The comprehension test scores of the two given texts are showed in Table 6.

CHAPTER FOUR RESULTS

This chapter presents the results from this study. Section 4.1 presents the results of the reading comprehension tests on the two articles of different levels of difficulty.

Section 4.2 presents the results of the subjects’ vocabulary retention tests. Section 4.3 explores the subjects’ lookup behavior. Section 4.4 probes the correlation among the variables. Section 4.5 presents the participants’ perceptions of using the online dictionary consultation system.

4.1 Results of Reading Comprehension

The comprehension test scores of the two given texts are showed in Table 6.

Table 6. Mean Scores and Standard Deviations (SD) for Comprehension Tests

Group

Text (H) Mean (SD)

Text (L) Mean (SD)

Grand Means O.D.

3.13 (1.08) 4.41(1.37) 3.77 P.D.

2.97(1.53) 3.87(1.84) 3.42

N.D.

2.95(1.67) 3.74(1.62) 3.35

Grand Means

3.02 4.01

Note: 1. Group (OD): online dictionary, Group (PD): conventional paper dictionary, Group (ND): no

dictionary access

2. Text (H)—the high difficult level text, Text (L)—the low difficulty level text

3. Comprehension test: 7 multiple choices, full mark=7

A two-way ANOVA was used to assess the effects of the independent variables, with three levels, online dictionary, conventional paper dictionary and no dictionary access, and text difficulty on the dependent variable, text comprehension.

Table 7. A two-way (2x3) ANOVA on Comprehension Test Scores

Source SS df MS F p

Between groups

Text difficulty 57.50 1 57.50 24.34 .000 Dictionary Check 7.92 2 3.96 1.67 .189 Text x Dictionary 2.57 2 1.28 .54 .581

Error 538.46 228 2.36

Total 3494.000 233

Although the dictionary consultation groups got higher scores on the comprehension tests than did the no-dictionary consultation group, there was no significant difference among groups on reading comprehension tests for each of the two texts. However, text difficulty significantly affected the subjects’

comprehension performance (F=24.34, p =.000 ＜.001). The subjects performed better on the easy text (grand mean of Text L was 4.01) than on the difficult text (grand mean of Text H was 3.02). In the study, no interaction effect between the text difficulty and dictionary check was found.

4.2 Results of Vocabulary Learning

One of the research questions is to find out whether the subjects can acquire vocabulary in reading. In this section, the results of the vocabulary retention tests are presented. Table 8 presents the mean scores the three reading groups received on the immediate vocabulary posttest.

Table 8. Mean Scores of Immediate Vocabulary Posttest

Group

Text (H) Mean (SD)

Text (L) Mean (SD)

Grand Means O.D. 6.62

(2.02) 8.35

(2.41)

7.49

P.D. 6.02

(2.65) 7.88

(1.88)

6.95

N.D. 5.30

(2.05) 5.81

(2.51)

5.56

Grand Means

5.98

7.34

Note: 1. a. means the adjusted model using vocabulary pretest as covariate 2. full mark of the vocabulary posttest =12

From Table 8, we found that the grand mean of Group (OD) was 7.49, higher than 6.95 of Group (PD) and 5.56 of Group (ND). The grand mean of Group (PD), 6.95, was higher than Group (ND), 5.56. The grand mean of Text (L), 7.34, was higher than that of Text (H), 5.98.

A two-way (2x3) ANCOVA was performed to examine whether there was a

significant difference among groups on immediate vocabulary posttests. The two

independent variables were dictionary check and text difficulty. The dependent

variable was immediate vocabulary posttest and the covariate was the vocabulary

pretest. The results were shown in Table 9.

Table 9 ANCOVA on Immediate Vocabulary Posttests

438.77 6 73.13 16.82 .000

2161.06 1 2161.06 496.90 .000

189.33 1 189.33 43.53 .000

152.92 2 76.46 17.58 .000

120.30 1 120.30 27.66 .000

22.55 2 11.28 2.59 .077

987.23 227 4.35

11826.00 234

Source Adjusted model Intercept Covariate GROUP TEXT

TEXT * GROUP Error

Total

SS df MS F P

According to the result of Table 9, we found that the dictionary check had a significant effect on the immediate vocabulary posttest (F = 17.58, p = .000 < .01). In other words, there was a significant difference on immediate vocabulary posttest among groups. Second, grand mean of Text (H) was 5.98 and grand mean of Text (L) was 7.34. As we can see from Table 9, text difficulty also had a significant effect on the immediate vocabulary posttest (F = 27.66 , p = .000 <.01). Third, there was no interaction effect between text difficulty and dictionary check (F = 2.59, p = .077

>.05).

