General Discussion

Some major findings were found in Sections 4.1.1 and 4.1.2. and the discussion about categorical and age factors is addressed as follows.

To begin with, Type 1, Type 2 and Type 4 were performed better than Type 3 and Type 5 by the subjects and the acquisition order was Type 2 >Type 1> Type 4 > Type 5 > Type 3. Second, the unmarked (Type 1, Type 2 and Type 4) conditionals were significantly easier than the marked conditionals (Type 3 and Type 5) for each experimental group except for the control group. These findings might be due to the positive evidence as it has been claimed in L1 acquisition, which consist of contextualized utterances in a non-tutored language surrounding (Schwartz and Gubala-Ryzak 1992). With regard to the higher percentage of the unmarked conditional types, it is highly likely that children were more likely to acquire these forms earlier than the marked ones because children are exposed to massive quantities of positive evidence (Casilde 2004, Sanz and Morgan-Short 2004) in their daily life.

Then, it was assumed in the present study that conditional adverbials might help trigger conditional interpretation and could be regarded as unmarked forms.

Nonetheless, the present study showed that conditional adverbials was not as crucial as we expected. In Section 4.1.1, the results showed that the mean scores for Type 4 were not significantly different those for Type 1. Type 4, a type without any

65

conditional adverbials, was considered a “plain” sentence which has a common topic chain for ease of reference because the two clauses are semantically related (Haiman 1978, Tsao 1982 and 1990). With regard to this, Type 4, with its smaller structure (Givón 1991 and 1995), is another representation of structural unmarkedness. In addition, this type has an optional but determining adverbial jiu ‘then,’ serving as a clausal connector for the conditionals (Li and Thompson 1981, Wu 1994) and assist children’s conditional processing. In other words, Type 4, without any conditional adverbials, is still an easy type respecting its linguistically economic forms since jiu ’then,’ suggesting the undesirability of the consequent clause (Su 2005),

stimulated our children’s conditional reasoning. Besides, our children’s better performance on Type 4 represented that they were capable of perceiving the implied conditional meanings without any conditional adverbials; in the meanwhile, our children’s general nonlinguistic understanding was re-geared to mental representation (Bowerman 1986). Consequently, it is implied that grammar does “guide our elaborate conceptual work with an admirable economy of overt indications”

(Fauconnier 2003:251).

The insignificant effects of conditional adverbials on conditional acquisition, which was represented by our subjects’ better performances on Type 4 as aforementioned, were further expressed by the subjects’ poor performance on Type 3, having a clause-final conditional adverbial dehua ‘if’ in the antecedent clause. As shown in Figure 4-1, Type 3, a marked type, is the most difficult type among the five types of Chinese conditionals. It was found that the children were unfamiliar with the type because there was no positive evidence provided in the subjects’ textbooks.

Concerning the issue of processing, the antecedent clauses (i.e., if-clauses) themselves served as vehicles for the action performed (Su 2005) and even set up a mental space distinct from our base space (Fauconnier 1985) in Chinese conditionals. Thus, the

66

occurrences of conditional adverbials (i.e., if-words) influenced children’s processing of conditionals. Therefore, Type 3, a type with a clause-final conditional adverbial, is like a sentence structure similar to garden path constructions, which employ more referential mapping skills than clause-initial adverbials. Hence, the subjects’

performance on this type was undesirable in terms of its mental effect and our children might need output drills or exercises to make the type ‘automatic’ (Liming 1990, Krashen 1991), and then ‘learned’ the competence.

