Developmental Psychology 2005, Vol. 41, No. 3, 464 – 478

Copyright 2005 by the American Psychological Association 0012-1649/05/$12.00 DOI: 10.1037/0012-1649.41.3.464

Narrative Development in Bilingual Kindergarteners: Can Arthur Help? Yuuko Uchikoshi Harvard University This study examined the effects of the children’s TV program Arthur on the development of narrative skills over an academic year for Spanish-speaking English-language learners. In October, February, and June of their kindergarten year, children were asked to tell a story, in English, prompted by 3 pictures. Before the 2nd and 3rd assessments, half of the 108 children were randomly assigned to view Arthur 3 times a week during school hours, and the other half, which formed the control group, viewed the children’s program Between the Lions on the same schedule. Individual growth modeling analysis showed that children who viewed Arthur had steeper growth trajectories than those who viewed Between the Lions. Boys displayed better English narrative skills than girls but no difference in narrative growth rate. The results suggest that certain educational TV programs can assist in some aspects of the language development of bilingual children. Keywords: narrative development, educational television, English language learners, kindergarten, growth modeling

ability to mark the significance of events and their ability to represent informational context in expository discourse predicted 8-year-olds’ reading comprehension skills. Dickinson and Tabors (2001) found kindergarten narrative production scores to be correlated with fourth- and seventh-grade receptive vocabulary as well as reading comprehension. In addition, Mason, Stewart, Peterman, and Dunning’s (1992) study showed that at-risk children without narrative problems do much better on later reading achievement than do children with narrative delays. Given that narratives have been identified as the critical link to later school success (Paul & Smith, 1993), narratives are often demonstrated in classroom situations such as show-and-tell, even by kindergarteners and first graders. Yet, although children from mainstream backgrounds enter school familiar with the type of narrative structure that is valued in schools (Heath, 1982), children from other backgrounds, such as ELL children, do not. Research shows that narrative skills are associated with cultural background and ethnic-group membership (Dart, 1992; Heath, 1982, 1986; Minami & McCabe, 1991; Ninio, 1980; Shiro, 1995; Silva & McCabe, 1996; Tannen, 1980; Wang & Leichtman, 2000). For example, Latino children tend not to follow a linear model when telling narratives (Rodino, Gimbert, Perez, Craddock-Willis, & McCabe, 1991, as cited in Silva & McCabe, 1996). They tend to deemphasize event sequencing yet emphasize description and evaluation (Silva & McCabe, 1996). Latino children also tend to produce narratives focused on personal and family relationships rather than on events as monolingual English-speaking children do (Silva & McCabe, 1996). Thus, Spanish-speaking ELL children are at a disadvantage from school entry as a result not only of their weak English skills but also of their different narrative styles. Consequently, they are projected to have later academic problems. In order to prevent this, interventions that can be conducted prior to school entry, particularly for the growing Spanish-speaking ELL population, need to be created and examined. More research is crucial to give these ELL children an equal start in first grade.

The number of children who have limited proficiency in English in U.S. schools has risen dramatically over the past 2 decades. Moreover, 53% of these children are concentrated in kindergarten through Grade 4; nationally, at least 1 in 12 kindergarteners comes from a home in which English is not the primary language (August & Hakuta, 1997). In addition, 77% of the English-language learner (ELL) children in the United States have Spanish as their first language (Office of English Language Acquisition, Language Enhancement, and Academic Achievement for Limited English Proficient Students, n.d.). Despite a dramatic increase in the number of young children learning English as a second language, there has been little systematic research on the development of their English narrative skills. Narrative skills have been pointed out as being strong predictors of later language and literacy achievement for monolingual English children (e.g., Dickinson & Tabors, 2001; Griffin, Hemphill, Camp, & Wolf, 2004; Paul & Smith, 1993). Griffin et al. (2004) showed that children’s oral narrative abilities at age 5 predicted reading comprehension skills at age 8. In particular, 5-year-olds’