Table 10 Post Hoc Scheffe Comparison among Groups on Immediate Vocabulary Posttests

Text (H) Text (L) df F p df F p Between groups 2 3.781 .026 2 14.694 .000

Within groups 114 114

From Table10, it is found that there was significant difference among the three reader groups (p

= .0 26＜ .05, and p

= .000 ＜.05).

Though no significant interaction effect between dictionary check and text difficulty on the immediate vocabulary posttest was found, there seemed to be a slight interaction effect (F= 2.59 , p=.077, almost close to p < .05). Thus, the Post Hoc Scheffe was performed to compare between groups.

Table 11. Post Hoc Scheffe Comparison between Groups Group

I J

Mean Difference(I-J)

Std.

Error Sig. 95% Confidence Interval Lower Bound Upper Bound Text (H) OD PD

ND

.60 1.32*

.46 .47

.20 .01

-.32 1.51 .39 2.24

PD ND .721

.47

.12

-.20 1.64

Text (L) OD PD ND

.47 2.54*

.49 .49

.341 .000

-.51 1.44 1.57 3.51 PD ND

2.07* .48

.000

1.11 3.02

Note: 1. Group (OD): online dictionary, Group (PD): conventional paper dictionary, Group (ND): no

dictionary access;

2. Text (H)—the difficult text, Text (L)—the easy text

We can see from Table 11 that on Text (L), there was a significant difference between Group (OD) and Group (ND), and between Group (PD) and Group (ND).

However, no significant difference was found between Group (OD) and Group (PD).

On Text (H), significant difference was only found between Group (OD) and Group

(ND). In conclusion, we didn’t find a significant difference between Group (OD) and

Group (PD) on vocabulary posttests of the two texts. In other words, although

dictionary check will significantly enhanced vocabulary retention, no significant

effect of different types of dictionary check on vocabulary retention was found in the present study.

A delayed vocabulary posttest was conducted two weeks after the immediate vocabulary posttest and the mean scores were presented in Table 12.

Table 12. Mean Scores of Delayed Vocabulary Posttest

Group

Text (H) Mean (SD)

Text (L) Mean (SD)

Grand Means O.D. 6.73

(2.22) 7.87

(2.07)

7.3

P.D. 5.91

(3.32) 7.48

(2.78)

6.69

Grand Means

6.32

7.67

Note: a. means the adjusted model using immediate vocabulary posttest as covariate

The grand mean of Group (OD) was 7.3, higher than 6.69 of Group (PD). The grand mean of Text (L) was 7.67, higher than 6.32 of Text (H).

A two-way (2x2) ANCOVA was performed to examine whether there was

significant difference between the two dictionary access groups on the delayed

vocabulary posttest, as shown in Table 13. The independent variables were dictionary

check and text difficulty. The dependent variable was the delayed vocabulary posttest

and the covariate was the immediate vocabulary posttest.

Table 13. ANCOVA on Delayed Vocabulary Posttests D

elayed ^Source

342.59 4 85.65 15.71 .000

127.58 1 127.58 23.40 .000

238.54 1 238.54 43.74 .000

14.97 1 14.97 2.75 .100

4.46 1 4.46 .82 .367

2.29 1 2.29 .42 .518

823.41 151 5.45

8810.000 156

Adjusted model Intercept Covariate GROUP TEXT

TEXT * GROUP Error

Total

SS df MS F P

.

From Table 13 we found some trends. First, there was no significant difference between the two groups on the performance of the delayed vocabulary posttest (F=

2.74, p = .100 ＞.05). In other words, the two dictionary access groups didn’t performed differently on the memory of the word learned while reading. Second, the text difficulty didn’t affect the subjects’ long term retention performance (F= .82, p =.367>.05). Third, there was no interaction effect between text difficulty and

dictionary types (F= .419, p = .518 >.05) on delayed vocabulary posttests.

