Chapter 3 Methodology
3.3 Coding scheme
3.3.3 Reliability
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
29
3. Evaluation: Utterances which are non-events and contain only evaluation such as descriptions of internal states (emotions, cognitions, and physical states), intentions, intensifiers, compulsions, explicit negatives, etc. (e.g., Mother: You were scared at that time, right? or Child: I was very happy.).
4. Speech: Utterances which include direct or reported character speech (e.g., Mother: The doctor said you should take the medicine. or Child: The uncle told her that she should be careful.).
5. Coda: Utterances which serve to mark an explicit conclusion (e.g., Mother:
That’s all. or Child: That’s all what I remembered.).
6. Non-narrative talk: Utterances which are completely unrelated to the narration of the story (e.g., Mother: Do you want to eat the pudding now? or Child: I want to eat the bread.). Also included are utterances which are related to the narrative in a tangential way, such as general knowledge talk, fantasy talk, and future talk.
3.3.3 Reliability
About one-third of the mothers’ data in the co-constructed narratives and one-third of the children’s data in the independent narratives to the researcher were randomly selected and then coded independently by another trained rater in
accordance with the coding schemes described earlier. The rater is also a Mandarin
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
30
speaker. Then, Cohen’s Kappa was used to calculate the reliability of the data. The inter-coder reliability reaches 0.86, which represents a considerably high agreement between the two coders.
‧
Analysis of mothers’ ways of co-constructing narratives with children and analysis of children’s narrative performance in researcher-child conversations will be shown in section 4.1 and section 4.2, respectively. The effect of mothers’ ways of co-constructing narratives with children on children’s narrative performance will be presented in section 4.3.
4.1 Analysis of mothers’ ways of co-constructing narratives with children At first, we presented the number of co-constructed narratives in mother-child conversations, the number of total utterances per narrative, and the number of
utterances that the mothers and the children produced per narrative when the children were 3;6, as shown in Table 1.
Table 1. Number and length of narratives in mother-child conversations
Note. The ratios of maternal/child utterances per narrative to total utterances per narrative are given in parentheses.
Category
Utterances per narrative 28.6 33.9
Maternal utterances per narrative 18.0 (62.9%) 20.6 (60.7%) Child utterances per narrative 10.6 (37.1%) 13.3 (39.2%)
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
32
Table 1 shows that in conversation between D’s mother and D, they
co-constructed 21 narratives, while in conversation between Z’s mother and Z, they co-constructed 16 narratives. As for the ratio of maternal utterances per narrative to total utterances per narrative, D’s mother is 62.9% and Z’s mother is 60.7%. This suggested that the two mothers produced quite similar proportion of utterances per narrative. As for the ratio of child utterances per narrative to total utterances per narrative, D is 37.1% and Z is 39.2%. This indicated that the two children were also quite similar about the proportion of utterances that they produced per narrative when they were 3;6. In addition, it is worth noting that the proportion of utterances that both mothers produced per narrative is higher than the proportion of utterances that both children produced per narrative, suggesting that mothers contributed more utterances than children did in co-constructed narratives.
In the following section, we will present the findings of the two mothers’ ways of co-constructing narrative with their children in terms of the conversational styles, which will be shown in section 4.1.1 and in terms of the narrative information types, which will be presented in section 4.1.2.
4.1.1 Conversational styles
Maternal utterances were classified into six types. The mean frequencies and distribution of utterance types per narrative are presented in Table 2.
‧
Table 2. Mean frequencies of maternal utterance types per narrative
Table 2 revealed that the highest mean frequency of the two mothers’ utterances was elaborations. This suggested that they often elaborated on the narrative topic when they talked about past events. Repetitions seldom occurred in the two mothers’
utterances. With regard to clarification questions and evaluations, D’s mother clarified and evaluated her child’s utterances more often than Z’s mother did. As for associative talk and off-topic talk, both mothers seldom made these kinds of utterances.
Furthermore, in order to determine the two mothers’ conversational styles on the basis of Reese, Haden, & Fivush’s (1993) classification, the ratio of elaborations to
repetitions is presented in Table 3.
Table 3. Mothers’ elaborations to repetitions ratio
Category D’s mother Z’s mother
Elaborations
Ratio of elaborations to repetitions 6.92 7.42
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
34
Based on Reese, Haden, & Fivush’s (1993) study, if mothers used at least twice as many elaborations as repetitions, they would be classified as high-elaborative mothers; if mothers’ use of elaborations was nearly equal to the use of repetitions, they would be classified as low-elaborative mothers. As a result, according to this classification, the results in Table 3 reveal that the two mothers were both
high-elaborative mothers because the ratios of elaborations to repetitions were higher
than 2. That is, both mothers used more than twice as many elaborations as repetitions.
