Chapter 1 Introduction
1.3 Organization of the thesis
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
5
events will be discussed by examining their connectedness, types of connections, and markers for connectivity. Finally, the present study expects to answer the following research questions:
(1) How are events in the four-year-old children’s narratives connected?
(2) How do four-year-old children mark narrative connectivity with linguistic devices?
1.3 Organization of the thesis
The present study will investigate the connectivity in narratives produced by Mandarin-speaking four-year-old children. The thesis is organized as follows: Chapter 2 will present a review of related studies; Chapter 3 will describe the analytical framework and methodology of the present study; Chapter 4 will display research results; Chapter 5 will discuss the findings; and Chapter 6 will conclude the thesis with a summary, limitations, and suggestions for further research.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
6
Chapter 2 Literature Review
Research has established fundamental features of connectivity (e.g., Ellis, 1983; Berman
& Slobin, 1994). Cross-linguistic studies have revealed similarities and differences in the way linguistic units in texts connect among various languages, especially in narratives (e.g.,
Berman & Slobin, 1987, 1994; Bamberg & Marchman, 1990, 1991; Slobin, 1993). A number of these studies have investigated children’s acquisition and development of connectivity in
narratives (e.g., Peterson & McCabe, 1991; Berman & Slobin, 1994; Sah, 2013).
This chapter will review and present related studies on connectivity in the following order. Section 2.1 defines what connectivity is and introduces cognitive abilities of connectivity and the linguistic realization of connectivity. Section 2.2 discusses studies on connectivity in the genre of narratives, including narrative coherence, relating events in narratives, and cross-linguistic differences in narrative connectivity. Section 2.3 introduces features of connectivity in Mandarin Chinese. Finally, important studies related to children’s acquisition of connectivity in Mandarin Chinese, especially in narratives, are presented in Section 2.4.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
7
2.1 Connectivity
The term “connectivity” is used to represent the notion of relating linguistic units in a
text. To construct connectivity in a text, cognitive abilities and linguistic capacities are required. Connectivity is realized through additive, adversative, temporal, and causal aspects and through various linguistic devices in a text.
2.1.1 Coherence and connectivity
The concept of connectivity is regarded as similar to “coherence” (e.g., Ellis, 1983;
Samet & Schank, 1984). The notion of coherence is one of the properties of discourse structuring in general (Halliday & Hasan, 1976; Ellis, 1983; Samet & Schank, 1984; Berman
& Slobin, 1994). Halliday and Hasan (1976) described coherence as the “relations of meaning
existing within a text and defining a text” (p. 4). Accordingly, in terms of a coherent text, Samet and Schank (1984) put it another way, stating that “something is coherent if it ‘hangs together’; if it is unified” (p. 59). Moreover, “[t]he interpretation of coherence requires
linguistic units beyond the sentence proposition and is based on the interpretation of one sentence, clause, or utterance relative to another,” which is “the fundamental notion of
‘connectiveness’” (Ellis, 1983: 233). The present study has adopted the term “connectivity” to
denote the concept of connectiveness. The notion of connectivity is seen in many studies on
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
8
relating events in discourse or extended discourses (e.g., Ellis, 1983; Samet & Schank, 1984;
Trabasso & Sperry, 1985; Berman & Slobin, 1994; Sah, 2013). In terms of texts such as
narratives, Samet and Schank (1984) considered the notion of connectivity as “internal connectivity” (p. 59), which represents the function of connections between units smaller than
narratives. In other words, connectivity represents the relation between a linguistic unit and some other unit (Ellis, 1983).
To sum up, connectivity is crucial for understanding the coherence of a text (Ellis, 1983;
Samet & Schank, 1984). This implies that how a discourse is coherent is analyzable by specifying the connectivity among units of a text (e.g., Trabasso & Sperry, 1985; Diehl et al., 2006).
2.1.2 Cognitive abilities and linguistic capacities
To construct the connectivity of a text, corresponding cognitive abilities and linguistic capacities are required (deBeaugrande, 1980; Ellis, 1983; Kemper, 1984; Berman & Slobin, 1994). A text should be produced following a topic; thus, the notion of topicality has been used to define the connectivity of sentences and the coherence of discourse (Ellis, 1983).
DeBeaugrande (1980) indicated that topicality lies in both sequential and conceptual connectivity. Sequential connectivity means that to elaborate on a topic, utterances should be
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
9
organized; thus, back-tracking abilities, on-line verbal production, explicitly stating relations between clauses, local linking, and integration and organization of the text units are necessary (deBeaugrande, 1980; Ellis, 1983; Berman & Slobin, 1994). Conceptual connectivity refers to different levels of abstraction from sentence topicality; thus, elements in a discourse should relate to the topic and relate to each other in terms of the elaboration of the topic (Ellis, 1983;
Berman & Slobin, 1994).