Besides, the statistics showed that adverbials might serve as decisive impulses to trigger conditional meanings when the sentence structures did not consist of any if-words. For instance, the performance on Type 5, a type with a wh-word regarded as

an indefinite pronoun (Cheng and Huang 1996, Chierchia 2000) in the antecedent clause but including jiu ‘then’ in the consequent clause, was not performed statistically worse than Type 3. However, the construction itself indicated its complicated characteristic. In the literature, it has been discussed that the L1 acquisition of pronouns emerge early (Hickman and Schneider 1993) but children’s full control of pronouns appeared to be late until the age of six (Hendriks and Spenader 2006) or ten (Hickman and Schneider 1993). These findings indirectly implied that the indefinite pronouns in conditional sentences seem to be more complex on account of their referential meanings of indefinite pronouns and the conditional processing without any if-words (Cheng and Huang 1996, Cherchia 2000).

Such complexity reflected that Type 5 was an example of cognitive complexity according to Markedness Theory in terms of its mental effort (Eckman 1977).

Nevertheless, to determine the conditional interpretation of Type 5, our children reorganized their preliminary conditional knowledge even though they did not fully master the reasoning of indefinite pronouns in Type 5. Therefore, they performed better on this type, reflecting that our children’s optimal interpretation of the marked

67

type highly conformed to their optimal processing of the unmarked forms in the same contexts (Hoop and Krämer 2006), in which the adverbial jiu ‘then’ in Type 5 propelled our subjects’ conditional readings as prevailingly as clause-initial adverbials in Mandarin Chinese conditionals. In consequence, jiu ‘then’ strongly triggered a conditional interpretation in Type 5 while dehua ‘if,’ due to its low frequency of input in the subjects’ textbooks and its processing load, obstructed the subjects’ conditional processing.

In addition to presenting the effects of conditional adverbials on the five types of Mandarin Chinese conditionals, the age issues in conditional acquisition are addressed in the following. Firstly, G4’s (i.e., the six-year-olds) performance showed a transitional stage of the subjects’ conditional acquisition because they performed much better than G1-G3. In addition, the post hoc analysis of the mean scores for G4 did not show any significant differences to the mean scores for G5, whose subjects were the oldest in the experimental groups, and the control group. In consequence, it was found that G4, as well as G5, reached the adult grammar and simultaneously provided an empirical support for the generally-accepted arguments that children’s competence of comprehension and production would affect with each other and sometimes even showed a U-shaped curve since the second graders (i.e., G5) had their own syntactic judgments of conditionals. Nonetheless, the competence and performance would gradually improve with age (Reilly 1982 and 1986, Bowerman 1986, Crutchley 2004).

Moreover, it was found that not all the seven-year-olds (i.e., the subjects of G5) performed best as we expected in the statistics¹. The results showed that G4 was capable of correctly responding to the two major types. The prominent performance        

1  Group 5 got lower scores (unmarked: 0.87 / marked: 0.8) than Group 4, getting the highest mean scores (unmarked: 0.91 / marked: 0.86) among all the other groups even though the comparison of these two groups did not show any significant effects.

68

of G4 and the slightly lower scores found in G5 indicated that our children did not fully acquire the complex structures and needed more linguistic and cognitive incorporation to control the construction until the age of eight ( Reilly 1986) and even in adolescence (Barrouillet and Lecas 2002, Crutchley 2004). Aside from this reason, it was also found that the older children’s syntactic awareness affected their performance in the experiments. In the literature about syntactic errors, it has been claimed that school-age children preferred tomaintain the semantic intent of the message rather than criticallysearch for grammatical errors (Sutter and Johnson 1990).

Furthermore, the syntactic awareness appeared to constitute a higher-level of language skills related to other aspects of language performance (Bowey 1986).

Hence, the argument implied that our children might be more aware of the linguistic interpretation of the target sentences but neglected the trivial differences in wordings.

In addition, there were no significant differences found between G1 and G2 in response to the unmarked and marked conditionals. The two groups performed similarly since they got only 55% of correct responses on the two major types while the other experimental groups (G3-G5) got mean scores around 0.7 or even higher.

The distinction between 0.55 and 0.7, however, reached a significant level. Therefore, one pivotal phenomenon was that the children’s conditional development abruptly rose at age six (G4), which was later compared with the previous arguments about conditionals (Harris, German and Mills 1996, Perner, Sprung and Steinkogler 2004).

