Effects of Semantic Mapping Strategy on Reading Comprehension

CHAPTER FOUR RESULTS AND DISCUSSION

This chapter aims to present and discuss the results of the study. In accordance with the research questions, the results of the effects of semantic mapping strategy on participants’ reading comprehension and on different proficiency levels are presented first. Whether text difficulty and text structure have any effects on the results of using semantic mapping strategy are presented next. Then, the experimental group’s

responses to the semantic mapping training and application, including the

post-treatment questionnaire and the interviews of the experimental group will be described. All the results of the study will be discussed in detail in the last section.

Effects of Semantic Mapping Strategy on Reading Comprehension

In this section, the test results of the control and experimental groups are

presented. In order to investigate the effects of semantic mapping strategy on reading comprehension, all the scores on recall protocols and multiple-choice comprehension tests were translated into percentage, and then SPSS Multivariate Tests (one-way MANOVA) were employed for the analysis.

Written Recall Protocols

The means and standard deviation of the three the written recall protocols between the experimental and the control groups are listed in Table 9.

Table 9

Means and Standard Deviation of the Written Recall Protocols of the Two

Groups

(total score for each article= 100) Experimental Group (N=64)

M SD

Control Group (N=64) M SD

Article 1 67.00 10.30 65.72 10.72

Article 2 48.58 14.57 38.72 19.41

Article 3 33.86 14.42 27.31 15.26

Average 49.91 13.10 43.92 15.13

According to Table 9, the experimental group had higher means than the control

group. The results of one-way MANOVA (see Table 10) showed that treatment (semantic mapping strategy) had a significant overall effect on the written recall protocols, F(3,124)= 4.341 (p < .01). That is, there was a significant difference between the two groups, and the experimental group significantly outperformed the control group. Thus, semantic mapping activity facilitated students to recall idea units after reading these articles.

Table 10

Summary of Multivariate Analysis for the Written Recall Protocols of the Two Groups

Effect Value Hypothesis df Error df F

Treatment Pillai's Trace .095 3 124 4.341**

** p < .01

Multiple-choice Comprehension Test

Table 11 presents the means and standard deviation of the three multiple-choice comprehension tests between the experimental and the control groups.

Table 11

Means and Standard Deviation of the Multiple-choice Comprehension Tests of the Two Groups

(total score for each article= 100) Experimental Group (N=64)

M SD

Control Group (N=64) M SD

Article 1 64.41 16.24 59.33 16.64

Article 2 66.41 20.03 58.59 22.37

Article 3 64.06 20.53 55.93 16.01

Average 64.96 15.60 57.95 18.34

As shown in Table 11, the mean scores of the experimental group were higher than those of the control group. However, the results of one-way MANOVA (see Table 12) indicated that treatment (semantic mapping strategy) had no significant overall effect on the multiple-choice comprehension tests, F(3,124)= 2.641 (p > .05);

no significant difference was found between the experimental and the control groups.

Table 12

Summary of Multivariate Analysis for the Multiple-choice Comprehension Tests of the Two Groups

Effect Value Hypothesis df Error df F

Treatment Pillai's Trace .060 3 124 2.641

Since no significant overall effect of treatment on the multiple-choice

comprehension tests existed, the subsequent Univariate F tests was conducted to test if treatment had effect on each article (see Table 13). The results showed that no significant effect between the two groups was found on Article 1 (F= 3.045, p > .05).

But there were significant differences between the two groups’ performance on Article 2 (F= 4.333, p < .05) and Article 3 (F= 6.235, p < .05), and the experimental group significantly outperformed the control group in comprehending Article 2 (M=66.41 vs.

58.59) and Article 3 (M= 64.06 vs. 55.93). Accordingly, though semantic mapping strategy did not have significant overall effect on students’ multiple-choice

comprehension tests, it still enhanced the experimental group to comprehend two of the three articles.

Table 13

Summary of ANOVA for the Multiple-choice Comprehension Tests between the Two Groups

SS df MS F

Article1 Between Groups 825.195 1 825.195 3.045

Within Groups 30407.547 126 270.219

Article2 Between Groups 1953.125 1 1953.125 4.333*

Within Groups 56796.875 126 450.769

Article3 Between Groups 2112.500 1 2112.500 6.235*

Within Groups 42687.500 126 388.790

* p < .05

Based on the results presented in Tables 9 to 13, it was obvious that semantic mapping strategy was facilitative in promoting students’ reading comprehension.

Effects of Semantic Mapping Strategy on Different Proficiency Levels

In order to see the effects of semantic mapping strategy on learners of different

proficiency levels, SPSS Multivariate analysis was conducted on each proficiency level.

Written Recall Protocols

Table 14 shows the mean scores and standard deviation of the two high proficiency groups’ written recall protocols.

Table 14

Means and Standard Deviation of the Written Recall Protocols of the Two High-proficiency Groups

(total score for each article= 100) Experimental Group (N=32)

M SD

Control Group (N=32) M SD

Article 1 73.56 7.64 72.16 7.72

Article 2 58.31 11.24 47.97 18.41

Article 3 43.97 10.47 35.44 15.40

Average 58.61 9.78 51.86 13.84

According to Table 14, the high-proficiency students in the experimental group (HE) had higher recall scores than those in the control group (HC) after reading the three articles. Besides, the results of one-way MANOVA analysis (see Table 15) also showed a significant overall effect of treatment between the two groups, F(3,60)=

3.275 (p < .05). That is, there was a significant difference between the written recall protocols of high proficiency students in experimental and control groups, and semantic mapping strategy enhanced the high-proficiency students’ recall performance after reading these articles.