Yuuko Uchikoshi, Graduate School of Education, Harvard University. This research was supported by dissertation grants from the International Reading Association and the Harvard University Graduate School of Education. I am thankful to Catherine Snow, Terry Tivnan, Maria Carlo, and Lowry Hemphill for their valuable comments on previous versions of this article. I also thank Carlo Santos, Frida Gomez, Kim Keith, Elisa Jazo, Kaytie Dowcett, Jill Jacobs, Jody Clarke, Jude Higdon, Patti Sullivan-Hall, Sara Roberts, Robyn Viloria, and Elizabeth Willmott for their assistance with data collection and Elisabeth Duursma for her assistance with coding reliability. In addition, I am grateful to television station WGBH as well as the principals, teachers, staff, and children of the schools for their support and participation in this study. Correspondence concerning this article should be addressed to Yuuko Uchikoshi, who is now at the School of Education, University of California, Davis, 2059 Academic Surge, Davis, CA 95616. E-mail: yuchikoshi@ ucdavis.edu 464

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS

Labov (1972) defined a narrative as at least two sequential independent clauses describing a single past event. Narratives are a form of decontextualized extended discourse. The ability to produce a narrative demonstrates a child’s ability to sustain talk about the world beyond the here and now (Snow, Tabors, Nicholson, & Kurland, 1995). Furthermore, narratives are not streams of unrelated words or sentences. They require the use of several utterances or turns that must be linked to build a coherent linguistic structure (Snow et al., 1995). Story structure (Chang, 2004; Labov, 1972; Peterson & McCabe, 1983), evaluation (Chang, 2004; Labov, 1972; Peterson & McCabe, 1983), temporality and reference (Chang, 2004; Labov, 1972; Peterson & McCabe, 1983), and storybook language (Purcell-Gates, 1988, 2001) are the key dimensions of narrative skills. Much of the past literature in this area has focused on the development of these dimensions in the narratives of English monolingual children. The components of a well-formed story structure include an abstract (a summary of the narrative or a title), an introduction (a conventional opening to the narrative), an orientation (a description of the characters, setting, time, and activity to set the stage for the narrative), events (actions that advance the storyline, including the problem), a resolution (a termination of complicating events), and a coda (an ending of the narrative; Chang, 2004; Labov, 1972; Peterson & McCabe, 1983; Snow et al., 1995; Willenberg & Kang, 2001). It is also important to focus on the evaluation component of narratives. When telling a story, the narrator not only tells the listener what happens in the story but also talks about the meaning of those events. In other words, the narrator includes evaluation devices to signal the point of the story from the narrator’s perspective (Chang, 2004; Labov, 1972; Peterson & McCabe, 1983). Evaluative devices include intensifiers (e.g., “all the other bears”), adjectives (e.g., “the little bears”), negatives or defeats of expectations (e.g., “The bear could not reach the kite”), references to emotional states or cognitions (e.g., “The bear was scared”), references to physical states (e.g., “The bear was hurt”), intentions (e.g., “The bear wanted to get the kite”), causal markers (e.g., “The bear fell because he slipped out of the tree”), and words with high evaluative content (e.g., “The bear came crashing to the ground”; Beck, Coker, Hemphill, & Bellinger, 2003; Peterson & McCabe, 1983; Willenberg & Kang, 2001). Peterson and McCabe (1983) found 21 types of evaluation devices in the narratives of 4- to 9-year-old American children, with the older children using a larger variety of evaluations than the younger children. Temporality and reference are the two main elements in the organization of a narrative. In order for narratives to be coherent and follow a time line, linguistic temporal devices such as connectives are used. Peterson and McCabe (1991) found that even 31⁄2-year-old children used connectives such as and in their narratives. In addition, how the narrator first introduces the story characters in the narrative influences the organization of the story (Beck et al., 2003). Ideally, the character should be introduced in a way that does not assume any prior knowledge on the part of the listener. For example, a story that begins with an unspecified pronoun such as he would have an unclear beginning. A narrative that starts with a nonpresupposing introduction using an indefinite article and a noun, such as a bear, would be more coherent (Beck et al., 2003; Willenberg & Kang, 2001).