To sum up, on the basis of the quantity of talk that the two mothers elaborated on the topic, they could be viewed as displaying the same conversational style:
high-elaborative style. In order to gain more insight into how the two mothers
co-constructed narratives about the past with their children, we further analyzed the type of narrative information that they provided or requested from their children in the co-constructed narratives.
4.1.2 Narrative information types
The mean frequencies and standard deviations of the narrative information that were provided by or requested by the mothers per narrative are shown in Table 4.
‧
Table 4. Mean frequencies and standard deviations of mothers’ narrative information types per narrative
* p < .05
Note. t values refer to tests for the significance of the difference between the two mothers.
As shown in Table 4, the two mothers both frequently provided or requested information about complicating action and orientation, which were the two frequently mentioned information types. But the mean frequencies of the two narrative
information types in the two mothers’ utterances were quite different. Z’s mother specifically focused on orientation (M = 10.1), with more than twice as many as complicating action (M = 4.9). In contrast, D’s mother placed more emphasis on complicating action (M = 7.6) than on orientation (M = 5.1). Furthermore, differences between the two mothers in orientation (t = -2.07, p < 0.05) reached statistical
significance. With regard to evaluation and non-narrative talk, the two mothers sometimes provided or requested these types of narrative information. As for speech
D’s mother Z’s mother
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
and coda, the occurrences of these information types were all very low in the two mothers’ utterances. In sum, Z’s mother and D’s mother emphasized different narrative information. Z’s mother focused on the orientation information of the narratives. In contrast, D’s mother, while providing or requesting some orientation information, focused on describing the actions and events in the narratives.
To sum up, the two mothers were similar in that they frequently elaborated on the narrative topic under discussion. That is, they demonstrated the style termed high-elaborative by Reese, Haden, & Fivush. But there were differences in how they
elaborated on the topic. Z’s mother was particularly concerned about orientation information. She frequently provided or requested information about the context of the narratives. In contrast, D’s mother placed more emphasis on complicating actions.
As a result, although the two mothers could be classified as having the same high-elaborative style, they differed in the type of narrative information that they
placed more emphasis on when co-constructing narratives about the past with their children.
4.2 Analysis of children’s narrative performance in researcher-child conversations
We firstly presented the number of narratives in researcher-child conversations, the number of total utterances per narrative, and the number of utterances that the
‧
researcher and children produced per narrative at three time points, as shown in Table 5.
Table 5. Number and length of narratives in researcher-child conversations at three time points
Note. The ratios of researcher/child utterances per narrative to total utterances per narrative are given in parentheses.
Table 5 shows that in conversations between the researcher and D/Z, the number of total utterances per narrative increased over time. In addition, the ratio of child utterances per narrative to total utterances per narrative for both children is higher than the ratio of researcher utterances per narrative to total utterances per narrative at three time points. That is, both children contributed more proportion of utterances per narrative than the researcher did at each time point. This suggested that the major narrator in researcher-child conversations was the children, not the researcher.
In the following section, we will present the findings of the two children’s Conversation between the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
38
narrative performance in researcher-child conversations in terms of the conversational contributions, which will be shown in section 4.2.1 and in terms of the narrative information types, which will be presented in section 4.2.2.
4.2.1 Conversational contributions
The utterances that the two children produced in the narratives when having conversations with the researcher were classified into six types. And the
conversational contributions were analyzed through the classification of utterance types. One way ANOVA was performed for each type to determine if there was significant difference in the two children’s utterance types over time. The mean frequencies and standard deviations of the utterance types per narrative at three time points (3;6, 3;9 and 4;0) are presented in Table 6 and Table 7. Table 6 and Table 7 exhibit the findings of the child D and the child Z, respectively.
‧
Table 6. Mean frequencies and standard deviations of D’s utterance types per narrative
* p < .05
Note. F values refer to tests for effects of time.
Table 7. Mean frequencies and standard deviations of Z’s utterance types per narrative
* p < .05
Note. F values refer to tests for effects of time.