Table 15

Summary of Multivariate Analysis for the Written Recall Protocols of the Two High-proficiency Groups

Effect Value Hypothesis df Error df F

Treatment Pillai's Trace .141 3 60 3.275*

* p < .05

As to the two low-proficiency groups, Table 16 shows their means and standard

deviation of the written recall protocols after reading the three articles.

Table 16

Means and Standard Deviation of the Written Recall Protocols of the Two Low-proficiency Groups (total score for each article= 100)

Experimental Group (N=32) M SD

Control Group (N=32) M SD

Article 1 60.44 8.27 59.28 9.41

Article 2 38.84 10.45 29.47 15.80

Article 3 23.75 10.10 19.19 9.99

Average 41.01 9.61 35.98 11.73

As indicated in Table 16, the low-proficiency students in the experimental group (LE) still had higher means than the low-proficiency students in the control group (LC). The results of one-way MANOVA analysis (see Table 17) also revealed that the treatment had a significant overall effect on the three written recall protocols,

F(3,60)= 2.830 (p < .05). Accordingly, semantic mapping strategy also facilitated the

low-proficiency students to recall more idea units after reading these articles.

Table 17

Summary of Multivariate Analysis for the Written Recall Protocols of the Two Low-proficiency Groups

Effect Value Hypothesis df Error df F

Treatment Pillai's Trace .124 3 60 2.830*

* p < .05

Multiple-choice Comprehension Tests

Table 18 presents means and standard deviation of the three multiple-choice comprehension tests between the two high proficiency groups.

Table 18

Means and Standard Deviation of the Multiple-choice Comprehension Tests of the Two High-proficiency Groups

(total score for each article= 100) Experimental Group (N=32)

M SD

Control Group (N=32) M SD

Article 1 73.94 10.73 68.66 15.21

Article 2 80.47 13.82 67.97 20.31

Article 3 75.63 15.01 62.50 14.14

Average 76.68 13.19 66.38 16.55

As shown in Table 18, HE still had higher means than HC in the three multiple-choice comprehension tests. Table 19, the results of one-way MANOVA analysis, indicated that a significant overall effect of treatment on the multiple-choice comprehension tests was found, F(3,60)= 6.215 (p < .01). That is, semantic mapping strategy enhanced the high-proficiency students’ comprehension of these articles.

Table 19

Summary of Multivariate Analysis for the Multiple-choice Comprehension Tests of the Two High-proficiency Groups

Effect Value Hypothesis df Error df F

Treatment Pillai's Trace .237 3 60 6.215**

Table 20 reveals the means and standard deviation of the three multiple-choice comprehension tests between the two low-proficiency groups.

Table 20

Means and Standard Deviation of the Multiple-choice Comprehension Tests of the Two Low-proficiency Groups

(total score for each article= 100) Experimental Group (N=32)

M SD

Control Group (N=32) M SD

Article 1 54.86 15.26 50.00 12.32

Article 2 52.34 14.70 49.22 20.56

Article 3 52.50 18.84 49.38 15.23

Average 52.23 16.27 49.53 16.04

As shown in Table 20, LE and LC had very similar means after reading the three articles. The results of one-way MANOVA analysis showed no significant overall effect of treatment between the two low-proficiency groups’ multiple-choice

comprehension tests, F(3, 60)= .856 (p > .05). Besides, the results of the subsequent

Univariate F tests (see Table 21) revealed that treatment had no significant effect on

each article (p > .05). Obviously, semantic mapping strategy did not enhance the

low-proficiency students’ comprehension on the three articles.

Table 21

Summary of ANOVA for the Multiple-choice Comprehension Tests between the Two Low-proficiency Groups

SS df MS F

Article1 Between Groups 380.250 1 380.250 1.977

Within Groups 11925.500 62 192.347

Article2 Between Groups 156.250 1 156.250 .489

Within Groups 19804.688 62 319.430

Article3 Between Groups 156.250 1 156.250 .533

Within Groups 18187.500 62 293.347

Based on the results presented in Tables 14 to 21, the high-proficiency students in the experimental group significantly outperformed those in the control group both in recall protocols and multiple-choice comprehension tests. However, the

low-proficiency students in the experimental group only performed significantly better than control group in the recall protocols. Clearly, it was the high-proficiency students that benefited more from semantic mapping instruction.

Effects of Difficulty Level and Structure of a Text on Semantic Mapping Strategy

In order to investigate whether the difficulty level and the structure of a text influenced the effects of semantic mapping strategy on students’ comprehension of the three articles in Practice Phase, the effect of treatment (semantic mapping strategy) on each article was further analyzed.

Difficulty Level of a Text and Semantic Mapping Strategy

According to readability level measured by Flesch-Kincaid Grade Level, the difficulty level of the three articles used in Practice Phase was Article 2 (8.0) >

Article 3 (7.6) > Article 1 (6.3). Article 2 and Article 3 were classified as Challenging;

while Article 1 was Easy.

