Finally, the background information of the subjects is given in section 4.4

(1)

4.5 Summary

In this chapter, I have described the details of present research. In section 4.1, we know the

construction can either be present reference or past reference. The research questions are reiterated in section 4.2. And from section 4.3, we know that we have designed three tasks to study the learning of the target construction. Finally, the background information of the subjects is given in section 4.4.

Chapter 5

Results and Discussion

In this chapter, the findings of each research question will be presented and further interpreted in the context of SLA study. In section 5.1, the results of the three tasks that have been processed statistically will be given. And in section 5.2, the “were-subjunctive” will be discussed. Sections 5.3 and 5.4 are about the validity and limitations of this study. Finally, in sections 5.5 and 5.6, the discussion and conclusion will be given, followed by pedagogical implications and suggestions for future studies in sections 5.7 and 5.8.

5.1 Results

From section 5.1.1 to section 5.1.3, the research questions will be answered one by one.

(2)

After the statistic treatment, it is found that the results from Situation and Grammaticality Judgment Tasks are identical, both indicating that the counterfactual conditional (present reference) is the most difficult construction for the subjects while the result from Elicited Translation Task shows the counterfactual conditional (past reference) is the hardest one among the three constructions. Further, Error Analysis of Elicited Translation Task shows that the most frequent error is the lack of the backshifted tenses in both present reference and past reference. The second most frequent is the one which arises from the incomplete learning of the past perfective modals. In terms of error distribution, the most difficult part lies in the main clause of the past-referenced counterfactual conditional.

5.1.1The Result and the Discussion of the First Research Question

The Situation Task (Task One) attempts to answer our first research question: Do Chinese EFL learners at the high school level display difference in their ability to identify the contexts in which the non-counterfactual or the two types of counterfactual conditionals constructions should be used? If so, is there a hierarchy of degree of difficulty?

From Figure 5-1, it can be easily seen that the three groups of different achievement levels found it most difficult to identify the contexts which require the use of counterfactual

conditionals (present reference) and then the degree of difficulty declines from the counterfactual conditionals (past reference) to the non-counterfactual conditionals.

(3)

Based on the result of the treatment, we may propose two conclusions in terms of students’ ability to identify the contexts in which the non-counterfactual or the counterfactual conditionals constructions should be used. The hierarchy of the degree of difficulty is presented in Table 5-2.

1. The counterfactuals (present reference) are harder than counterfactuals (past reference).

2. The counterfactuals (present reference) are harder than non-counterfactuals.

Table 5-2: The hierarchy of difficulty degree from Task One

5.1.2 The Result and the Discussion of the Second Research Question

The Grammaticality Judgment Task (Task Two) is designed to determine whether the two kinds of counterfactual conditionals in English (present reference and past reference) syntactically constitute different levels of difficulty for Chinese EFL learners at the high school level. More specifically, does the backshifted tense principle or the “were-subjunctive”

cause problems? The mean scores of the three levels of achievers are displayed in Figure 5-2 below. The result is identical with the counterpart from Task One.

1. Counterfactual (present reference) > counterfactual (past reference) 2. Counterfactual (present reference) > non-counterfactual

(4)

5.15

4.74

3.86 4.74

4.38

3.57 4.96

4.59

4.24

0.00 1.00 2.00 3.00 4.00 5.00 6.00

high intermediate low

NCC PRESENT PAST

Figure 5-2: the mean scores of the three groups from Task Two

The statistical procedure, paired t-test, is repeatedly applied to find out if there is a statistical significance between two groups. The result is shown in Table 5-3.

Table 5-3: The result of the paired t-test from Task Two

mean scores SD df t value significance

Paired NCC-present .3537 1.0815 81 2.961 .004*

Paired NCC-past 2.439E-02 1.2568 81 .176 .861

Paired present-past -.3293 1.4745 81 -2.022 .046*

(5)

Based on the results in Figure 5-2 and Table 5-3, we come to the conclusion that for Chinese EFL learners at the high school level counterfactual conditionals (present reference) constitute a higher level of difficulty than counterfactual conditionals (past reference) and non-counterfactual conditionals. The result is presented in table 5-4.

Table 5-4: The hierarchy of degree of difficulty from Task Two

Let’s take a closer look at which clause of the two counterfactual conditionals constitutes the greatest difficulty for students. The mean scores shown in Table 5-5 suggest that the main clause of the counterfactual (present reference) is the hardest one for students since the clause obtains the lowest mean score among the four clauses. The overall mean scores are still consistent with the results of Tasks One and Two. The result is that counterfactual conditionals (present reference) are harder than the counterfactual conditionals (past reference).