D
Elaborations 3.57 0.53 6.66 3.26 10.83 5.56 6.50*
Repetitions 0.28 0.48 0.50 0.83 2.50 1.76 7.25*
Clarification questions 0.00 0.00 0.00 0.00 0.00 0.00 -
Evaluations 0.85 0.69 1.50 0.54 2.16 2.56 1.19
Associative talk 0.28 0.48 0.33 2.04 0.83 2.89 0.38
Off-topic talk 1.14 3.02 1.50 3.67 0.83 2.04 0.07
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
40
Table 6 and Table 7 revealed that within each time point, elaborations was the most frequently occurred utterance type for both children. In comparison to
elaborations, the mean frequencies of repetitions were much lower at each time point.
This implied that the two children both frequently elaborated on the topic under discussion. That is, they contributed unique pieces of information about past events more than they participated without adding new information in researcher-child conversations. As for clarification questions, both of the children did not produce this kind of utterance.
Across time, the mean frequencies of elaborations, repetitions, evaluations, and associative talk increased as the two children grew older. According to the analysis of variance performed, differences in frequencies of elaborations (D child: F = 6.50, p <
0.05; Z child: F = 9.46, p < 0.05) and repetitions (D child: F = 7.25, p < 0.05; Z child:
F = 4.34, p < 0.05) across the three time points reached statistical significance. This
indicated that the children provided significantly more pieces of new information when they were older, and they also showed a greater willingness to engage in the conversation even when not providing new memory information. With regard to off-topic talk, the mean frequencies decreased to the lowest at Time 3 (4;0) for both children, suggesting that the two children had better ability to focus on the topic under discussion when they were older.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
41
In sum, the findings suggested that the two children were very similar about the type of utterances that they contributed in the narratives when having conversations with the researcher. More specifically, they both frequently produced the type of elaborations, which was much higher than repetitions at each time point. In addition, there were significant differences in the amount of elaborations and repetitions over time. Therefore, in terms of the conversational contributions, the two children
displayed very similar performance. In the following, we will further analyze the type of narrative information that the two children included in the narratives to gain more insight into their narrative performance in conversations with the researcher.
4.2.2 Narrative information types
The utterances that the two children produced in the narratives when having conversations with the researcher were classified into six narrative information types.
One way ANOVA was performed for each type to determine if there was significant difference in the two children’s information types over time. The mean frequencies and standard deviations of the information types per narrative at three time points (3;6, 3;9 and 4;0) are presented in Table 8 and Table 9. Table 8 and Table 9 exhibit the findings of Child D and Child Z, respectively.
‧
Table 8. Mean frequencies and standard deviations of D’s information types per narrative
* p < .05
Note. F values refer to tests for effects of time.
Table 9. Mean frequencies and standard deviations of Z’s information types per narrative
* p < .05
Note. F values refer to tests for effects of time.
D
Complicating action 2.14 1.06 3.50 1.97 7.33 3.01 10.20*
Orientation 0.57 0.78 2.16 1.47 2.83 3.37 1.99
Evaluation 0.85 0.89 1.50 1.37 1.83 2.31 0.62
Speech 0.00 0.00 0.00 0.00 0.33 0.51 2.73
Coda 0.28 0.48 0.00 0.00 0.00 0.00 2.02
Non-narrative talk 2.28 3.09 3.33 3.77 4.83 4.70 0.70
Z
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
43
Table 8 revealed that within each time point, complicating action was the most frequently included information type for Child D. And non-narrative talk was the second most frequently included information. As for Orientation and evaluation, Child D sometimes produced utterances about these information types. With regard to speech and coda, the mean frequencies were quite low. Across time, the mean
frequencies of complicating action, orientation, evaluation, speech, and non-narrative talk increased as Child D grew older. According to the analysis of variance performed, differences in mean frequencies of complicating action (F = 10.20, p < 0.05) across the three time points reached statistical significance. This indicated that Child D produced significantly more utterances about the events or actions which occurred in the narratives when she was older.
Table 9 showed that the information type which had the highest mean frequency at Time 1 (3;6) and Time 2 (3;9) was different from the one at Time 3 (4;0). That is, complicating action was the highest one at the first two time points, while at Time 3 (4;0), orientation became the highest one. This indicated that at first, Child Z placed more emphasis on complicating action than orientation, but afterwards she became more concerned about the orientation than complicating action. With respect to non-narrative talk, the mean frequency was quite high at Time 1 (3;6) but it decreased to 1.2 at Time 3 (4;0). As for evaluation, speech, and coda, Child Z sometimes
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
44
produced utterances about these information types at each time point. In addition, across the three time points, differences in mean frequencies of complicating action (F
= 7.93, p < 0.05), orientation (F = 6.69, p < 0.05), and speech (F = 4.22, p < 0.05) reached statistical significance. This indicated that Child Z produced significantly more utterances about the actions, context, or reported speech in the narratives when she was older.