465

In addition, when children pretend to read storybooks out loud, they use a discourse different from the one they use when they are telling a personal narrative. That is, when they pretend to read, children use the language of storybooks (Purcell-Gates, 1988, 2001). They tend to use a lexicon that is more literary and varied as well as to use the syntax found in written texts. For example, narrators may include direct or indirect quotes as well as conjoined phrases. However, not all of these traits appear in the narratives of children before the age of 6 (Peterson & McCabe, 1983). Narratives of 4- and 5-year-old monolingual children contain little evaluation of story events (Bamberg & Damrad-Frye, 1991; Eaton, Collis, & Lewis, 1999; Peterson & McCabe, 1983). The most common narrative structures of 4-year-old monolingual Englishspeaking children have been described as either leap-frog narratives, in which children jump from one event to another and leave out major events, or chronology narratives, in which they provide only a simple recounting of successive events (Peterson & McCabe, 1983, pp. 43– 45). At age 5, the most common narrative patterns are the chronology narrative and the ending-at-the-highpoint narrative, in which there is no resolution (Peterson & McCabe, 1983, pp. 41– 45). At age 6, the classic narrative pattern finally becomes dominant (Peterson & McCabe, 1983, pp. 36 – 41). The classic narrative pattern has been defined as one in which the narrative is built around a high point. “The narrator builds up to a high point, evaluatively dwells on it, and then resolves it” (Peterson & McCabe, 1983, p. 36). Yet the narrative patterns of ELL children—in particular, those of Spanish-speaking ELL children— and how their English narratives develop are still unexplored. Most studies of Spanish–English bilinguals’ narrative performance, such as the one by Gutie´rrez-Clellen (2002), have focused on crosslanguage comparison, not on charting the developmental process involved. Various environmental factors, such as parental responsiveness, affect the development of children’s narrative skills (Peterson, Jesso, & McCabe, 1999). The types and number of parent– child conversational exchanges during book reading shape children’s narrative development (McCabe & Peterson, 1991; Peterson & McCabe, 1992; E. Reese, Haden, & Fivush, 1993). Some parents make detailed references to past events and ask multiple questions, whereas others ask simple questions and frequently switch topics. These differences in parental style have been shown to influence both the quantity and quality of narratives produced by preschool children. Frequency of book reading is also associated with narrative development (Purcell-Gates, 2001). Books display narrative structures particularly clearly, and furthermore, some interventions use book reading as a carrier for their interactions. Well-read-to monolingual kindergarten children can learn linguistic registers that are specific to social contexts and use more formal and booklike narrating language when pretending to read a storybook (Adams, 1990; Heath, 1982; Ninio, 1980; Purcell-Gates, 1988, 1991, 1992, 1996, 2001; Purcell-Gates & Dahl, 1991; Scarborough & Dobrich, 1994; Scarborough, Dobrich, & Hager, 2001; Sulzby, 1985). Preschool-age children whose parents spend more time in narrative conversation, in asking more open-ended and context-eliciting questions, and in encouraging longer narratives through bookreading sessions tend to produce more context-setting descriptions

466

UCHIKOSHI

about when and where the events in a narrative took place (Peterson et al., 1999). However, research shows that Latino preschool children are less likely to have frequent book-reading experiences than are White, non-Latino children (Nord, Lennon, Liu, & Chandler, 1999). Given the evidence from past research with monolingual children demonstrating that exposure to narratives through book reading during the preschool years is positively associated with narrative skills in kindergarten and subsequent literacy outcomes, the impact of mechanisms for supplementing Spanish-speaking ELL children’s access to experiences with narrative needs to be evaluated. Educational TV programs, particularly shows based on books, could provide children with the desired extra exposure to language and literacy environments as well as to English narrative conventions.1 Past research suggests that educational TV programs, such as Sesame Street and Between the Lions, can be a source of language learning for monolingual English-speaking children (e.g., Linebarger, 2000; Rice, Huston, Truglio, & Wright, 1990; Rice & Woodsmall, 1988; Van Evra, 1998). Children have acquired new vocabulary and developed phonological awareness skills by viewing these shows. In addition, child language research shows that the number of conversations and the variety of words that children hear affect the speed of their language and literacy growth (Tabors, Beals, & Weizman, 2001). Studies show that, on average, children between the ages of 3 and 5 years watch 19 –20 hr of TV per week (Huston, Wright, Rice, Kerkman, & St. Peters, 1990); it is important to know whether, if these hours were concentrated on educational TV programs, it would assist in the English-language and literacy development of ELL children. Children from nonmainstream backgrounds might become more accustomed to mainstream English narrative styles that are prevalent in school settings by viewing educational TV programs that are based on books and follow the classic narrative pattern. Like book reading, TV viewing can be a routinized activity. Many shows follow a routine format. Moreover, children can view reruns or videotaped versions of the same episode. In addition, researchers have found that with monolingual English-speaking children, parent– child interactions around TV programs can replicate aspects of interactions around books (Lemish & Rice, 1986). Yet there have been no studies investigating the impact of book-based educational programs on either monolingual or ELL children’s narrative development. A representative of a bookbased show is Arthur. Arthur is a 30-min educational program appearing on Public Broadcasting Service (PBS) stations across the country and targeted to audiences of preschool and kindergarten children. Arthur tells a story to the viewers in each episode and is popular with kindergarten-age children. In Arthur, each episode presents two stories, each with a plot, a conflict, and a resolution. As each story is based on a storybook, each story follows the components of a well-formed story structure as defined by Labov (1972) and Peterson and McCabe (1983). The Arthur TV series exposes children to various stories with moral points of interest to them. The show is about growing up. The characters in Arthur learn to make thoughtful decisions and resolve problems in each episode. The problems Arthur and his friends face are similar to the ones the viewers may face at home and at school. In addition, evaluation is emphasized through the characters’ speech and actions.