In conclusion, the long term retention performance of vocabulary learning in the

two types of dictionary check groups didn’t show significant difference on the two

texts. Besides, the results of the present experiment did not suggest that readers

tended to use electronic glosses for short-term purposes, as the investigations of Black

et al. (1992) indicated. Instead, the findings of the present study indicated that words

learned through dictionary consultation could be long-term retention. In other words,

dictionary consultation not only had a significant effect on vocabulary gain but the retention could be long term memory as well.

4.3 The Results of Frequency of Lookup Behavior

The frequency of the unknown words looked up in the dictionary was also a topic that the present study explored. Table 14 indicated the frequency of unknown words looked up in the two dictionary check groups.

Table 14. Mean Number of the Words Looked up

Group

Text (H) Mean (SD)

Text (L) Mean (SD)

Grand Means O.D. 36.28(17.75) 22.49(11.34)

29.38 P.D. 8.13(4.32) 7.87(2.81)

8 Grand Means

22.2 15.18

Note: 1. Group (OD)—online dictionary, Group (PD)—conventional paper dictionary 2. Text (H)—the difficult text, Text (L)—the easy text

From Table 14, it was found that Group (OD) looked up about three to four

times more words than Group (PD), which corresponded with the study conducted by

Aust et al.(1993). Aust’s study also indicated that hyper-reference users consulted

over two times as many word definitions as conventional dictionary users. The online

dictionary apparently increased users’ likeliness for elaboration, for it provided users

with more efficient access. One online dictionary user made 65 unique consultations

out of the 270-word article (Text H), whereas the highest number of consultations in

the paper dictionary condition was 24.

Table 15 presented the two-way ANOVA (2x2) for the effect of text difficulty and reader groups on the subjects’ lookup behavior.

Table 15. A two-way ANOVA on Lookup Behavior

Source SS df MS F p

Between groups

Text difficulty 1925.026 1 1925.026 16.372 .000 Dictionary type 17834.769 1 17834.769 151.680 .000

Text x Dictionary 1787.077 1 1787.077 15.199 .000

Error 17872.359 152 117.581

Total 93926.000 156

From the Table 15, it was found that text difficulty significantly affected the lookup behavior(F= 16.37, p = .000 ＜.01). When the subjects read the different difficulty levels of texts, their lookup behavior presented significant difference. The grand mean of Text (H) was 22.2 and that of Text (L) was 15.18. When they read the difficult text (H), the frequency of lookup behavior significantly became higher than if reading the easier text (L).

Second, different types of dictionary use significantly affected the frequency of lookup behavior (F= 151.68, p = .000 ＜.01). The grand mean of frequency of words looked up in Group (OD) was 29.38 and that in Group (PD) was 8. Obviously, Group (OD) consulted three times more words than Group (PD). Namely, online lookup

system indeed significantly motivated the subjects to do more lookup behavior.

Finally, interaction effect between the text difficulty and dictionary types (F=15.20, p=.000<.01) also significantly affect the lookup behavior of the subjects.

Figure 1 Interaction between Dictionary Types and Text Difficulty on Lookup Behavior

group

paper dictionary online dictionary

Mean (the num ber of words looked up)

40

30

20

10

0

text

dificult easy

According to Figure 1, it was found that the fact that text difficulty had a

significant effect on the frequency of lookup behavior only existed in the online

dictionary group. It was not true for the paper dictionary group. We found that when

reading the difficult text, the online group obviously looked more words than if

reading the easier text. On the other hand, the paper dictionary group showed a

ceiling effect on the frequency of lookup behavior for the two texts of different

difficulty levels. In other words, the frequency of the words looked up on the two

texts of different levels of difficulty didn’t indicate differences for Group (PD). This

was mainly because the subjects in Group (PD) had to spend time flipping through

pages of the dictionary and writing down what they had consulted on the paper. The effort and reading time limit discouraged them from looking up the unknown words they encountered in the given texts.

In the paper-printed dictionary group, the subjects were required to write down on the paper the definition or translation of the unknown word they encountered while reading the text. The paper must be handed to the researcher before they took the comprehension test. From the paper collected, the researcher had to check if the subjects wrote down the correct meaning that was appropriate in the context. The number of correct and inappropriate definition or translation of the words was counted.