The results in Table 9 showed the significant overall effect of treatment

(semantic mapping strategy) on the three written recall protocols between the two

groups, F(3,124) = 4.341 (p < .01). The subsequent LSD Pairwise Comparisons (see

Table 22) showed that no significant difference was found between the two groups in the recall protocol of Article 1 (p > .05). But, the experimental group performed significantly better than the control group in the recall protocols of Article 2 (p < .01) and Article 3 (p < .05). Clearly, semantic mapping strategy helped the experimental group to recall more idea units of the challenging texts (Article 2 and Article 3).

TABLE 22

Summary of LSD Pairwise Comparisons for the Written Recall Protocols of the Two Groups

(total score for each article= 100) (I) Means of

Exp. Group

(J) Means of Ctrl. Group

Std. Error Mean Difference (I-J)

Article 1 67.00 65.72 1.86 1.28

Article 2 48.58 38.72 3.03 9.86**

Article 3 33.86 27.31 2.62 6.55*

* p < .05 ** p < .01

As to the effect of treatment on three multiple-choice comprehension tests, Table 12 showed no significant overall effect of treatment between the two groups (p > .05).

While, based on the means (see Table 11) and the results of subsequent univariate F test (see Table 13), the experimental group and the control group also showed significant differences in the multiple-choice comprehension tests of Article 2 (M=

66.41 vs. 58.59, p < .05) and Article 3 (M= 64.06 vs. 55.93, p < .05). It was obvious that the experimental groups performed significantly better than the control group in reading the two challenging texts, Article 2 and Article 3. That is, semantic mapping strategy promoted students’ comprehension of challenging texts.

The results mentioned above indicated that semantic mapping strategy did not

facilitate students’ reading comprehension when the text was easy (Article 1), but it

did help students to comprehend the challenging texts both in terms of recall protocols

and multiple-choice questions. Clearly, the difficulty level of the text influenced the

effects of semantic mapping strategy on reading comprehension.

Structure of a Text and Semantic Mapping Strategy

According to Carrell (1984b), readers comprehended a text better when the structure of the text was more organized. Since semantic mapping strategy was an effective technique to help students see the structure of a text (Armbruster et al., 1984;

Reutzel, 1986), it was worthwhile to investigate the relationships between the structure of a text and semantic mapping strategy; that is, whether the structure of a text influenced the effects of semantic mapping strategy on reading comprehension. In this study, Article 2 represented a well-organized text and Article 3 was a

less-organized text. The following discussion focused on the effects of semantic mapping strategy on the comprehension of these two articles.

The results of Table 22 showed that the experimental group significantly outperformed the control group in the recall protocols of Article 2 (M= 48.58 vs.

38.72, p < .01) and Article 3 (M = 33.86 vs. 27.31, p < .05). Table 13 showed that the scores of the multiple-choice comprehension tests of Article 2 and Article 3 between the two groups were also significantly different (F = 4.333 & 6.235, p < .05) with the experimental group performed significantly better than the control group (see Table 11). Obviously, semantic mapping strategy did facilitate the experimental group to recall and comprehend both the well-organized text and the less-organized text.

Besides, the experimental group recalled the well-organized text (Article 2, p < .01) better than the less-organized text (Article 3, p < .05)

In addition, taking the proficiency levels into consideration, the results in Table 15 showed that treatment had a significant overall effect on the recall protocols of the two high-proficiency groups. The follow-up LSD Pairwise Comparison (see Table 23) indicated HE performed significantly better than HC in the recalls of Article 2 (M=

58.31 vs. 47.97, p < .01) and Article 3 (M= 43.97 vs. 35.44, p < .05). Clearly,

semantic mapping strategy enhanced the high-proficiency students’ recall

Table 23

Summary of LSD Pairwise Comparisons for the Written Recall Protocols of the Two High-proficiency Groups (total score for each article= 100)

(I) Means of Exp. Group

(J) Means of Ctrl. Group

Std. Error Mean Difference (I-J)

Article 1 73.56 72.16 1.36 1.40

Article 2 58.31 47.97 2.70 10.34**

Article 3 43.97 35.44 2.33 8.53*

* p < .05 ** p < .01

performance after reading Article 2 and Article 3.

As to the low-proficiency groups’ recall protocols, the results in Table 17 showed a significant overall effect of treatment. Besides, the subsequent LSD Pairwise

Comparisons (see Table 24) revealed LE did not significantly outscored LC after reading Article 3 (p > .05). But LE performed better than LC in Article 2, and a significant statistical difference was found (Mean difference= 9.37, p < .01). That is, semantic mapping strategy only helped the low-proficiency students recall more idea units of Article 2, the well-organized text.

Table 24

Summary of LSD Pairwise Comparisons for the Written Recall Protocols of the Two Low-proficiency Groups (total score for each article= 100)

(I) Means of Exp. Group

(J) Means of Ctrl. Group

Std. Error Mean Difference (I-J)

Article 1 60.44 59.28 1.57 1.16

Article 2 38.84 29.47 2.37 9.37**

Article 3 23.75 19.19 1.78 4.56

** p < .01

Concerning the multiple-choice comprehension tests, the results presented in

Table 19 showed that treatment had a significant overall effect on the multiple-choice

comprehension tests of the two low-proficiency groups (p < .05). Table 25, the

subsequent LSD Pairwise Comparisons, showed that HE significantly outscored HC

in the multiple-choice comprehension tests of Article 2 and Article 3 (p< .01).

