Chapter 2 Literature review
2.3 Children’s language performance in conversation and narration
Previous studies on children’s motion encoding have employed conversational and narrative data respectively to analyze children’s language patterns with different purposes. Narrative data is the most extensively used data in studying motion
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
encoding. The researchers used either wordless picture books or motion pictures to elicit children’s motion expressions (Slobin 1996, 2004; Özyurek & Özçaliskan 2000; Chen 2005; Pourcel 2005; Parafragou et al. 2006; Lin 2006; Guo & Chen 2009; Hickmann 2006). These studies usually concerned different language-specific speech patterns — including the typological issue — and “thinking for speaking”
issue. On the other hand, conversational data can also be found in several studies over the years (Choi & Bowerman, 1991; Slobin, 2003; Zheng & Golding-Meadow, 2002; Hohenstein, 2013). The researchers analyzed children’s spontaneous use of language in daily conversations. One of the advantages of using conversational data is that it is a more effective way to collect the speech of those children younger than three years old (Choi & Bowerman, 1991; Slobin, 2003; Hickmann & Hendriks, 2010). Conversational data can also be used to collect motion expressions from deaf children (Zheng & Golding-Meadow, 2002) or the motion expressions in child-directed speech (Hohenstein, 2013). It seems that narrative and conversational data both have their advantages with respect to studying different aspects of children’s motion encoding. The disadvantage, however, is that it is unclear if the language patterns found in one genre can also be observed in another. The factor of genre remains to be a potential variable that might affect the results of the studies.
There are few studies that focus on the factor of genre in children's motion encoding. Slobin (2000) examined adult’s motion expressions in conversation and narration in order to elaborate the theory of “thinking for speaking”. It was found that adults of different languages encoded motion event construction in an identical way in narration and conversation in terms of verb type usage. He found consistent results
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
for the S-language adult speakers (in the study, English speakers) who used more types of Manner verbs than Path verbs in both narration as well as conversation.
Similarly, how children talked about motion has also been studied by Slobin. Two of Slobin's study (1996b; 2000) collected children’s narrative data and conversational data respectively. The first study of Slobin’s (1996b) applied narrative data, and it was found that children’s motion verb type was consistent in conversation and narration. V-language children used fewer types of Manner verbs, while S-language children use more. These results with narrative data are comparable to Slobin’s second study (2000) with conversational data, in which he compared the Manner verb type of children speaking typologically different languages in conversation. In other words, according to two of Slobin's study (1996b; 2000), S-language children used more types of Manner verbs and V-language children used more types of Path verbs in both conversation and narration. These two studies have implied that children’s narrative and conversational motion encoding both showed language-specific characteristics in terms of verb type. Although the purpose of Slobin’s studies did not aim to solve the issue of genre differences in motion encoding, he did pin out the importance to examine actual language use in different genres, and his studies have proposed that certain similarity, namely verb type usage, might be found in the children’s expressions of motion events in both genres.
Another study of Selimis and Katis (2010) takes the genre differences of speech into account as a factor of affecting children’s motion expression’s usage. The researchers studied English-speaking and Greek-speaking children’s motion expressions by examining the data collected from three different methods: picture
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
narratives, film narratives, and spontaneous conversation. They found that, despite that Greek and English are typologically different languages, children in both language groups used more types of Manner lexeme in conversation than in narration. Therefore, they claimed that Manner is more salient in conversation than narration. According to Selimis and Katis's study, it is expected in the current study that more Manner types can be found in conversation than in narration in Mandarin children’s motion expressions.
The above two studies of Slobin (1999b, 2000) and Selimis and Katis (2010) only examined the usage of verb type. However, the analyses concerning the speech habit of motion encoding from other previous studies were not confined to the usage of verb type. The usage of verb token and verb construction has also been analyzed in other previous studies on motion expressions. In fact, a language-specific characteristics of motion expressions was said to be even stronger when verb token was being analyzed. For example, Özçalışkan and Slobin (1999) found systematic differences in verb token between V-languages and S-languages, but not in verb type.