In this study, my aim was to investigate whether the book-based educational TV program Arthur would influence narrative development among ELL children from Spanish-speaking homes. I attempted to answer the following question: Does viewing Arthur have a beneficial effect on narrative skills so that ELL children can be better prepared for first grade? To control for effects of watching educational TV, I used Between the Lions, another book-based educational TV show, as a control.2 Between the Lions is also a 30-min book-based educational program that is broadcast on PBS stations across the country and targeted to kindergarten-age children. Between the Lions introduces a book in each episode, but instead of focusing explicitly on the narrative, the show puts more weight on text structure, individual words, and other print features (Linebarger, 2000). Phonological awareness and reading fluency are heavily emphasized. Each Between the Lions episode follows a “whole–part–whole” framework, adopted as the approach to literacy instruction. The story line of each Between the Lions episode begins with a readaloud experience as the “whole,” in which portions of the text are displayed on the screen and words are highlighted as they are read (Rath, 2000). Then the “parts” are emphasized to point the viewers’ attention to such topics as phonological awareness, letter– sound correspondence, word meanings, punctuation, and other conventions of written English (Rath, 2000). Most of the show is centered on this “parts” part. At the end, the “whole” text is revisited and the “parts” are reviewed (Rath, 2000). Hence, Arthur focuses more on the narrative structure, whereas Between the Lions focuses more on phonics and story mechanics. By showing a program that is educational but not entirely focused on narratives, such as Between the Lions, to the control group, I could control for any effects that could be attributed to TV viewing. Thus, in this study I examined the effect of viewing Arthur versus Between the Lions on ELL children’s English narrative growth within the context of a standard kindergarten program. In addition, I investigated the patterns of growth in oral English narrative skills in young ELL kindergarteners. Such research is needed to fully understand the narrative skills development of Spanish-speaking ELL children and to prepare them for school entry. Moreover, in this study I used individual growth modeling techniques (Singer & Willett, 2003; Willett, 1994) to analyze the children’s narrative development. As individual growth modeling makes use of repeated waves of data and conceptualizes change as a continuous process of development (Willett, 1994), it yields a more accurate picture of change over time than do traditional techniques, such as ordinary regression methods. 1

Educational TV refers to shows that have a core educational or informational purpose. In 1990, the Federal Communications Commission passed the Children’s TV Act, which required commercial broadcasters to air programming that has a core educational and informational purpose targeted to children under age 16 (Hill-Scott, 2001). Such programs must be regularly scheduled, weekly programs of at least 30 min and be aired between 7:00 a.m. and 10:00 p.m. 2 Between the Lions focuses on phonics and story mechanics. How Between the Lions viewing affects the development of ELL children’s phonological awareness and story mechanics, with Arthur viewing as a control, is not discussed in this article.

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS

467

average parental educational level being some secondary education. The top 10% of all parents had received some higher education, whereas the bottom 10% had not completed primary education. The remaining 80% fell somewhere in between, with the majority completing primary education. The English vocabulary levels of the target children were measured with the Peabody Picture Vocabulary Test—Third Edition (PPVT–III; Dunn & Dunn, 1997) in October before the intervention started. The results of the test indicated that, on average, in October these native Spanish speakers scored at the level of a monolingual English child 3 years 2 months of age (Dunn & Dunn, 1997). The Spanish vocabulary levels of the target children were also measured in October with the Test de Vocabulario en Ima´genes Peabody (TVIP; Dunn, Padilla, Lugo, & Dunn, 1986). The results of this test revealed that, on average, these native Spanish-speaking entering kindergarteners achieved scores that would be expected from Spanish monolingual children residing in Mexico and ranging in age from 4 years 8 months to 5 years 0 months, according to the age norms of the test (Dunn et al., 1986). As the average age of the children in this study was 5 years 7 months, these ELL children had vocabulary levels that were slightly lower than those of their Spanish monolingual counterparts. (See Table 1 for a summary.) Raw scores are reported for initial vocabulary levels in both Spanish and English for later use in growth modeling analysis. It should be noted that there are fewer items on the TVIP than on the PPVT–III, which may explain the lower Spanish raw scores.