Table 25

Summary of LSD Pairwise Comparisons for the Multiple-choice Comprehension Tests of the Two High Proficiency Groups

(total score for each article= 100) (I) Means of

Exp. Group

(J) Means of Ctrl. Group

Std. Error Mean Difference (I-J)

Article 1 73.94 68.66 2.33 5.28

Article 2 80.47 67.97 3.07 12.50**

Article 3 75.63 62.50 2.58 13.13**

** p < .01

Accordingly, semantic mapping strategy facilitated the high-proficiency students to comprehend both well-organized and less-organized texts. As to the low-proficiency students’ performance on the multiple-choice comprehension tests, the results in Table 21 showed that no significant effect was found on Article 2 and Article 3 (p > .05).

That is, semantic mapping strategy did not help the low-proficiency students comprehend either the well-organized text (Article 2) or the less-organized text (Article 3).

Based on the results discussed above, semantic mapping strategy helped the high-proficiency students to recall and comprehend both well-organized and

less-organized texts, but it only facilitated low-proficiency students to recall the idea units of the well-organized text.

Students’ Attitudes toward Semantic Mapping Strategy

This section presented the experimental group’s responses to the instruction that incorporated semantic mapping strategy. Their results on the post-treatment

questionnaire and their opinions elicited from the interviews are reported below.

Responses on the Post-treatment Questionnaire

The questionnaire contained ten items. Items 1 to 7 were used to evaluate students’ attitudes toward the semantic mapping strategy training in reading class.

Items 8 to 10 aimed to investigate students’ responses toward their independent

application of semantic mapping strategy in Practice Phase. Students’ responses were described and discussed as the positive responses or the negative ones. The former included “strongly agree” and “agree” responses. The latter contained responses of

“disagree” and “strongly disagree.” Table 26 presents the frequency and a percentage distribution of participants’ responses. The means (M) and standard deviation (SD) between the two different proficiency levels are included.

Concerning the participants’’ attitudes toward the semantic mapping strategy training in the reading class (see Table 26, Items 1 to 7), Item 5 received the highest score (M = 4.18), then Item 2(M = 4.08), Item 6 (M = 4.05), Item 1 (M = 3.98), Item 3 (M = 3.83), and Item 7 (M =3.72). Item 4 scored the lowest (M = 2.97). To be more specific, thirty-one HE students (97%) and twenty-seven LE students (85%) agreed on the effect of semantic mapping strategy on their understanding of text structure (Item 5); no students disagreed. With regard to whether the semantic map gave them a general idea of the content before reading (Item 2), twenty-nine HE students (91%) and twenty-eight LE students (88%) agreed and no one disagreed. Besides, fifty-two (H: 27, 84% & L: 25, 78%) students agreed that the semantic map helped them remember the content (Item 6), and only 3% of the HE students disagreed. More than eighty percent of HE students (81%) and LE students (85%) expressed that semantic mapping strategy helped them comprehend the text in the reading class (Item 1). Only one HE student (3%) and one LE student (3%) did not think so. More HE students (81%) than LE students (65%) thought that the semantic map helped them predict the content of the text (Item 3), but one HE student (3%) and one LE student (3%) did not.

As to whether they liked their teacher to use semantic mapping strategy to teach

reading (Item 7), twenty-four HE students (76%) and twenty-two LE students (69%)

expressed positive responses, while two HE students (6%) and two LE students (6%)

did not like the instruction. Concerning vocabulary learning (Item 4), only twelve HE

Table 26

Summary of Descriptive Statistics of the Participants’ Responses to the

Post-treatment Questionnaire (N= 32 for each proficiency level)

Items 5

4 3 2 1 M SD

High 8(25)^b 18(56) 5(16) 1(3) 0(0) 4.03 .74 Low 4(13) 23(72) 4(13) 1(3) 0(0) 3.94 .62 1. Using the semantic mapping

strategy in the reading course helps me comprehend

the text. _{Total 12 41 9 1 0 3.98 .68}

High 8(25) 21(66) 3(9) 0(0) 0(0) 4.16 .57 Low 4(13) 24(75) 4(13) 0(0) 0(0) 4.00 .51 2. The teacher-student

interactive map used in class helps me understand the content of the text briefly

before reading. ^{Total 12 45 7 0 0 4.08 .54}

High 6(19) 20(63) 5(16) 1(3) 0(0) 3.97 .69 Low 2(6) 19(59) 10(31) 1(3) 0(0) 3.69 .64 3. The teacher-student

interactive map used in class assists me to predict the content of the text.

Total 8 39 15 2 0 3.83 .68

High 2(6) 10(31) 9(28) 11(34) 0(0) 3.09 .96 Low 1(3) 8(25) 8(25) 15(47) 0(0) 2.84 .92 4. The teacher-student

interactive map used in class assists me to guess and learn the new words in the text.

Total 3 18 17 26 0 2.97 .94

High 12(38) 19(59) 1(3) 0(0) 0(0) 4.34^c .55 Low 6(19) 21(66) 5(16) 0(0) 0(0) 4.03 .59 5. The teacher-student

interactive map used in class helps me understand the

structure of the text. _{Total 18 40 6 0 0 4.18 .59}

High 10(31) 17(53) 4(13) 1(3) 0(0) 4.13 .75 Low 6(19) 19(59) 7(22) 0(0) 0(0) 3.97 .65 6. The teacher-student

interactive map used in class helps me memorize the

content of the text. _{Total 16 36 11 1 0 4.05 .70}

High 4(13) 20(63) 6(19) 2(6) 0(0) 3.69 .78 Low 4(13) 18(56) 8(25) 2(6) 0(0) 3.75 .76 7. I like my teacher to use

semantic mapping strategy in the reading class.