To sum up, the results suggested that at Time 1 (3;6), the two children were not very different in the mean frequencies of orientation that were included in the narrative. But, at Time 3 (4;0), the mean frequency of orientation for Child Z was much higher than that for Child D. And there was significant difference in the mean frequency of orientation over time for Child Z (F = 6.69, p < 0.05), while there was no significant difference in the mean frequency of orientation over time for Child D (F = 1.99, p < 0.05). This suggested that Z increasingly paid much more attention to include orientation information than D did as she got older. Furthermore, at Time 1 (3;6) and Time 2 (3;9), D and Z were similar in that complicating action was higher than orientation. However, at Time 3 (4;0), Z’s orientation was higher than
complicating action, while D’s complicating action was still higher than orientation.
Therefore, in terms of the narrative information type, the two children displayed different performance at Time 3 (4;0) in that Z placed more emphasis on orientation
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
45
than complicating action, while D focused on complicating action just as she did at Time 1 (3;6) and Time 2 (3;9).
4.3 The effect of mothers’ ways of co-constructing narratives with children on children’s narrative performance
In this section, we aim to see whether differences between mothers’ ways of co-constructing narratives with children would be reflected in children’s narrative performance in conversations with the researcher. There are two sub-sections included.
The first section shows the effect of mothers’ conversational styles on children’s conversational contributions, and the second section presents the effect of mothers’
provision/request for narrative information types on children’s inclusion of narrative information types.
4.3.1 Effect of mothers’ conversational styles on children’s conversational contributions
Although the two mothers showed the same high-elaborative conversational style, there was difference between the two mothers’ percentage of the two utterance types (elaborations and repetitions) which were used to determine the conversational style. Hence, we were then interested in analyzing whether the difference between the two mothers’ percentage of elaborations and repetitions would be reflected in
children’s contributions of the two utterance types in conversations with the
‧
researcher. Table 10 shows the distribution of D’s mother’s elaborations and repetitions in conversation with D and the distribution of D’s elaborations and repetitions in conversation with the researcher. Table 11 presents the distribution of Z’s mother’s elaborations and repetitions in conversation with Z and the distribution of Z’s elaborations and repetitions in conversation with the researcher.
Table 10. Distribution of D’s mother’s and D’s elaborations and repetitions per narrative
Note. The mean frequency of each category per narrative is given in parentheses.
The total utterances per narrative of D’s mother are 18, and the total utterances per narrative of D at each time point are 6.14, 10.5, and 17.16, respectively.
D’s mother D
‧
Table 11. Distribution of Z’s mother’s and Z’s elaborations and repetitions per narrative
Note. The mean frequency of each category per narrative is given in parentheses.
The total utterances per narrative of Z’s mother are 20.6, and the total utterances per narrative of Z at each time point are 7.83, 18.33, and 33.4, respectively.
Table 10 showed that in D’s mother’s utterances, the percentage of the type of elaborations at Time 1 (3;6) was 50%, and Table 11 revealed that in Z’s mother’s utterances, the percentage of the type of elaborations at Time 1 (3;6) was 68.8%. This suggested that when having conversations with children, Z’s mother elaborated on the topic under discussion much more often than D’s mother did at Time 1 (3;6). As for children’s performance in conversations with the researcher, Table 10 showed that in D’s utterances, the percentage of elaborations at three time points was 58.3%, 63.4%, and 63.1%, respectively. And Table 11 revealed that in Z’s utterances, the percentage of elaborations at three time points was 68.1%, 81.8%, and 80.8%, respectively. These results indicated that at each time point, the percentage of Z’s elaborations was higher
Z’s mother Z
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
48
than the percentage of D’s elaborations. At Time 1 (3;6), D and Z did not differ much.
But at Time 2 (3;9) and Time 3 (4;0), the difference between the two children’s percentage of elaborations was more apparent. This suggested that the difference between the two mothers’ elaborations was reflected in their children’s contributions of elaborations more obviously at later time points.
With regard to the type of repetitions, Table 10 showed that the percentage of D’s mother’s repetitions at Time 1 (3;6) was 7.2%, and Table 11 revealed that the
percentage of Z’s mother’s repetitions at Time 1 (3;6) was 9.3%. This suggested that when having conversations with children, Z’s mother repeated her own utterances a
percentage of Z’s mother’s repetitions at Time 1 (3;6) was 9.3%. This suggested that when having conversations with children, Z’s mother repeated her own utterances a