There was also no acquisition gap among the three (G1: mean: 3; 8) to the-four-year-olds (G3: Mean: 4;9) on account of their preliminary linguistic knowledge base.

In summary, this section presents some notable issues of the categorical and age factors in the acquisition of Mandarin Chinese conditionals. The statistics will be further examined in Section 4.3.

69 4.2 Hypothetical vs. Counterfactual Conditionals

This section is to discuss the differences between the hypothetical and counterfactual circumstances for our children’s conditional acquisition, aiming to answer the second research question in the present study. Section 4.2.1 presents the results of the raw data. In Section 4.2.2, a second look at the scenario differences, excluding some problematic cases in the imitation task (i.e., the IM task), is shown. In Section 4.2.3, the discussion and analysis of the hypothetical and counterfactual scenarios are addressed.

4.2.1 Comparison of the Hypotheticals and Counterfactuals

In the previous research, hypothetical and counterfactual scenarios have been widely discussed in the acquisition of conditionals (Riggs and Peterson 2000, Robinson and Beck 2000, Perner, Sprung and Steinkogler 2003). The present study aims to compare these two scenarios in L1 acquisition of Mandarin Chinese conditionals. The subjects’ mean scores for the hypotheticals and counterfactuals are shown in Figure 4-3:

Figure 4-3 A Comparison of the Hypotheticals and Counterfactuals

As can be seen in Figure 4-3, the mean scores for hypothetical and counterfactual conditionals for each group were: G1: 0.52 vs. 0.5; G2: 0.56 vs. 0.52; G3: 0.78 vs.

70

0.75; G4: 0.91 vs. 0.89; G5: 0.86 vs. 0.85; the control: 0.98 vs. 0.98 ). Besides, the within-group comparison indicated all the experimental groups did not show significant differences in the hypothetical and counterfactual conditionals (p >.05) but they performed relatively better on the hypotheticals than on counterfactuals.

Therefore, it suggested a tendency that the hypotheticals were prior to the counterfactuals. Concerning the within-group responses, G1-G3 performed better on hypotheticals than on counterfactuals: hypotheticals > counterfactuals. G4 and G5 performed on the two scenarios nearly alike but still showed better performances on the hypotheticals: hypotheticals > counterfactuals. As for the control group, the statistics showed significant effects (p= 0.04 < .05) on the two scenarios, indicating that the adults could demarcate the scenario differences and made correct judgments.

However, the control group performed significantly better on counterfactuals than on hypotheticals: counterfactuals > hypotheticals).

The between-group comparison of the hypothetical conditionals by one-way ANOVA showed that the differences reached a significant level (F (5, 102)= 49.436, p < .05). The post hoc analysis also showed that G1 and G2 performed alike and G4

performed the best while there was no significance differences among G4, G5 and the control group.

With regard to the between-group comparison of the counterfactual conditionals, the results showed there were significant differences (F (5, 102)= 42.389, p < .05).

The post hoc analysis, however, demonstrated that G1 and G2, showing no significant differences, performed similarly on the counterfactual conditionals. In the meanwhile, G4 still performed better than G5 while the differences did not yield any significance on the counterfactual conditionals.

As addressed above, the results showed that the distinction between the two conditional scenarios was not significant. Nonetheless, the tendency that hypotheticals

71

were acquired prior to counterfactuals in our children’s conditional development was still in accord with the documented findings in the literature (Harris, German and Mills 1996, Riggs and Peterson 2000, Robinson and Beck 2000, Perner, Sprung and Steinkogler 2003). Nonetheless, it appeared that the developmental relation of conditional scenarios was indistinct.