Table 5-5: The mean scores of the different four clauses from Task Two 1. Counterfactual (present reference) > counterfactual (past reference) 2. Counterfactual (present reference) > non-counterfactual

(6)

5.1.3The Result and the Discussion of the Third Research Question

The two tasks above are primarily concerned with comprehension. In order to have a more comprehensive understanding of the target structure, the third research question we raise is how the errors that Chinese EFL learners at the high school level make would be categorized qualitatively and quantitatively, according to the related SLA studies.

The task designed for the answers of the question is fifteen Elicited Translation test items (Task Three). And all the data obtained from this task can be divided into quantitative and qualitative categories.

First, let’s take a look at the quantitative category. Similar to the presentation of the answers to the first two tasks, the mean scores and the result of paired t-test will be examined.

In terms of the mean scores, Figure 10 displays a result different from the results of the mean scores overall mean scores

Counterfactual (present reference)

(if-clause) 2.3049

(main-clause) 1.9878

4.2927

Counterfactual (past reference)

(if-clause) 2.0366

(main-clause) 2.5858

4.6224

(7)

previous two tasks. Counterfactual (past reference) replaces counterfactual (present reference) and makes the most difficult structure among the three for students to learn.

5.65 5.44

4.14 5.80

5.25

4.14 5.15

4.25

3.33

0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00

high intermediate low

NCC PRESENT PAST

Figure5-3: The mean scores of the three groups from Task Three Further paired t-test (Table 5-6) confirms that the counterfactual (past reference) is significantly harder than the counterfactual (present reference) and the non-counterfactual

conditionals.

Table 5-6: The result of the paired t-test from Task Three

mean scores SD df. t value significance

Paired NCC-present 3.049E02 1.1148 81 .248 .805

Paired NCC-past .8354 1.1278 81 6.707 .000*

Paired present-past .8659 1.3541 81 5.790 .000*

*P<. 05

Based on the outcome of this task, the hierarchy of difficulty is summarized in Table 5-7.

(8)

Table 5-7: The hierarchy of difficulty degree from Task Three

After the quantitative interpretation, two qualitative observations are made during the process of data analysis. First, it will be shown that the lack of the backshifted tenses is the most problematic for students. Then, second to the lack of the backshifted tenses, incomplete learning of the past perfective modals is also found to constitute a difficulty for students. Some examples of the incorrect forms collected from the Elicited Translation Task will be given to illustrate this point.

(I) The lack of the backshifted tenses:

The lack of the backshifted tenses proves problematic for the three levels of language achievers. The lack of the backshifted tenses can be found in the four categories of the two kinds of counterfactual conditionals. As the percentages shown in Table 5-8, the low

achievers’ group suffers from the lack of this principle more frequently than the intermediate and high achievers’ groups. The intermediate achievers’ group, likewise, suffers from the lack of this principle more frequently than the high achievers’ group. That may suggest that the lack of the backshifted tenses is the main source of difficulty for Taiwanese EFL students.

1. Counterfactual (past reference) > counterfactual (present reference) 2. Counterfactual (past reference) > non-counterfactual

(9)

Table 5-8: The percentage of the lack of the backshifted tenses in the four categories between the two types of counterfactual conditionals

(II) Incomplete learning of the past perfective modals:

The incomplete learning of the past perfective modals can be best reflected from the mean scores in Table 5-9. The ungrammatical sentences collected from the eighth test item in the Elicited Translation Task will be used as examples to illustrate this point.

假如他當時沒把錢丟掉，他就能買腳踏車。(test item 8)

*If he had not lost the money at that time, he would bought a bike.

*If he had not lost the money at that time, he would has bought a bike.

*If he had not lost the money at that time, he would have buy a bike.

From the answers of the eighth test item, obviously, our subjects did not encounter certain degree of difficulty with the past perfective modals (would/could/should/might + have + Past Participle). This observation can also be proven through statistics. Among the four mean

scores in Table 5-9, the main clause in the past-referenced counterfactual obtains the lowest score. That indicates that the complicated structure of past perfective modals required in the

High Intermediate Low Counterfactual (present reference)

(if-clause) 1.23% 5.88% 15.87%

(main clause) 4.93% 16.66% 41.26%

(if-clause) 8.64% 17.64% 25.39%

(main clause) 14.81% 33.33% 57.14

(10)

main clause constitutes the greatest difficulty.