Table 16 indicated the percentage of the correct and of the inappropriate meanings of the words, which were consulted in the paper-printed dictionary group.

Table 16. Percentages of Correct and Inappropriate Meanings of the Words Consulted in the Paper Dictionary Group

Correct Meaning Inappropriate Meaning

Text (H) 87% 13%

Text(2) 91% 9%

From Table 16, we found that the percentage of the word meaning fitting the context

got lower on the difficult text, Text (H). It was implied that when the difficulty level

of the text was raised, it would become harder for the subjects to find the appropriate

meanings of the unknown words in the dictionary check.

4.4 Results of Correlation among the Variables

The researcher was also concerned about the correlation among variables.

Based on the data gained from the online dictionary group reading Text (L), the easier article, inspection of the correlation coefficient was shown in Table 17 and Table 18.

Table 17. Correlations Among the Dependent Variables for Group (OD)

N.of Words I.V. Posttest Comp. Test Number of Words Looked

up

1.00

.41**

.26 Immediate Vocabulary

Posttest

1.00

.62**

Comprehension Test 1.00

** p < .01

Table 18. Correlation Among the Dependent Variables for Group (PD)

N.of Words I.V. Posttest Comp. Test

Number of Words Looked up

1.00

.34 **

.06 Immediate Vocabulary

Posttest

1.00

.59 **

Comprehension Test 1.00

** p < .01

Inspection of the correlation coefficient (see Table 17 and Table 18) revealed some

trends. First, there was a significant positive correlation between the number of the

unknown words looked up and immediate vocabulary posttests. That is, the more

unknown words were consulted, the higher the immediate vocabulary posttest the

subjects scored. Second, there was a significant positive correlation between the immediate vocabulary posttest and the comprehension test. Namely, the higher score the subjects received on the vocabulary-retention tests, the higher score they received on the comprehension tests. Third, the study did not find correlation between the frequency of lookup and the comprehension test.

Inspection of the correlation coefficient among variables in Text (H) illustrated the same results (see Table 19).

Table 19. Correlation Among the Dependent Variables for Text (H)

N.of Words I.V. Posttest Comp. Test

Number of Words Looked up

1.00

.37 *

.14 Immediate Vocabulary

Posttest

1.000

.53**

Comprehension Test 1.00

* p < .05

** p < .01

4.5 Participants’ Perceptions of Using Online Learning System

After taking the reading task, the subjects in the online dictionary group had to do a post-study questionnaire. The questions on the questionnaire focused on the subjects’ response to the online learning program, The Online Digital English Learning Center. Table 20 illustrated the results of the subjects’ response to the

online reading system after they used it to help enhance text comprehension.

Table 20. Percentages Showing Participants’ Perceptions of Using Online Learning System

Opinion number (percentage) Statement

Positive Negative Neutral

I like the Online Learning System. 30 (77%) 5 (13%) 4 (10%)

The online dictionary consultation system motivates me to look up unknown words.

29 (74%) 8 (21%) 2 (5%)

The learning system helps me

enhance the text comprehension. 21 (54%) 8 (20%) 10 (26%)

I enjoy reading on the computer

screen. 13 (33%) 17 (43%) 9 (24%)

Online dictionary consultation is more beneficial than conventional paper dictionary consultation.

25 (64%) 13 (33%) 1 (3%)

(total number=39)

One of the topics the researcher intended to explore in the present study was

how the online dictionary consultation system influenced the subjects’ text

comprehension. A high percentage of the subjects (77%) thought that the online

learning program would benefit their language learning and enhance their interest in

learning English. Also, 74% of the participants thought the learning program would

motivate them to look up unfamiliar words. Based on the response of the subjects,

54% participants voted strongly for the positive effect of the online dictionary

consultation on the text comprehension. Although the majority of subjects agreed on

the efficiency of the online dictionary consultation system, 43% of the subjects held a

negative attitude toward reading a passage on the computer. They thought that

reading a passage presented on the computer screen would distract them from the

content of the article and would make their eyes tired. According to the questionnaire,

it was found that the subjects’ learning attitude toward the online learning system had

much to do with their performance on the reading comprehension test and vocabulary

retention test demonstrated in the present study.