One of the necessary bases in building these abilities is the concept of causality. Earlier researchers have suggested that causal relation is the basic foundation of argumentation and explanation (e.g., Kemper, 1984; Byrnes, 1991; Berman & Slobin, 1994; Hsu, 1996). The concept of causality emerges at the age of two (Hood & Bloom, 1979). Hood and Bloom found that two- to three-year-old children can put sentences together that include both cause and effect, without applying explicit linguistic devices to mark the relationship between cause and effect. In developing extended discourse ability, such as narrative ability, one must be equipped with the concepts of temporality and causality, which enable the construction of plots and themes (Kemper, 1984).
In the circumstance of storytelling narratives, a narrator must be able to introduce the protagonist’s action, reveal the protagonist’s motives, display the following actions or states, and show the results of the protagonist’s actions; these elements enable narrators to connect
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
10
events in stories and make unified narratives (Samet & Schank, 1984; Berman & Slobin, 1994). Many researchers have indicated that the concept of causality plays a role in narrative skill and narrative coherence in terms of telling a story (e.g., Applebee, 1978; Samet &
Schank, 1984; Peterson & McCabe, 1991; Shapiro & Hudson, 1991; Berman & Slobin, 1994).
For example, Berman and Slobin (1994) elicited narratives with the aid of a picture book called Frog, Where Are You? (Mayer, 1969). In the picture book, the first three pictures display (1) a boy and a dog looking at a frog in a jar, (2) a frog climbing out of the jar while the boy and dog are sleeping, and (3) the awakened boy and dog and an empty jar. Still other series of events exist in the story, and narrators with different levels of cognitive maturity might relay different interpretations of the events in terms of their causal relations. Berman and Slobin indicated that young children were found to have had no difficulty in recognizing the above scenes, but they might have been unable to figure out what happened to the protagonists in the latter series of pictures, which features (1) the boy climbing on a rock, (2) the boy clutching what seems to be branches of a tree, and (3) the branches lifting, which are actually antlers on the head of a deer.
Temporality is encoded in causality (Byrnes, 1991; Hsu, 1996; Wu & Tsai, 2006). The use of causal connectives not only expresses the causal relation between two clauses but also signals their temporal relation. For example, “if” encodes the co-occurrence of cause and
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
11
effect, while “because” encodes the inconsistency of cause and effect in time, namely the
cause brings about the effect (Byrnes, 1991). The expression of causality between events lies on understanding the principle of covariance, namely their temporal priority continuity (Byrnes, 1991; Hsu, 1996; Wu & Tsai, 2006). Other researchers have posited that temporal succession characterizes causal connections (Trabasso & Sperry, 1985; Trabasso & van dan Broek, 1985). In addition, many researchers have found that temporality plays a role in narrative skill (e.g., Applebee, 1978; Peterson & McCabe, 1991; Shapiro & Hudson, 1991;
Berman & Slobin, 1994; Chang, 2004; Sah, 2007, 2013).
2.1.3 Linguistic devices for connectivity
Berman and Slobin (1994) defined connectivity as a structured characteristic of events in narratives and they described connectivity as “knitting the fabric of narrative discourse” (p.
19). Berman and Slobin suggested that formal expressions of connectivity include syntactic conjunctions, syntactic subordinations (such as relative clauses), nonfinite verb forms, nominalizations, and topic ellipses. These formal expressions belong to the cohesive devices proposed by Halliday and Hasan (1976). In their influential work on cohesive devices in English, Halliday and Hasan investigated the linguistic system that constructs the unity and consistency in a text. The cohesive devices were categorized into five classes: conjunctions,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
12
substations, ellipses, references, and lexical cohesion. Various types of conjunctions, namely various functions that connectives serve, include additive, adversative, temporal, and causal relations (Halliday & Hasan, 1976; Peterson & McCabe, 1991; Hsu, 1996). An additive
relation is the coordination of sentences or clauses when extra information or exemplification is added. An adversative relation refers to something “contrary to expectation”; that is, the
second-occurring linguistic element is contrary to the first-occurring one in meaning. A temporal relation, often shown in two successive sentences, is the relation in time in terms of semantic content; for example, simultaneousness, sequenceness, or perviousness. Common expressions are “then,” “afterward,” “simultaneously,” “finally,” etc. A causal relation
includes result, reason, purpose, conditional, and respective relations between linguistic elements. They are often expressed through “so,” “thus,” “consequently,” “because of that,”
“then,” “under those circumstances,” “in this respect,” etc. (Halliday & Hasan, 1976).
The use of connectives has been a prevailing issue in the study of language in recent decades (e.g., Schiffrin, 1986; Peterson & McCabe, 1988, 1991; Biq, 2001; Wang & Huang, 2006). Peterson and McCabe (1991) elicited personal narratives from children three-and-a-half years old to nine-and-a-half years old. They investigated connective use and how the use of this linguistic device reflected narrative macrostructure. Their results showed that most of the uses of the connectives “because,” “and,” “so,” “but,” and “then”
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
13
appropriately matched their canonical meanings, although a number of them did not.