Table 5-9: The mean scores among the four different clauses from Task Three²

In summary, the statistic treatment outcome from the Task Three (Elicited Translation Task) shows that the counterfactual conditional (past reference) was significantly the hardest among the three ones. From the result of this task, the lack of the backshifted tenses constituted the major problem, which was consistent with the findings of Ziegeler (1995) and Wang & Chang (1999). And for the two references, backshifted tenses are applied less in the past reference than in the present reference. A plausible explanation would be that students lack a clear idea of the usage of the past perfective tense and therefore they do not know that the past perfective is one step back to the past tense. Also, the result shows that high school students have more

2 The statistic numbers in the second column repeatedly indicated the same finding with the Grammaticality Judgment Task that the counterfactual conditional (past reference) is harder than (present reference).

mean scores overall mean scores

NCC conditional (if-clause) 2.3841

NCC conditional (main clause) 2.7927 5.1768

(if-clause) 2.7622

(main clause) 2.3841

5.1463

(if-clause) 2.5000

(main clause) 1.8110

4.311

(11)

difficulty mastering the modal system since most errors of this kind occur in the main clause of the counterfactual conditional (past reference). This result is not unexpected since the structure of this clause (would/should/could/might+ have +past participle) is more complex than the structure of the if-clause of the past reference (had +past participle).

In order to gain a more comprehensive understanding of the answer to the third research question, let’s turn our attention to the qualitative data after having analyzing the quantitative data, With the degree of difficulty predicted as medium by the Contrastive Analysis, the Elicited Translation Task shows that the most prevalent error, as mentioned earlier, is the lack of the backshifted tenses. The lack of backshifted tenses also can be of two kinds.

(I) The use of present tense as a substitute for past tense.

Take test items (5) and (14) in the Elicited Translation Task for example:

假如現在我有自己的公司，我將會娶 Mary。(test item 5)

*If I have a company of my own, I would marry Mary.

假如他有錢，他就能買它了。(test item 14)

*If he had money, he will buy it.

(II) The use of simple past tense as a substitute for past perfective tense.

For example:

假如當時我有自己的電腦，我早就是程式設計師了。(test item 6)

*If I had a computer of my own, I would have become a programmer.

假如他當時沒把錢丟掉，他就能買腳踏車。(test item 8)

(12)

*If he had not lost the money at that time, I would buy a bicycle.

Even though the lack of the backshifted tenses has been acknowledged as a common source of errors (Ziegeler, 1995；Wang & Chang, 1999), the finding of the present study is different from that of Wang & Chang (1999). They identify the following three causes as sources of errors: (a) the substitution of simple present for past perfective; (b) the substitution of simple past for past perfective; (c) a mixture of present and past tenses.

5.2 The Conventional Usage of the “Were-Subjunctive”

If students have learned to apply the backshifted tenses to denote the counterfactual expression, the past tense of the copula would be was or were. But conventionally we use the

“were-subjunctive,” regardless of the subjective-verb agreement in the sentence. This conventional usage makes the researcher hypothesize that the learning the “subjunctive”

would be much easier since many Chinese EFL students would make mistakes about the agreement between subjects and verbs. This hypothesis, in other words, is about whether the

“were-subjunctive” would cause less difficulty for learners because of its salient form.

Therefore, in the Grammaticality Judgment Task and in the Elicited Translation Task, two test items are respectively designed for two purposes: first to detect whether students would get confused between the “were-subjunctive” and the regular tense of copula; second, the comparative difficulty between the “were-subjunctive” and the past perfective tense of the copula (had + been). The results from the two test items are presented as follows.

(13)

In Grammaticality Judgment Task, the following two test items are designed to detect the possible confusion between the “were-subjunctive” and the regular tense of copula.

( ) If I had been a good student at that time, I will go to National Taiwan University. (test item 7)

( ) If you were a fish, the cat would eat you. (test item 9)

The correctness rates of the seventh test item (about had + been) and the ninth questions (about the “were-subjunctive”) are 72% and 88% respectively, suggesting the degree of difficulty of the past perfective form of the copula (had +been) is higher than the

“were-subjunctive.”³

The comparative difficulty between the “were-subjunctive” and the structure (had + been) is also detected through the two following Elicited Translation test items:

假如我是你的話，我不會那樣做。(test item 1)

If I were you, I would not do that.

假如我當時在家裡的話，我就會遇見 Tom。(test item 2)

If I had been at home at that time, I would have met Tom.

The data from the two test items repeatedly reveals the past perfective form of the copula (had +been) presents more difficulty than the “were-subjunctive.” The correction rate of the

3 This conclusion is contradictory with the overall one of the Grammaticality Judgment Task. But allow I remind you that this conclusion is more limited to the copula that appears in the if-clause and the number of the test items that the conclusions drawn from are also different (The number of test items in this section is 2 while the overall one is 12).