Total 8 38 14 4 0 3.72 .77

High 2(6) 21(66) 9(28) 0(0) 0(0) 3.78 .55 Low 2(6) 17(53) 10(31) 3(9) 0(0) 3.56 .76 8. In Practice Phase, creating

the map by myself after reading the text facilitates me

to comprehend the text. _{Total 4 38 19 3 0 3.67 .67}

High 4(13) 20(63) 7(22) 1(3) 0(0) 3.84 .68 Low 5(16) 13(41) 11(34) 2(6) 1(3) 3.59 .95 9. In Practice Phase, creating

the map by myself after reading helps me answer the

multiple-choice questions Total 9 33 18 3 1 3.72 .83 High 5(16) 21(66) 5(16) 1(3) 0(0) 3.94 .67 Low 4(13) 17(53) 8(25) 3(9) 0(0) 3.69 .82 10. In Practice Phase, creating

the map by myself after reading helps me recall the content of the text

Total 9 38 13 4 0 3.81 .75

a 5 = strongly agree, 4 = agree, 3 = no idea, 2 = disagree, 1 = strongly disagree

b The number in parenthesis is the percentage and has been rounded to the nearest whole number

c The numbers in bold indicated a significant difference existed, p = .32 < .05.

students (37%) and nine LE students (28%) agreed on the helpfulness of the semantic map in guessing and learning the unknown words; but eleven HE students (34%) and fifteen LE students (47%) did not.

As to the participants’ responses toward their independent application of semantic mapping strategy in Practice Phase (see Table 26, Items 8 to 10), Item 10 received the highest score (M = 3.81), then Item 9 (M = 3.72), and Item 8 (M= 3.67).

In more details, twenty-six HE students (81%) and twenty-one LE students (65%) expressed that semantic mapping strategy helped them recall the content of the text (Item 10), but one HE student (3%) and three LE students (9%) expressed a negative attitude. With respect to whether semantic mapping strategy helped them answer the multiple-choice questions (Item 9), twenty-four HE students (76%) and eighteen LE students (57%) considered semantic mapping strategy effective, but one HE student (3%) and three LE students (9%) did not. Besides, twenty-three HE students (72%) and nineteen LE students (59%) thought that their independent application of

semantic mapping strategy in the testing period facilitated their comprehension (Item 8), while three LE students (9%) did not think so.

Based on the findings mentioned above, the students in general expressed positive attitudes toward the semantic mapping training and application. They considered semantic mapping strategy effective to improve their knowledge of text structure and promote their reading comprehension.

In order to see if any significant differences existed between the two proficiency levels in their responses to each item of the post-treatment questionnaire, an

independent-sample t-test analysis was conducted. As can been seen in Table 20, the

means of each item between the two proficiency levels were much the same (no

significant differences existed) except for responses to Item 5 (p < .05). More

high-proficiency students than the low-proficiency students (31 vs. 27) thought the

semantic mapping training promoted their understanding of text structure, and more low-proficiency students expressed “no idea” (5 vs. 1).

Findings from Interviews

One week after Practice Phase, an interview was conducted to sixteen students chosen randomly from the experimental group, eight from each proficiency level.

Their opinions toward semantic mapping instruction were analyzed and categorized as the positive opinions and negative ones. The results are described below.

When asked “Does the application of semantic mapping strategy introduced in reading class facilitate your reading comprehension?” (Question 1), twelve

students (H: 7, L: 5) thought that semantic mapping strategy facilitated their reading comprehension. The positive opinions were categorized and coded as follows. (1) Semantic mapping strategy helped them find the key words, topic sentence, and main idea. (2) The content analyzed and categorized by semantic mapping strategy was easier to understand and memorize. (3) It was much easier to memorize the graphic form than the words. (4) Semantic mapping strategy helped them understand the structure (organization) of the texts, which made them comprehend the content better, and have a general idea of writing. (5) The semantic mapping strategy activity made them more attentive in class and improved their understanding of the content.

Four students (H: 1, L: 3) did not consider semantic mapping strategy helpful in comprehending a text. Their negative opinions included (1) Semantic mapping strategy was not helpful because it was different from the way students decoded and memorized information from a text. (2) Semantic mapping strategy was not helpful for students because they did not understand the meanings of the unknown words, phases, and the whole sentence. (3) It was time-consuming to organize a map.

When asked “When does semantic mapping strategy help you most, in

pre-reading, during-reading or post-reading stage?” (Question 2), six students

thought that semantic mapping strategy helped most when it was used in the

pre-reading stage; three students favored the application in the during-reading state;

four students chose the post-reading stage; and three students preferred both

pre-reading and post-reading stages. Their reasons were summarized below. (1) When used in pre-reading stage, the semantic map gave students a general idea of the

content and also helped them predict the content. It also helped students find out the key words and the main idea of the content. Besides, it was helpful in analyzing and categorizing (organizing) the content. (2) When used in during-reading stage, the semantic mapping activity made students focus on the important information and be more attentive in the reading class. (3) When used in post-reading stage, the semantic map made students pay attention to text structure, the important parts of the content, rather than minor details. It helped students recall and memorize the content logically and systematically.