4.2.2 A Second Look at the Hypothetical and Counterfactual Conditionals

With regard to the results that the hypotheticals were not significantly easier than counterfactuals presented as in Section 4.2.1, it is important to see if there were some exceptional cases in the two tasks. After examining the raw data and voice recordings, some problematic cases in the imitation task (the IM task) were found, as shown in (1), (2) and (3):

‘If I were a little pig, I would like to build a solid house in the forest.’

(Q13⁴)        

2  Q1, one of the hypothetical target sentences (Type1), refers the first question of the imitation task.

3 Q6, one of the counterfactual target sentences (Type 4), refers the sixth question in the imitation task.

4 Q13, one of the hypothetical target sentences (Type 3), refers the thirteenth question in the imitation task.

72

The above test items yielded the subjects’ inconsistent performances compared with the other items. Example (1), one of the hypothetical Type 1 sentences, consisted of much semantic loading (gaoxingde ‘happily’ and pao-lai-pao-qu ‘run around’);

therefore, most of the subjects did not perform well on this question while they performed much better than other target sentences of Type 1. Example (2), one of the counterfactual Type 4 sentences, made the subjects add ruguo ‘if’ or dehua ‘if’ but they did not show the tendency for other Type 4 sentences. Example (3), one of Type 3 target sentences including a relatively difficult word jiangu ‘solid,’ resulted in most of the young children’s partial production in the consequent clause on account of unfamiliarity but their production of other Type 3 sentences did not show the same trend. Figure 4-4 presents the results after removing these three test items:

Figure 4-4 A Revised Comparison between Hypothetical and Counterfactual

The difference in responding to the hypothetical and counterfactual conditionals, as shown in Figure 4-4, was clear (G1: 0.56 vs. 0.51; G2: 0.58 vs. 0.52; G3: 0.79 vs.

0.75; G4: 0.93vs. 0.89; G5: 0.87 vs.0.85; the control group: 0.97 vs. 0.99).

Nonetheless, the within-group comparison in the paired sample t-test demonstrated that these two scenarios still did not reach a statistically significant level ( p >.05) except for the control group (p =0.04 <.05) but there was a tendency in general as addressed before: hypotheticals > counterfactuals (excluding the performance of the

73

control group). G5 responded to the two scenarios similarly while the control group still performed better on the counterfactuals than on the hypotheticals.

As for the between-group analysis of the hypothetical conditionals, one-way ANOVA showed a significant difference (F (5, 102)= 35.696, p < .05). G1 and G2 were at a similar performance level according to the post hoc analysis. G4, who performed the best, showed no significances to G5 and both groups did not show significant differences with the control group.

Regarding the between-group comparison of the counterfactual conditionals, the results revealed a significant difference as well (F (5, 102)= 42.389, p < .05). G1 and G2 performed alike. In addition, G4 and G5 performed quite similarly compared to the control group.

4.2.3 General Discussion

In this section, some major findings addressed in Sections 4.2.1 and 4.2.2 are discussed as follows.

First, the results showed that the differences between the hypothetical and counterfactual conditionals were not significant. This result, however, still showed a trend that the hypotheticals proceeded the counterfactuals in L1 acquisition of Mandarin Chinese conditionals, corresponding to the findings of the superiority of the hypotheticals over the counterfactuals in L1 acquisition of English conditionals. In the literature, it was consistently found that the subjects’ performance on the hypothetical scenario was better than on the counterfactual scenario because such ability was in accord with their cognitive development. Hypothetical reasoning employed the ability of creating imaginative state of affairs, like children’s competence in the play that a brick is a car or a house to a certain extent, arose at around age 2 or 3 (Riggs and Peterson 2000). Moreover, the previous research argued that the children’s imagined

74

reasoning might not be in conflict with their realistic state (Beck et al. 2006). Most of the results, obviously, showed that their children were more prone to answer correctly on the hypothetical conditionals than on the counterfactuals, which required them to imagine an alternative situation replacing what was known to be false (Mitchell and Lacohee 1991, Russell 1996, Taylor and Mitchell 1997, Robinson and Beck 2000).