(14)

second question is 64%, while the first test item is 79% and furthermore, the result from the first question also suggests the students would not tend to use the regular copula tense was to replace the “were -subjunctive” with only 21% suffering from the misuse of was for were4.

Therefore, at this point, we can make two tentative conclusions: (a) the “were -subjunctive” is easier than the past perfective form; (b) the confusion between regular past tense was and the “were-subjunctive” is not obvious. In sum, our original hypothesis that the

“were-subjunctive” is comparatively easier to learn because of its salient form is partly supported

5.3 Two Methodological Concerns: Task Effect and Group Effect

In this section we will take a look at two important issues in SLA research. That is, task effect and group effect. Task effect is derived when there might be a bias caused by the one single experiment task. In this study, for the comprehension part, both Situation Task and Grammaticality Judgment Task demonstrate identical results. That is to say, task effect is not found in this research. On the other hand, group effect concerns the grouping of the subjects involved in the experimental tasks. In the study, the subjects are divided into three groups.

Paired t-tests show that statistical significances can be found between the groups. Therefore,

4 Even though the use of was is acceptable, as we mention in section 2.1.4.1 of Chapter 2, this usage is not included in the textbook used in high schools. Therefore, here we regard the use of was as a piece of evidence that subjects fail to apply “the were-subjunctive.”

(15)

we can affirm the existence of a group effect in this study and conclude that the grouping of the subjects is of methodological significance.

5.4 Discussion

The outcomes from Task One (Situation Task) and Task Two (Grammaticality Judgment Task) indicate that the counterfactual conditional (present reference) is significantly harder than the other two structures. While the outcomes from the Task Three (Elicited Translation Task) show that the counterfactual conditional (past reference) is significantly the hardest one among the three.

As for the incompatible outcome from Task Three, the cause would be attributed to the comprehension/production distinction. As we can see in the raw data, the second most

frequent mistakes (second to the mistakes of the lack of the backshifted tenses) come from the structure of past perfective modal (would/could/should/might + have + Past Participle in the main clause). That indicates that when required to produce, students would suffer from some difficulty in learning this structure. In sum, my findings can be briefly summarized into three points:

(1) The predicted degree of difficulty would be modified as the following:

For the comprehension of the counterfactual conditionals, the counterfactual conditional (present reference) presents more difficulty than the counterfactual conditional (past reference) and non-counterfactual conditional.

(16)

Table 5-10: The degree of difficulty in terms of comprehension

For the production of the counterfactual conditionals, the counterfactual conditional (past reference) is harder than the counterfactual conditional (present reference) and

non-counterfactual conditional.

Table 5-11: The difficulty degree in terms of production

(2) Students find it comparatively difficult to identify the contexts requiring present-referenced counterfactual conditional.

(3) The reason for the difficulty in the past-referenced counterfactual conditional is L2 complexity (to be more specific, the structure of past perfective tense).

(4) For the learning of the counterfactual conditionals, using of the backshifted tenses would constitute the greatest problem.

The first two findings are not completely consistent with the hypothesis proposed in section 4.3 of Chapter 4. My explanation is as follows. As I have mentioned in the summary of Chapter 2, since the counterfactual expressions (past reference) in Chinese require the aid

Present reference > past reference

Present reference > non-counterfactual conditional

Past reference > present reference Past reference > non-counterfactual conditional

(17)

of time temporal and the internal semantic logic of the sentence, they would not be so difficult for Chinese subjects to identify the contexts which must be expressed

counterfactually in the past-referenced tense.

On the other hand, the pragmatic functions of the past reference are more limited to fewer expressions such as regrets and re-evaluation, compared with the more diversified functions of the present reference, which can also be used to convey wish making and advice giving. Therefore, the degree of difficulty of the counterfactual conditionals (past reference) would not be as great as expected.

The finding of the degree of difficulty can also be interpreted as evidence to support the

argument that high school students in Taiwan learn counterfactual conditionals as explicit knowledge⁵. The distinction between explicit knowledge and implicit knowledge (Bialystok, 1978) closely corresponds to the learning/acquisition hypothesis of Krashen & Terrel (1983) and Attention–Processing of McLaughlin (1978). As Brown (1994) indicates, explicit

knowledge is “the facts that a person knows about language and the ability to articulate those facts in some way.” The property of this knowledge, as described in Krashen & Terrel (1983), is learned consciously. These consciously-learned grammars require several conditions for successful application: (a) knowledge of the rule (b) concern with correctness or focus on

5 We can infer that high school students may learn counterfactual conditionals through rote learning since the rote learning of the structure patterns (if + S + V-ed, S + would/should/might/could + V or if + S + had+ PP, S would/should/might/could + have + PP) is a very common way of presenting these two structures in the teaching material.