Method Participants A total of 108 children (47 girls and 61 boys) attending six public schools in a large urban district located on the East Coast participated in the study. The average age of these children in October was 5 years 7 months. The average age for the boys was 5 years 7 months; for the girls it was 5 years 6 months. Spanish–English bilingual kindergarten classrooms were selected from these schools. In all classrooms, instruction occurred in both English and Spanish. All kindergarten teachers were fluent in both English and Spanish. All children were from primarily Spanish-speaking homes and lived in neighborhoods heavily populated by Spanish-speaking people. Participants were recruited through these kindergarten classrooms. School district demographics and school data indicate that 80% or more of the participating students qualified for free lunch. To gain background information about the students, I sent home with them a parental questionnaire in both English and Spanish (see Table 1 for a summary of these data). The number of older siblings of the target child ranged from 0 to 5, with the average child having 1.2 older siblings. Years living in the United States ranged from 3 months to 7 years, with the average being 4.9 years. The majority of those whose parents responded had been born in the United States; only 22% of these children had been born outside of the United States. Although variation in the number of children’s books in the home was large—from 0 to 300 books (SD ⫽ 35 books)—parents responded that on average there were 21, including both English and Spanish books. The mode was 10 books. Concerning prekindergarten experience, 56 of 108 parents responded that their target child had gone to either prekindergarten or a Head Start program. A third (34 of 108 parents) said they took their target child to libraries on a frequent basis. In terms of home viewing, 73.4% of the children viewed Arthur at home, whereas 41.9% watched Between the Lions at home. Information about the mothers was also collected. The educational levels of the mothers ranged from no education to professional degrees, the

Design On the basis of a stratified random sampling, half of the students in six classrooms (51 children) were assigned to watch Arthur during school hours, and the other half in the same six classrooms (57 children) were assigned to watch Between the Lions during school hours. In each classroom, the children were first grouped according to gender, and then they were rank ordered on the basis of their October English PPVT–III vocabulary scores. I then randomized assignment to the two viewing conditions,

Table 1 Background Information and Vocabulary Scores for All Children (N ⫽ 108) and for Children by Viewing Group and Gender Total Variable Parental educationa No. of older siblings Years lived in the U.S. No. of Spanish books at home No. of English books at home No. of total books at home

Between the Lions

Boys

Girls

M

SD

n

Range

M

SD

n

M

SD

n

M

SD

n

M

SD

n

3.69 1.16 4.89 12.40 9.01 21.47

1.69 1.16 1.80 27.32 11.22 35.07

93 92 99 91 92 92

0–8 0–5 0.3–7 0–250 0–55 0–300

3.76 1.09 4.87 7.39 5.60 13.11

1.76 1.27 1.85 6.79 5.44 9.55

45 44 48 46 47 47

3.63 1.23 4.90 17.51 12.57 30.02

1.63 1.06 1.78 37.77 14.29 47.70

48 48 51 45 45 45

3.49 1.19 5.02 8.73 8.06 16.94

1.79 1.12 1.69 9.14 10.16 17.72

53 53 57 52 53 53

3.95 1.13 4.71 17.28 10.31 27.51

1.52 1.22 1.96 40.16 12.54 49.26

40 40 42 39 39 39

No. of yes Preschool experience (yes/no) Library experience (yes/no) Arthur home viewing (yes/no) Between the Lions home viewing (yes/no)

Receptive vocabulary score Initial English vocabulary Initial Spanish vocabulary

Arthur

n

No. of yes

n

No. of yes

n

No. of yes

n

No. of yes

n

56 34 69

89 92 94

26 13 35

44 47 46

30 21 34

45 45 48

28 21 39

49 52 54

28 13 30

40 40 40

36

86

17

50

19

42

21

50

15

36

M

SD

n

Range

M

SD

n

M

SD

n

M

SD

n

M

SD

n

39.69 32.05

20.38 13.29

108 108

0–85 2–69

41.37 32.65

20.74 12.49

51 51

38.19 31.51

20.13 14.06

57 57

44.30 31.28

18.37 13.24

61 61

33.72 33.04

21.49 13.44

47 47

Note. Background information was obtained from parental questionnaires, and vocabulary scores from the Peabody Picture Vocabulary Test-Third Edition and the Test de Vocabulario en Ima´genes Peabody. a Parental education was on a scale from 0 to 8.