Therefore, in this current study, in addition to verb type, verb token and verb construction are also included and analyzed.
Although few studies discuss the differences of motion verb token in conversation and narration, previous studies did imply that children’s linguistic expressions can vary in some respects. Selimis el al.(2010) has pointed out that speaker’s habitual speech “can vary depending upon factors such as the type of discourse, its context and content”. There are several differences between conversation and narration. The first difference is that the two discourses demand
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
different communication responsibilities. Narration demands greater communication responsibility as opposed to conversation (Logan et al., 2011). In conversation, there is at least one interlocutor interacting with the target child, while in narration, the participation of the other interlocutors, if any, is little. The joint participation in message development will influence children’s language use (Logan et al., 2011).
The second difference is the degree to which the motion element can be inferred. In conversation, children may talk about the current motion that can be observed by both the speaker and the listener. However, in narration, the motion that the children talk about is depicted in pictures, in which the motions are not shown in real world. The way speakers talk about motions seems to be influenced by whether the motion is presupposed or not by the audience. For example, Papafragou et al.
(2006) found that when Manner was not inferable by the hearer, Greek speakers were more likely to mention this element, even though Greek is a V-framed language that prefer to use Path. If so, then the difference of the visibility of motion in narration and conversation might affect how children talk about Path and Manner.
Finally, it was found that children produced more complex utterances in narration than in conversation (Dollaghan et al., 1990). For example, children’s MLU was higher in narration, and the frequency of bound morphemes was also higher in narration, compared to conversation (Leadholm & Miller, 1992). Therefore, it is also expected to find difference between the motion construction in narration and in conversation in this current study. In terms of motion construction, it is possible that children might use more complex construction, i.e., more number of verbs, in narration than in conversation.
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
A language like Mandarin, in which Manner and Path are equally encodable, allows children to have more freedom to choose what to mention and what not to.
With the freedom provided by Mandarin, what affect children to attend to certain aspect of motion might lie in context and in language use, just like the previous studies on Mandarin children’s motion expression mentioned in section 2.2.2 suggested. Although previous studies suggested that there have been some distinctions in Mandarin children’s motion expression patterns in different genres, the comparison needs to be done in a more consistent way.
In sum of section 2.3, how context in different genres affects children’s motion expressions does not seem to have a consistent and extensive conclusion, no matter in Mandarin or other languages. In order to cope with this issue, the current study re-examined and compared the results collected from conversation and narration data respectively. The comparison aims to answer the last research question “How do the motion encodings of Mandarin-speaking children ages 3, 4 and 5 differ in conversation and in narration?” The possible differences might be found based on the suggestions of previous studies. First, in terms of verb type, previous studies have shown that children used more types of Manner verbs in conversation than in narration. Second, children’s motion expressions can be different because of the following characteristics: the communication responsibility is higher in narration, and the salience of the motion elements is higher in conversation. Last, in terms of motion constructions, previous studies suggested that children’s utterances were more complex in narration, so more complex motion construction might be observed in narration in the current study.
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Chapter 3 Method
As mentioned in Chapter two, previous studies concerning children’s motion expressions collected data from two genres: eliciting narration and spontaneous conversation. In this current study, with the purpose of examining children’s expressions of motion events in these two different genres in a consistent way, both types of the data will be collected, namely the conversational data from natural spontaneous interactions and the narration data elicited by story-telling.