(18)

form (c) sufficient time (Krashen, 1982). Thus students’ failure to make the correct counterfactual conditionals may indicate that the three above conditions are not all met.

5.5 Conclusions

This research has tried to answer the three research questions which derive from the broader question: Whether the lack of overt counterfactual linguistic categories in Chinese language causes problems for Chinese EFL learners at the senior high school level. We find that counterfactual conditional (present reference) is more difficult than counterfactual conditionals (past reference) and non-counterfactual conditionals. The findings also show that the major problems arise from the use of backshifted tenses and the incomplete learning of the modals in English. These two problems can be best reflected from the construction of past perfective tense that appears in the main clause of the counterfactual conditional (past reference).

On the other hand, this study also seems to suggest that when learning grammar, the comprehension and production ability cannot transfer well to each other since the conclusions drawn from the first two tasks and the third task are not consistent.

From this research, we cannot find any Interlingual errors, Intralingual errors and Developmental errors. We may conclude this study by saying that the lack of overt

counterfactual linguistic categories in Chinese does not particularly constitute difficulty of the learning of counterfactual conditionals for Taiwanese EFL students at the senior high

(19)

school level. The difficulty probably arises from the complexity of the structure.

5.6 Pedagogical Implications

The findings of this study have two pedagogical implications. First, students display

difficulty in identifying counterfactual and non-counterfactual contexts, as the results of the Situation Task shows. This means the distinction between real and unreal condition may still need to be emphasized in the teaching of counterfactual conditionals. Second, the lack of backshifted tenses and the incomplete learning of the past perfective modals suggest that students lack a comprehensive understanding of the tense and modal systems in English.

Thus, it is suggested that the past perfective tense and the past modal systems in English should be learned before the counterfactual conditionals since the result of this study indicates that students encounter some difficulty in the past perfective tense and the modal systems.

5.7 The Limitations of This Study and Suggestions for Future Study

It has to be stressed that the conclusion of this research is by no means deterministic, due to the inherent difficulty of SLA study. During the process of research, a number of constraints have been perceived. First of all, even though we find that there exists a hierarchy of difficulty among the three kinds of conditionals and that counterfactual

conditionals (present reference) constitute the greatest difficulty for students, still this cannot

(20)

give us a clear idea of the learning of counterfactual conditionals. Perhaps, more detailed semantic analysis of Chinese and English can provide a new perspective for this kind of study.

Another limitation is that the distinction between “error” and “mistakes” cannot be clearly distinguished. Finally, the last limitation may come from the research methods, the experimental tasks in particular. Tasks considered to be able to detect students’ learning conditions may, to some extent, suffer from this kind of limitation. That is, students may sense the purpose of the experimental tasks and perform as they are expected and consequently, the result of the experimental tasks would be misleading. In other words, students’ “consciousness” about the rules of grammar might have been raised (about consciousness raising, refer to Krashen (1999)).

The results and conclusions of this research alone cannot provide us with the whole picture of what the learning of the counterfactual conditionals is like. There are still many questions unexplored. First of all, we are still not clear about the learning of counterfactual conditionals in terms of their communicative values. As we know, the communicative function of counterfactual conditionals includes the expression for possibilities, guesses, wishes, regrets etc. Two counterfactual conditionals with identical time references probably do not have the same communicative function. For example,

(I) If I were you, I would not do that.

(21)

(II) If I were eighteen years old, I would play on the basketball team again.

The communicative value of the first sentence can be interpreted as an indirect way of giving advice while for the second one, the function is likely to be purely an expression of an unrealistic wish. Accordingly, we may have a good reason to infer that in terms of

communicative values, the learning of counterfactual conditionals would be a quite different story from the conclusions we have drawn from this research.

On the other hand, future studies could focus on more detailed semantic analysis and study whether the semantic difference is the source of difficulty. Or future studies may try to find out the reason for the lack of backshifted tenses. Can the lack be ascribed to the deficiency of clear counterfactual expressions in Chinese? If so, what is the learning like for speakers of other Indo-European language for whom the counterfactual expression is clearly indicated? If the lack of the backshifted tenses is not found in other Indo-European language speakers, we may have more reasons to argue that the lack of the overt counterfactual linguistic categories in Chinese really cause the problem of the lack of the backshifted tenses in English for EFL Chinese speakers.