Materials

The picture verification task was adopted in the comprehension task. There were 18 trials in total, 6 of which were practice trials, 8 of which were experimental trials and 4 of which were filler trials. The practice trials consisted of 2 RC trials, 2 AC trials and 2 CC trials in order. The RC and AC test items in the practice trials were designed to make sure that subjects know how to respond to the filler trials, which were extracted from the RC and AC test items. The 8 CC experimental trials were selected from the whole set of CC test items. In other words, there were 12 CC test items in total, 8 of which were chosen to be the experimental trials for the CC comprehension task, and 4 of which were averagely adopted to be the filler trials for the RC and AC comprehension tasks. The 12 CC test items could be separated into 2 sets, each of which was composed of 6 sentences with CCs. A half of the 6 sentences belonged to the left-branching CC type, namely, CCs in the subject position of a matrix clause, and the other half were the members of the right-branching CC type, namely, CCs in the object position of a matrix clause. In each branching type were 3 CC types: statements, WH-questions and A-not-A questions. Of the statement CCs, only CCs with raising predicates were taken into consideration. For all the complex sentences with CCs, the matrix clause and the CC were designed to carry one proposition, respectively, making the complex sentence contain two propositions.⁴ Table 2.1 illustrates the distribution of the test items.⁵

4 Since complex sentences with RCs and complex sentences with ACs also embrace two propositions, for the sake of being on the equal footing, complex sentences containing CCs with two propositions were adopted in the dissertation.

5 Prof. Tsao (personal communication) has pointed out that the majority of our test items belong to the ECM (exceptional case-marking) construction since most of the complement clauses in Mandarin occur in the ECM (exceptional case-marking) construction and the serial verb construction.

39

Table 2.1 The Distribution of the 12 Test Items for the CC Comprehension Task

Branching Direction CC Type Set I Set II

Left-branching

Statement 1 1

WH-question 1 1

A-not-A question 1 1

Right-branching

Statement 1 1

WH-question 1 1

A-not-A question 1 1

Each test item was accompanied with one picture, in which two main characters were chosen from Xiyangyang & Huitailang, a popular cartoon that almost all children aged from 3 to 12 could easily name the two. Each picture would either match with its corresponding test item or not. Half the sentences in the practice, experimental and filler trials should be judged as ‘true’ in response to the matched pictures, while the other half should be judged as ‘false’ in response to the unmatched pictures. All the false pictures corresponding to the left-branching CCs were designed to wrongly describe the actions mentioned in the CCs, while the false pictures corresponding to the right-branching CCs were designed to wrongly regard the subjects of the CCs as the objects of the matrix verb.

All test items were recorded with a normal speed via Praat in a sound-proof studio beforehand. In addition, all the pictures were presented in the computer screen.

The test items and their corresponding pictures in the practice, experimental and filler trials are presented in Appendix I. In addition, the whole set of the test items in order is given in Appendix II.

2.3.2.2 Production

Following Hsu’s (2006) study on children’s RC production, we adopted the elicited task to investigate children’s CC production. The materials for the production task consisted of 6 practice trials, 8 experimental trials and 4 filler trials. Like the

40

comprehension task, the practice trials consisted of 2 RC trials, 2 AC trials and 2 CC trials. The 8 CC experimental trials were selected from the whole set of CC test items, that is, 12 CC test items in total, in which the other 4 were averagely adopted to be the filler trials for the RC and AC production tasks. The CC test items could be divided into 2 sets, each of which was composed of 6 expected responses with their sample utterances and elicited questions. The expected responses were the same with the test sentences in the comprehension task.

In each test item, there were two sample utterances. The first utterance, composed of an animate subject and an inanimate object, was always the base sentence for making a CC in response to the elicited question.⁶ The second utterance was always the sentence for making a matrix clause. The matrix verbs for the constructions with the left-branching CCs were raising predicates, viz. hen nan ‘very difficult’ and hen rongyi ‘very easy,’ which allowed the subjects of CCs to be topicalized. The matrix verbs for the constructions with the right-branching CCs were verbs which could take NPs or clauses as their complements. The verbs included kanjian ‘see’ and jide ‘remember,’ which were commonly used matrix verbs that took complement clauses in Mandarin (Huang, 2003). One test item from each branching type was given in (9) and (10).

6 Fikkert and Hoop (2009) maintained that in the comprehension process, animacy is an important source of information for young children. In the world languages, generally, subjects are often animate, whereas objects are often inanimate. When a sentence tallies with the pattern, it is easier to comprehend. A sentence with inanimate subject and animate object leads to difficulties in comprehension. On the other hand, animacy does not contribute to the optimization of the form in production because the animacy of the subject and the object is already given in the input. When producing a sentence, young children only follow the word order constraint PRECEDENCE, which requires the subject to precede the object. Since animacy does not fall in the inspection of the current study, either in production or in comprehension, the collocation of the subjects and objects is controlled as animate subjects and inanimate objects, providing semantic aid for the comprehension of the sample utterances in the production task and the comprehension of the test sentences in the comprehension task.

41

(9) Left-branching: Statement Sample Utterances

Xiyangyang chi da xigua. Zheyang hen nan.

Xiyangyang eat big watermelon this very difficult

‘Xiyangyang eats a big watermelon. This is very difficult.’

Elicited Question

Qingwen: Sheme hen nan?

question what very difficult

‘Question: What is very difficult?’

Expected Response

Xiyangyang chi da xigua hen nan.

Xiyangyang eat big watermelon very difficult

‘That Xiyangyang eats a big watermelon is very difficult.’

(10) Right-branching: Statement Sample Utterances

Huitailang ti qiu. Xiyangyang kanjian le.

Huitailang kick ball Xiyangyang see LE

‘Huitailang kicked a ball. Xiyangyang saw it.’

Elicited Question

Qingwen: Xiyangyang kanjian sheme?

question Xiyangyang see what

‘What did Xiyangyang see?’

Expected Response

Xiyangyang kajian Huitailang ti qiu.

Xiyangyang see Huitailang kick ball

‘Xiyangyang saw that Huitailang kicked a ball.’

42

All test items were recorded via Praat in a sound-proof studio. Each test item was accompanied with three pictures: The first picture described the content of the first sample utterance, the second picture depicted the content of the second sample utterance, and the third picture illustrated the corresponding answer along with the elicited question so that the subjects would be led to describe the picture with the expected response. For each test item, the three pictures were presented with the audio test sentences, between which a one-second pause was inserted, and were broadcast by Powerpoint automatically.

The filler trials were consisted of 2 experimental trials from the RC production task and 2 experimental trials from the AC production task. The contents of the test items in practice, experimental and filler trials are shown in Appendix III. In addition, the whole set of the test items in order is given in Appendix IV.