When asked “Do you like the teacher to use semantic mapping strategy in the reading class?” (Question 3), thirteen students showed positive attitudes toward

using semantic mapping strategy in reading class. Their reasons were similar to their opinions toward Question 1: Semantic mapping strategy is helpful in improving their knowledge about text structure, memorizing the content, and making reading class more interesting.

Three students did not like the semantic mapping activity because it was time-consuming and it was tedious and useless when the text was easy. Besides, semantic mapping strategy interfered with their original reading methods.

When asked “Is it difficult for you to organize a semantic map after reading an article in Practice Phase? What is difficult for you?” (Question 4), only three

students from the high-proficiency level did not think it difficult to organize a map

after reading an article in Practice Phase. Thirteen students considered their

independent application in Practice Phase a tough task. Their reasons included (1) Time limitation: If they had enough time to read and create a map, it would not be difficult. (2) Vocabulary difficulty (especially for low-proficiency students): There were many words they did not know in Article 2 and Article 3, so it was difficult to understand the content and create a map. (3) Difficulty level of the text: It was difficult for students to analyze and organize the structure of the content (especially the supporting details) when the text was difficult and long (Article 3).

When asked “Will you use semantic mapping strategy in other subjects?”

(Question 5), three students did not want to use semantic mapping strategy in other subjects because it was not helpful for them in the reading class and they did not think it would be useful when used in other subjects. Besides, it was time-consuming.

Thirteen students would like to use semantic mapping strategy in other subjects, such as history and geography because semantic mapping strategy helped them to distinguish important information from unimportant details. Moreover, semantic mapping strategy, as note-taking and outlining skills would help them to organize the content more logically, which would make it easier for them to understand and remember the content.

When asked “Is semantic mapping instruction helpful to you? If yes, what have you learned from the instruction?” (Question 6), three students did not

consider semantic mapping instruction helpful to them. The other students (thirteen

students) all agreed on the helpfulness of semantic mapping instruction. They thought

that semantic mapping strategy was an effective reading technique, from which they

knew how to analyze and categorize the content briefly; how to organize the text

structure and used it to comprehend the text; and how to find the key points and

distinguish them from unimportant details.

Discussion

This section aims to further discuss the results of the study to explore the effects of semantic mapping strategy on EFL high school students’ reading comprehension.

The discussion will be presented following the research questions proposed in Chapter One.

Effects of Semantic Mapping Strategy on Reading Comprehension

The results of the recall protocols (see Table 9 and Table 10) and the

multiple-choice comprehension tests (see Tables 11 to 13) demonstrated that semantic mapping strategy did facilitate students’ reading comprehension. Such results support the previous studies, in which the students receiving mapping training outperformed the students receiving no mapping training (Alvermann, 1981; Berkowitz, 1986;

Carrell et al., 1989; Guastello, et al., 2000; Jau, 1997; Pappa et al., 2003).

Visual organization of the semantic map coincides with the ideas of schema theory, in which concepts and events are stored in human memory on the basis of semantic relationships. Thus, it is not counterintuitive to find out that semantic mapping strategy facilitates students to activate their existing content schemata, build up the new schemata, and to remember the content. Besides, the existing schemata (background knowledge) presented in the map also help students to make reasonable predictions about the content and confirm the predictions from reading, which is considered important skills in improving reading ability (Reutzel, 1986; Smith, 1985).As a result, the experimental group outscored the control group in the reading comprehension tests in this study. Besides, students’ responses to the questionnaire also indicated that they considered semantic mapping strategy helpful for them to predict the development of the content and learn the content.

In addition, students have to identify the main idea first when constructing a map.

Then, they analyze the content, determine (select) the second categories, the

supporting details, and organize the map. Finally, they have to evaluate and connect the new and the known information. Such a mapping process incorporates Mayer’s cognitive process and improves text comprehension (Armbruster, et al., 1991; Jau, 1997). Also, the semantic mapping training increases students’ sensitivity to text structure. This awareness of textual organization has been proved as one of the critical factors influencing reading comprehension (Carrell, 1992; Chen, 1990; Meyer, 1975;

Meyer et al., 1980). Accordingly, semantic mapping strategy improved students’

knowledge of text structure and led them to better understand difficult texts. Besides, students’ responses also showed their acknowledgement of the importance of text structure in comprehending texts in the present study. Majority of the students held positive attitudes toward the effects of semantic mapping strategy on their

understanding of text structure. Out of sixteen students selected for the interview, thirteen students regarded semantic mapping strategy as a useful technique to help them identify the main idea and text structure and to organize the content. Such findings also agree with many previous studies (Armbruster, Anderson, & Ostertag, 1987; Boothby and Alvermann, 1984; Guastello, Beasley, & Sinatra, 2000; Raphael, Englert, & Kirschner, 1986; Reutzel, 1986).

Also, to create a map independently after reading, students in the experimental group had to engage themselves in the reading process actively in the present study.

That is, students needed to identify the main idea, text structure, analyze, and organize

the content, which might result in their better text comprehension. Such results are in

agreement with Berkowitz’s (1986) point of view that the map strategy would be

more effective when students constructed the map independently. Moreover, students

also pay more attention to the salient and relevant information when conducting the

map, which might help to build up new schemata and contribute to the experimental

group’s superior performance. Based on students’ opinions in the interview, thirteen

students expressed that semantic mapping strategy helped them become more

attentive in class and improve their reading ability. It is consistent with Lipson’s (1995) conclusion that semantic mapping strategy kept students on their task so they could comprehend more information.