3.1 Subjects
The information of the subjects and the data are presented in Table 1. The age range of the participants chosen was between 3 and 5 considering children under three lacking the ability to narrate a whole story. Children under the age of 5 have been widely used as a sample in other studies, which means it is possible to compare the data found in this study with those works. Therefore, the three age groups of participants are 3, 4 and 5 years old. In conversational data, 8 children in total participated. In the group of age 3, there were 2 females and 1 male; age 4, 1 female and 1 male; age 5, 1 female and 2 males. The data length of each age group was roughly three hours. In accordance with the participants in natural conversation, children of 3, 4 and 5 years old were participated in narrative data. 22 children participated in total. In the group of age 3, there were 3 females and 4 males; age 4, 2 females and 5 males; age 5, 4 females and 4 males. The data length of each age
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
group was about 30 minutes. The data length from the two genres was determined by the motion clauses collected. The number of motion clauses from conversational data and narrative data reached a comparable quantity. The motion clauses from these two set of data can be found in Table 2 and Table 7 respectively. All the participants and their caregivers in both conversational and narrative data are native Taiwanese Mandarin speakers. The main language the children use at home and at school is also Taiwanese Mandarin. All the participants live in Taipei City.
3.2 Data collection
3.2.1 Motion expressions in natural conversation
The data was from the Language Acquisition Lab in the Graduate Institute of Linguistics in National Chengchi University directed by Professor Chiung-chih Huang. The data was collected in daily interaction between mother and child at home. The observer went to the participant’s house and recorded the natural interaction between mother and child with little intrusion. The situations include the
Table 1. Information of subjects and data length
Genre Conversation Narration
minutes 32 minutes 30 minutes 28 minutes
‧
child playing with toys, helping the mother with house chores, drawing pictures, etc.
The story-reading sessions were excluded from the conversational data. Each session lasted around one hour. All sessions were video-taped and transcribed in accordance with the CHAT Transcription Format.
3.2.2 Motion expressions in elicited narration
The data on elicited narration was also from the Language Acquisition Lab in the Graduate Institute of Linguistics in National Chengchi University. The wordless picture book Frog, where are you? (Mayer, 1969) was used to elicit children’s narration. The book has been widely used in eliciting motion expressions from adults and children of different languages (Berman & Slobin, 1994), since the storyline includes many movements of the characters. The book is about a boy and his dog finding a pet frog which escapes from his house. The data collection process followed these steps as follow: first, the picture book was shown to the child from the start to the end. Second, the researcher turned to the first page of the book and asked the participant to tell the story from the beginning. The observer only gave minimal instruction to encourage the child to speak more or to continue speaking, such as saying “And then?”, “What about the boy/the dog?” or “What happens here?” Each participant was tested individually. The whole process was recorded and the utterances were transcribed in accordance with the CHAT Transcription Format.
3.3 Coding
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Before coding, motion clauses were first identified in the transcription. All the motion clauses contain either one motion verb or a motion serial-verb, and it is either a Path verb, a Manner verb, a Deictic verb or a combination of any of the two or all three verbs. According to Talmy (2000) and Guo and Chen (2009), a motion clause includes that:
a) The Figure moves between two macro-locations.
The sentence below conveys that the Figure “the train” moves from its original point to another location “the grass”.
⽕火⾞車 到 草 裡⾯面 了。︒
huoche dao cao limian le train arrive grass inside PRT
“The train goes into the grass.”
b) The Figure changes posture.
As illustrated in this example, the Figure “the car” changes its posture by turning from one direction to another.
汽⾞車 就 轉 過 來。︒
qiche jiu zhuan guo lai car then turn pass come
“Then the car turns over.”
c) The Figure disappears or appears with respect to the ground.
The Figure “he” appears from a place that is not specifically pointed out, but it is obvious that the motion conveys the beginning point of the change of location — here.
他 從 這裡 鑽 出 來。︒
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
ta cong zheli zuan chu lai he from here dig out come
“He digs out from here.”
d) The Figure engages or disengages to or from the ground.
In this example, the Figure “the disk” is moved relative to the Ground, which is implied from the Path verb huí “return”.
把 光碟 放 回 去。︒
ba guangdie fang hui qu BA disk put return go
“Put the disk back.”
After the motion clauses were identified, the main verbs of each motion clause were examined. Previous studies often analyzed children’s motion expressions in several aspects: motion verbs, motion constructions, and motion syntax. This current study focuses on children’s motion verbs and motion constructions. Each motion verb and motion construction was coded. The coding categories and examples are presented below.
