanings of “the blue card on the yellow card” and “the yellow card on the blue card” and understood the construction that described those meanings in the Task-Oriented Activities. In the test, the participants were presented with constructions they were seeing for the first time, except for one pair of items (“the ice cream in the cup” and “the cup in the ice cream”). Thus, they had to apply what they had picked up in the Task-Oriented Activities and structured input activities to find the correct constructions.
The results on posttest 2 of Test B demonstrate good retention rates, because the
mean score did not decrease in the 6 weeks subsequent to the administration of posttest 1. As with Test A, the effectiveness of a combination of productive language activities (Task-Oriented Activities) and receptive language activities (structured input activities) is clear. The results on posttest 1 and posttest 2, after implementing the Task Activity, also show the effectiveness of that activity in helping students create sentences with postmodification.
The results on pretest 1 (Figure 10) showed that 55% of the participants scored seven or eight (full points) before the treatments. This result indicates that more than half of them might already have obtained some understanding of postmodification. Another view of these results is that rearranging three parts to create noun phrases might be easy for them. As Table 20 shows, there were two choices. For example, to answer item No. 1, the participants wrote
“the book on the notebook” or “the notebook on the book,” and none answered, “on the book the notebook” or “the book the notebook on.”
This study was conducted from December to January; therefore, it is possible that textbook or other language activities had provided the participants with exposure to prepositional phrases (e.g., “on the desk” or “in Japan”), and that they had learned how to produce prepositional phrases before the treatments. If so, it might be that combining a noun with a prepositional phrase on the test was relatively easy for the participants. However, they had not been exposed to postmodified structures. So it is reasonable to assert that they learned