Effects of Semantic Mapping Strategy on Different Proficiency Levels

According to the results shown in Tables 14 to 21, semantic mapping strategy had a significant overall effect on high-proficiency students’ recall protocols and multiple-choice comprehension; while it only had a significant overall effect on the low-proficiency students’ recall protocols. Clearly, high-proficiency students

benefited more from semantic mapping instruction than low-proficiency students. The finding contradicts with some previous studies, in which semantic mapping strategy were helpful to improve low-achieving students’ reading comprehension (Avery &

Baker, 1997; Guastello, Beasley, & Sinatra, 2000; Lipson, 1995). It is partially

contrary to Alvermann’s (1981) finding, in which students at both the upper and lower levels benefited from the use of mapping strategy.

Compared with the low-proficiency L1 participants in the previous studies

(Alvermann, 1981; Avery & Baker, 1997; Guastello, Beasley, & Sinatra, 2000; Lipson,

1995), these low EFL proficiency participants in the present study were not familiar

with English text structure. Thus, their limited vocabulary size and insufficient

linguistic knowledge contributed to the lack of facilitative effects of semantic

mapping strategy on the low-proficiency students’ reading comprehension in the

present study. Even though Chinese glosses of important words were provided in each

article used in Practice Phase, the three articles, especially Article 2 and Article 3,

still contained too many unknown words and unfamiliar sentence patterns for the L2

low-proficiency students to comprehend. Without adequate linguistic knowledge and

understanding of text structures, these low EFL proficiency students may not be able

to make use of the available contextual clues to figure out the meanings of these unknown words and make sense of the sentence patterns. It is hard for them to understand the content, let alone to create a map. Laufer (1997) argues that reading strategies, especially the top-down reading strategies, can’t work well when the English learners’ vocabulary size is below a threshold level, namely, 3,000 word families or 5,000 lexical items (Chen, 2004). As a result, semantic mapping strategy did not enhance low-proficiency learners’ reading comprehension in the present study.

Also, limited time in the testing stage may be another factor affecting the low-proficiency students’ performance. In Practice Phase, the experimental group needed to read an article, create a map, and then take the reading comprehension tests, including recall protocols and multiple-choice questions within 30 to 45 minutes. For low-proficiency students, such time limit seems to hinder their comprehension and performance. Roshan and Dwyer’s (1998) viewpoint that for mapping strategies to work, students might need to process at their own pace has found support here.

The low-proficiency students selected for the interview (8 in total) all also expressed that mapping task in Practice Phase was tough for them because they encountered too many unknown words and did not have enough time to read and create a map. Obviously, time is an important factor for semantic mapping activity to show effects.

Effects of Difficulty Level and Structure of a text on Semantic Mapping Strategy

Based on the results presented in Table 13 and Table 22, the experimental group

did not significantly outperform the control group after reading an easy article (Article

1), but significantly outscored the control group in two more challenging articles,

Article 2 and Article 3, on both recall and multiple-choice comprehension tests. The

results indicated that the difficulty level affected the effects of semantic mapping

strategy on reading comprehension.

Difficulty level of a text can also explain why the experimental group did not significantly outperform the control group in Article 1, an easier text. Article 1

(Flesch-Kincaid readability grade level: 6.3) is easy for senior high school students to comprehend, while Article 2 and Article 3 (Flesch-Kincaid readability grade level: 8.0 and 7.6 respectively) are challenging for them. Students’ opinions gathered from the interview were consistent with this classification. That is, they also considered Article 1 to be easy but Article 2 and Article 3 more difficult.

One advantage of semantic mapping strategy is to promote readers’ knowledge and awareness of text structure, which is a critical factor in comprehending a text.

Roller (1990) argues that text structure is an essential factor when comprehending

“moderately difficult texts” but it is less important if the language and concepts embedded in the text are simple. It is obvious that the results shown in Table13 and Table 22 are in agreement with Roller’s viewpoint that semantic mapping strategy (the visual display of text structure) may be superfluous for students in

comprehending an easy text (Article 1), but is helpful to comprehend the moderately difficult texts (Article 2 & Article 3). In other words, the difficulty level of a text influences the effects of semantic mapping strategy on reading comprehension. The easier the text is, the less helpful semantic mapping strategy will be to

comprehension.

The results presented in Table13 and Table 22 also indicated that semantic mapping strategy helped the experimental group to comprehend both the

well-organized text (Article 2) and the less-organized text (Article 3). Such results are

different from those yielded from Alvermann’s (1981) study, which showed that

mapping strategies did not enhance students’ comprehension of the comparison texts

better than the descriptive texts; the former was considered more organized. The

possible explanation for the difference is the participants’ formal schema, that is, their

knowledge of text structure. The participants in Alvermann‘s (1981) study are tenth-grade L1 learners, who are familiar with English text structure. Accordingly, mapping strategies are useless when these L1 students can use their existing formal schemata to comprehend the text. However, the participants in the present study are tenth-grade EFL learners with less knowledge of English text structure. Thus, the semantic map provides the EFL students with the visual organization of the text, which helps them to make use of the top-level text structure easily and facilitates their comprehension.