3.3.1 Motion verbs
After the identification of the motion clauses, the main verbs of each clause were coded. As mentioned in Chapter 2, previous studies on Mandarin-speaking children’s motion expressions coded motion verbs in different ways. In the current study, three categories of motion verbs are coded, namely Manner, Path, and Deixis.
The coding system of the current study is closer to the coding scheme of Huang’s
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(2012) study. However, Huang’s coding scheme was not explained and illustrated clearly enough, and further analysis in her study did not conform to the origin coding system. There are some reasons for the classification of the motion verbs in the current study. First of all, Manner included Cause and neutral verbs. The reason why both Cause verbs and neutral verb were both coded as Manner verbs was that there were not so many Cause verbs and neutral verbs in the data, and it was more convenient to combine them into one category to compare with Path. Cause was combined into Manner because Manner and Cause are both the co-event of motion, and both indicates how the Figure moves. Also, neutral verbs are merged into Manner because what Chen (2005) and Guo and Chen (2009) coded as a neutral verb acquires “the function and meaning of Manner verbs” (p. 198). For example, wen 聞
“smell” in wen chu qu 聞出去 “smell-pass-go” illustrates the Figure exiting with the manner of smelling. Since Cause and neutral verbs all share certain quality of Manner of motion, there was no need to further categorize them into smaller categories. Second, Deixis was coded separately from Path, because it is an important motion verb appearing frequently and productively in Mandarin serial verbs (Guo & Chen, 2009). Since the current study focused solely on Mandarin, Deictic became a rather important category. The examples and definition of each verb category are illustrated as follows:
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
a) Manner (M): the way the Figure moves, the co-events of motion (e.g 跑 pao
“to run”, 跳 tiao “to jump”, ⾶飛 fei “to fly”, ⾛走 zou “to walk”, 掉 diao “fall”). 1 2 Other verbs such as Cause verbs are the external force that moves the Figure (e.g. 拿 na “to take”, 放 fang “to put", 丟 diou “to throw”).
b) Path (P): the route that the Figure follows with respect to the Ground (e.g.上 shang “to rise”, 下 xia “to descend”, 進 jin “to enter”, 出 chu “to exit”, 回 hui
“to return”, ⾛走 zou “to leave”, 掉 diao “leave”).
c) Deixis (D): Li and Thompson (1989) mentioned that Deixis’ were used “when reaching a destination is implied”. It indicates the Path by implying what direction the Figure moves in, whether it is towards the speaker 來 lai “to come” or away from the speaker 去 qu “to go.
3.3.2 Motion constructions
The types of serial-verb constructions were coded and further categorized by number of the verbs based on all the constructions found in our data. The main verbs of a motion clause can be constructed with one to three verbs. The number of verbs
The word zou ⾛走 “walk” in some cases conveys the meaning of 離開 “away from” or “leave”
1
(教育部國語辭典). Therefore, some ⾛走 are coded as M if they actually mean “walk” (e.g. zou qu xuexiao ⾛走去學校 “walk to school”) and some are coded as P if they lose the meaning of the actual manner of moving slowly with feet (e.g. women zou ba 我們⾛走吧 “Let’s go.”).
The word diao 掉 can either mean “fall” or “leave, away” (教育部國語辭典). When it means
2
“fall”, it occupies the first slot of the serial motion verbs (e.g. diaoxialai 掉下來 “fall down”) and is coded as M. When it means “leave” or “away”, it occupies the second or third slot of the serial motion verbs (e.g. ta paodiao le 他跑掉了 “He ran away”) and is coded as P.
‧
and the types of verb construction are both coded. The examples are presented below:
a) One-verb construction: Deixis, Manner and Path are all allowed to exist alone as the main verb.
“The little rabbit jumps and jumps.”
這個 放 那裡。︒
b) Two-verb construction: the first slot of the two-verb structure is either Manner or Path, while the second slot is restricted to Deixis or Path. There are four