Moreover, these connectives not only marked semantic relations but also functioned
pragmatically, such as for utterance initiation. A feature of these connectives was that most of them served various functions. For example, “and” marked the temporal succession between
statements and coordinated events, and it was often applied in presenting physical causation between events. Peterson and McCabe also found that even though narratives are texts that use connectives heavily, intersentential relations were sometimes built in narratives without any use of markers. Peterson and McCabe further designed three versions of narratives to play to college students and asked them to rate their comprehensibility. The three narratives were designed with both the presence and the absence of connectives. The results showed that
comprehensibility did not differ much regarding the presence or absence of connectives.
Grices’ (1975) Maxim of Orderliness was adopted in their explanation:
[T]wo sequentially stated event clauses are assumed by the listener to be an accurate reflection of the order of occurrence of those events, that is, the act of stating or describing one event before the other is automatically assumed to mean that the event mentioned first actually occurred first (p. 31).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
14
In sum, connectives are considered obligatory in the production of a well-formed narrative but are not obligatory in the inference and comprehension of a narrative; moreover, connectives are obligatory only when the chronological order of events is stated in a violated order (Peterson & McCabe, 1988, 1991).
2.2 Connectivity in narratives
Earlier researchers have revealed that connectivity is a crucial element in constructing narratives (e.g., Labov, 1972; Samet & Schank, 1984; Trabasso & Sperry, 1985; Peterson &
McCabe, 1991; Shapiro & Hudson, 1991). Labov (1972) defined a minimal narrative as “a sequence of two restricted/independent clauses which are temporally ordered” (p. 360). A narrative, according to Peterson and McCabe (1991), is “a recounting of events that follow each other in time” (p. 30). A frequent method adopted in narrative studies is storytelling,
either telling fictional stories or personal experiences (e.g., Applebee, 1978; Trabasso &
Sperry, 1985; Peterson & McCabe, 1991; Shapiro & Hudson, 1991; Berman & Slobin, 1994).
2.2.1 Narrative connectivity and narrative structure
Researchers have described the connectivity in narratives in various ways, such as “a matter of the causally connected story-line at the center” (Samet & Schank, 1984: 65),
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
15
“knitting the fabric of narrative discourse” (Berman & Slobin, 1994), and “the interconnection and integration of the content of a narrative” (Nicolopoulou, 2008: 300). In sum, narrative
connectivity is a structured characteristic of events.
Researchers have proposed many ways to represent narratives based on the examination of relations between narrative elements (Applebee, 1978; Kemper, 1981, 1984; Trabasso &
Sperry, 1985; Trabasso & van dan Broek, 1985; Diehl et al., 2006). Applebee (1978) developed a six-stage narrative model based on Vygotsky’s (1962) stages of concept
development. Applebee examined stories told by children aged two to five and concluded that there are six stages to a story’s structure. The six stages showed a path similar to the general
developmental order for children proposed by Vygotsky (1962). In resemblance to Vygotsky’s stages of concept development, Applebee called the stages heaps, associative sequences, primitive narratives, unfocused chains, focused chains, and true narratives.
Trabasso and Sperry (1985) proposed procedures for deriving a causal network to assess connectivity between events in a narrative. In their study, a pause unit in a story narrative was considered an event. Trabasso and Sperry identified causal relations between all pairs of
events in a narrative through logical criteria of necessity (Mackie, 1980). A counterfactual test was used to examine necessity and dependency between events: “If not A then not B.” Given
that event A is the necessary element for the occurrence of event B, if event A has not
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
16
occurred, it is impossible for event B to occur. Accordingly, a pair of events that passed the test was identified as causally related.
To facilitate the judgment of relation between a pair of events, six kinds of causal relations described by Warren et al. (1979) were employed by Trabasso and Sperry (1985): (1) motivation, an event that causes a goal-directed action; (2) psychological causation, an event that causes a non-goal-directed action; (3) physical causation, naïve interpretations of the physical world or mechanical causality between objects and/or people; (4) enablement, actions or occurrences or states that are necessary but not sufficient to cause other actions or states; (5) temporal succession; and (6) temporal coexistence, any pair of linguistic units possessing a temporal sequence or coexistence but failing to pass the necessity test. Based on the connections between events, Trabasso and Sperry derived causal networks for narratives.
Moreover, they identified whether the event depicted by a statement was in a causal chain, which consists of opening, closing, and continuing. Trabasso and Sperry found that the number of direct operative links a statement has to other statements and whether a statement is in a causal chain were both given consideration in the assessment of connectivity in narratives. Trabasso and Sperry further compared an earlier study on the importance of each event to the dependencies between the particular event and others, namely the connections, and to whether or not an event was in a causal chain. In conclusion, the importance of story
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
17
events was highly related to causal relatedness.