In addition, Article 2 is more organized and therefore easier for students to be aware of its top-level text structure and comprehend the article better (McGee, 1982;

Meyer, Brandt, and Bluth, 1980; Taylor, 1980). This may account for the experimental group’s superior performance in Article 2 (shown in Table 22). Their performance of recall protocols on Article 2 was better than that on Article 3 (p < .01 vs. p < .05).

Besides, the results shown in Tables 23 to 24 indicated that both high- and low-proficiency experimental subgroups significantly performed better than the control group in recall protocols of Article 2 (well-organized). These findings support the results yielded from other studies. That is, when comprehending well-structured texts, readers who are aware of structure have better reading comprehension and recall more information than those who are unaware of it (Berkowitz, 1986; Carrell, 1985, 1992; Spires, et al., 1992).

In brief, difficulty level makes a difference in treatment effect for both proficiency groups as a whole. That is, the experimental groups significantly

performed better after reading two more challenging articles, Article 2 and Article 3 (see Table 13 and Table 22). But text structure makes a difference only for

low-proficiency group. Though semantic mapping strategy facilitates learners’

comprehension on both well-and less-organized texts, the effects will be better and

more distinct when comprehending a well-organized text, especially for the low-proficiency students (see Tables 23 to 24).

Students’ Attitudes toward Semantic Mapping Strategy

Students’ responses toward semantic mapping instruction were solicited through their responses on the post-treatment questionnaire and the interview survey. The responses to the questionnaire showed that the majority of the students in the mapping group held positive attitudes toward the training of semantic mapping strategy and considered the training effective to improve their reading comprehension. Namely, the semantic mapping training activated their content schemata, helped them make

reasonable prediction, improved their knowledge and awareness of text structure, and facilitated their comprehension. The findings are in agreement with other previous studies (Armbruster et al., 1987; Jau, 1997; Lipson, 1995; Sinatra et al., 1984). In general, over seven-tenths of the high-proficiency students and near seven-tenths of the low-proficiency students liked the semantic mapping training in the reading class.

When it comes to the effect of the semantic mapping training on vocabulary learning (Item 4), however, the majority of students didn’t think the training was helpful to their guessing and learning of unknown words. This result contradicted with previous studies (Heimlich & Pittelman, 1986; Johnson & Pearson, 1984).

However, 97% of the high-proficiency students and 85% of the low-proficiency students considered semantic mapping strategy effective to improve their awareness of the text structure (Item 5). This may be a result of the instructional focus of semantic mapping in the present study. The semantic mapping training in the present study emphasized the identification of main idea, secondary categories, and

supporting details. It also focused on activating the background knowledge, the

structure of text, and the relationship among different concepts. Seldom was the

attention drawn on vocabulary learning explicitly. This may be the reason why most

of the students disagreed that the training had effects on their guessing and learning of unknown words, whereas the majority of them agreed on its effects on the

understanding of text structure.

In addition, fewer low-proficiency students than high-proficiency students considered the training helpful in promoting their knowledge and awareness of text structure (Item 5, p < .05). The result may be attributed to students’ proficiency level.

Because of low-proficiency students’ deficiency in vocabulary, they may focus more on decoding the meaning of unknown words than the identification of main idea and important details when reading. This may influence their understanding of the text structure.

As to students’ responses to the effects of semantic mapping strategy on the reading comprehension tests, most of the students considered semantic mapping strategy useful to answering the multiple-choice questions and recalling the content of the article. The findings are in accordance with the previous studies (Jau, 1997;

Lipson, 1995; Reutzel, 1986; Sinatra et al, 1984).

Students’ opinions elicited from the interview also showed that they considered the training useful to their reading comprehension. Semantic mapping strategy made them focus on the main ideas, key information, and the relationship among them. In addition, semantic mapping strategy helped them understand the structure of a text by analyzing and organizing the content. Compared with a verbal picture, such visual representation of information is easier for students to remember. What’s more,

semantic mapping activity made the reading class more interesting. Such findings also support some previous studies (Avery & Baker, 1997; Guastello, Beasley, & Sinatra, 2000; Jau, 1997; Kuo, 2003; Lipson, 1995), which all conclude that semantic mapping activity can be an asset in reading class. However, students’ opinions from the

interview revealed that semantic mapping instruction may be time-consuming, both in

training and in their independent application. The mapping strategy would also be tedious and superfluous when the text is too easy. These findings are in agreement with other previous studies (Armbruster & Anderson, 1980; Holley and Dansereau, 1984; Roller, 1990).

When it comes to the difficulties that the students encountered when they created a map, most of the students (especially the low-proficiency students) expressed that they had difficulty understanding the unknown words in the articles and therefore it was difficult for them to comprehend the content and create a map. It seems that semantic mapping strategy can not work well when the students encounter too many unknown words. Laufer’s (1997) view that reading strategies, especially the top-down strategy, can not work well if the students’ vocabulary is limited is well-noted here.

Thus, teachers should take students’ vocabulary size into consideration when introducing semantic mapping strategy to low EFL proficiency students.

In short, this study found that semantic mapping strategy was an effective

reading strategy for senior high school students in Taiwan and should be introduced in class. Students’ responses and opinions from the questionnaire and the interviews not only substantiated the effects of semantic mapping strategy on EFL reading

comprehension, but also provided precious suggestions for future adaptation.