Furthermore, there was no doubt that our young children were ready for the hypothetical conditionals since other studies even found that many three-year-olds could distinguish future hypothetical questions more easily (Riggs et al. 1998, Beck et al. 2006) than counterfactuals. Though the present study did not show a significant difference between the two scenarios, the tendency still supports a common statement that the counterfactual reasoning intervened in children’s current view of reality (i.e., the prepotent response) when asked to think counterfactually (Riggs and Peterson 2000, Robinson and Beck 2000). In other words, in counterfactual reasoning, young children’s current knowledge base has to identify two possibilities at a time, one of which happens and the other of which could have happened. Furthermore, the results found in G1 and G2 on the counterfactuals showed that the preschoolers’ less control over the counterfactual conditionals, exhibiting that our young children could not suppress the salience of their own knowledge base (Russell 1996).

The reason why the differences of the hypothetical and counterfactual conditionals among the experimental groups were not significant is addressed as follows. A close scrutiny of the two scenarios showed that there was a crucial procedural factor found in the present study. In Kavanaugh et al. (1995) and Harris et al. (1996), as well as in the present study, an ordinary state of affairs was changed into a negative state by some accidents or mishap. Owing to these negative outcomes, children might be inclined to think of a means in which the misfortune might have been avoided and it prompted their counterfactual reasoning (Harris and Leevers

75

2000).

Regarding the situation that the control group performed slightly worse on the hypothetical conditionals, the inspection of raw data, nonetheless, showed their adept linguistic competence to avoid ambiguity. Some participants of the control group got lower scores in the hypothetical production task especially for Type 5 (i.e. wh-word conditionals) because the antecedent clause was ambiguous in that shei shi ‘whoever was,’ implying two possible readings, one of which is interrogative ‘who is’ and the other of which is conditional ‘if whoever was’. On the contrary, the sequence of the antecedent clauses for Type 5 in the counterfactual scenario was not ambiguous, such as Shei dai gushi shu,… ‘If whoever had brought a story book, …’ suggesting a much stronger conditional reading in processing. To anchor the conditional meaning of the hypothetical wh-word conditionals, the adults, thus, employed a clause-final adverbial dehua ‘if’ to avoid ambiguity of the antecedent clause while the addition made them

disobey the rules. Such a strategy demonstrated that our adults were apt to modify their mental representation and capable of differentiating two possible interpretations of the target sentences. Furthermore, their production data demonstrated their subjective evaluation to the actuality of the proposition (Akatsuka 1986), suggesting adults’ conversational concern and discourse susceptibility. However, our children were still at the threshold stage of differentiating hypothetical and counterfactual readings as possibilities through the actual events (Beck et al. 2006, Robinson and Beck 2000).

With regard to the mature level of scenario differentiation, the performance of G4 and G5 almost reached maturity. Such statement corresponded to the argument that children around 5 or 6 years old could acknowledge possibilities from things that did not happen or could have happened and genuinely employed an alternative to resolve hypothetical and counterfactual conditionals (Beck et al. 2006). In addition,

76

the 5-or-6-year-olds’ mature quality of scenario was consistent with the argument for adult grammar on the five types of Mandarin Chinese conditionals stated in Section 4.1.3.

4.3 Task Effects

The third research question aims to discuss the task effects on children’s L1

acquisition of Mandarin Chinese conditionals. In order to reduce the subjects’ biased responses, an interpretation task (the IT task) and an imitation task (the IM task) were designed to explore our children’s performance on the conditional constructions and investigate whether and how they performed differently on the two tasks. Then, the results of each task are presented in different formats, such as the unmarked vs.

marked types, hypothetical vs. counterfactual scenarios and the developmental trend of each task in Section 4.3.1. Finally, a general discussion of the task effects is

marked types, hypothetical vs. counterfactual scenarios and the developmental trend of each task in Section 4.3.1. Finally, a general discussion of the task effects is