Diehl et al., (2006) calculated the number of connections between related narrative events that had been identified following Trabasso and Sperry’s (1985) procedures. Events
coincided with communication-units (c-units), which consist of a verb and its arguments (Strong, 1998). The number of connections that en event had to other events was categorized into four levels: isolated events that have no causal connection to or from other events in a narrative (C0); an event that has only one connection to or from the other events in a narrative (C1); an event that has two connections with other events in a narrative (C2); and an event that has three or more connections with other events in a narrative (C3+). Levels of connectedness for three connections and above were collapsed because they were seldom found to be produced by children. Diehl et al. (2006) investigated narrative coherence in children with ADS (autism spectrum disorders) and children in a control group, with ages ranging from six to fourteen. In terms of narrative connectivity, relative to the children in the control group, the children in the ADS group were found to have produced a higher proportion of isolated events with no connections to or from the other events. They also had a lower proportion of highly connected events. Diehl et al. indicated that these patterns signaled that children with ADS produced less coherent stories relative to the children in the control group.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
18
Applebee’s (1978) model and Trabasso and Sperry’s (1985) causal network were both
applied in later studies that investigated narratives via narrative structure analysis. For example, Applebee’s (1978) model was adopted in Kemper’s (1984) study, along with high
point analysis (Labov, 1972). Trabasso and van dan Broek (1985) proposed a recursive transition network that integrated story grammar (Stein & Glen, 1979), causal network, causal
chain, and problem-solving approaches. Sah (2013) adopted Trabasso and Sperry’s (1985) and Diehl et al.’s (2006) methods of analysis, and further discussed the causal relation between
narrative events with the approach of goal plan analysis.
2.2.2 Developmental path of narrative connectivity
Given that researchers used various ways to investigate narrative connectivity, they interpreted children’s developmental path of narrative connectivity differently. In terms of Applebee’s (1978) six-stage narrative model,1 children’s ages corresponded to these six
1
Figure 1. Applebee’s (1978) model for developmental stages of narratives (p. 58)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
19
stages as they developed. In the first stage, heaps, two- to three-year-old children can list unrelated (linguistically not marked as related) characters, actions, or events. In the second stage, associative sequences, children three years of age are aware that there is a characteristic such as a central theme in the story, but they are unable to associate the sequence of events into a plot and merely describe what events have happened. In the third stage, primitive narratives, four- to four-and-a-half-year-old children are aware that there is a concrete core to the narrated events and are able to relate those events. Primitive story grammar is seen in the initiating event, action, and consequence. In the fourth stage, unfocused chains, four-and-a-half- to five-year-old children can present events related directly to each other.
However, they still could not conceive an overall plot, and thus the consistency of characters or the main theme in a story was missing. In the fifth stage, focus chains, four-and-a-half- to five-year-old children are able to describe a chain of events and how the chain of events relates to a central core, and the process of forming a narrative emerges. Finally, in the sixth stage, narratives, five- to seven-year-old children are aware that stories have a main theme, plot, and characters. They are aware of the hierarchical order and relations of events and are able to express them in a story. In the procedure, Applebee found that centering and chaining were two basic mechanisms in the process of forming narratives and that children often linked story elements through time sequence and causality.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
20
A developmental path was also concluded by Berman and Slobin (1994) in their study of children’s and adults’ performance of relating events in storytelling. In the first phase, very
young children tended to treat each scene as an isolated event, and they described each individual picture separately. For example, three-year-old children’s narrative style was deictic, as they usually identified, named, described and pointed at objects. In the second phase, three-year-old children could connect events sequentially. The relation chain between events was a linear chain, meaning that they could connect events in a temporal sequence and they could also introduce the characters and their possessive relationships. The four-year-old children were regarded to be in developmental transition between the three-year-old and the
older five-year-old preschoolers. The four-year-old children started to think about and relate the events from a protagonist’s perspective. They could also understand the protagonist’s
beliefs, mental states, and intentions. In the third phase, events were causally associated. For example, five-year-old children could introduce characters and indicate the characters’ goals and actions. However, causal relations in the five-year-old children’s narratives were still considered to be simple. In the last phase, the causal interrelations became hierarchically structured. For example, in the adults’ narratives, the purpose, actions, and outcomes became more complicated and richer in detail and emotional reactions.
beliefs, mental states, and intentions. In the third phase, events were causally associated. For example, five-year-old children could introduce characters and indicate the characters’ goals and actions. However, causal relations in the five-year-old children’s narratives were still considered to be simple. In the last phase, the causal interrelations became hierarchically structured. For example, in the adults’ narratives, the purpose, actions, and outcomes became more complicated and richer in detail and emotional reactions.