UCHIKOSHI

468

matching the children on their vocabulary scores, which yielded viewing groups with similar composition in gender and initial English vocabulary skills. Receptive vocabulary scores were chosen as a basis for stratification because children’s understanding of the shows would be most influenced by their English vocabulary. Both groups watched one 30-min episode three times a week in a classroom at school from October to the beginning of May, for a total of 54 episodes. A frequency of three episodes a week was chosen because of the importance of repetition in interventions (Galdwell, 2000), children’s liking for repetition and familiar events (Galdwell, 2000), and the feasibility and practicality of children viewing educational TV during school hours. Owing to time constraints and to keep the intervention consistent among classrooms, teachers and researchers only showed the videos and did not do follow-up activities based on the episodes with the children. All kindergarteners were assessed with the narrative tests described in the following section at three time points throughout the school year: October (before watching any episodes in the classrooms), February (after watching 27 episodes in the classrooms), and late May or early June (after watching an additional 27 episodes in the classrooms).3 The total testing time for each individual session was 30 – 45 min. Children were also assessed on English as well as Spanish receptive vocabulary knowledge.

Measures To measure narrative skills, I asked children to tell a “Bear Story” in English, prompted by three slides of pictures that depicted a family of teddy bears in an adventure involving a flyaway kite and a baby bear falling from a tree (from the SHELL [School-Home Early Language and Literacy] test battery in Snow et al., 1995). The child was allowed to look at the slides as long as needed but was asked to put them away before telling the story. However, when the child appeared hesitant to give a narrative, the child was allowed to see the “Bear Story” pictures as he or she related the story. Trained assessors transcribed the narratives dictated by the children on the spot. The children’s narratives were written down so that the assessor could read the story back to the child and confirm that the story had been correctly recorded. If the child began to describe the picture instead of providing a narrative, the assessor prompted the child with questions such as “What is happening in the story?” or “What happened next?” If the child appeared to be taking no definite course in the narrative or if the child ended the narrative abruptly, the assessors were instructed to ask, “Is that the end of the story?” or “How does the story end?”

(Beck et al., 2003; Snow et al., 1995; Willenberg & Kang, 2001): abstract, introduction, orientation, character delineation, problem, resolution, and coda. If the narrative contained all seven elements, the child was given a total of 7 points. The definitions of the elements, as well as examples from the actual data, are listed in the Appendix. Events coding. Narrative events relate the actions of the story characters and move the plot forward. Five categories of events from Willenberg and Kang (2001) were included in this section: bears playing, bears flying kite, bear climbed tree or bear attempts to get kite, bear fell or jumped, bear hurt or dead. Evaluation coding. Evaluative devices in narrative are used to signal the point of the story from the narrator’s perspective. Eight types of evaluation from Willenberg and Kang (2001) were included: intensifiers, adjectives, negatives or defeats of expectations, references to emotional states or cognitions, references to physical states, intentions, causal markers, and words with high evaluative content. See the Appendix for examples. Temporality and reference. Temporality and reference (e.g., use of connectives, first mention of story characters) were included from Willenberg and Kang (2001) as presented in the Appendix. Storybook language. The elements of storybook language examined came from previous studies concerning pretend book reading (Beck et al., 2003; Purcell-Gates, 2001; Snow et al., 1995). As most instances of storybook language contain syntactic constructions that are typically found only in older children (Labov, 1968, cited in Peterson & McCabe, 1983), only three elements of storybook language from Willenberg and Kang (2001) were examined in this study: direct or indirect quotes, -ly adverbs, and conjoined noun/verb or adverbial phrases (see the Appendix for examples). The points for each of these five dimensions of the child’s narrative were added up to form a combined narrative measure score. For reliability of coding, 20% of the narratives were randomly selected and independently scored by a trained researcher. The Cohen’s kappa statistic, which corrects for chance agreement (Bakeman & Gottman, 1997), was .82.

Statistical Analysis First, a descriptive analysis was conducted on the total number of words, the mean clause length, and the combined narrative measure. Then, to examine the difference in the level and rate of change on the combined narrative measure among individuals, I made use of individual growth modeling.

Total Number of Words and Mean Clause Length Because past research showed an increase in the length of stories as well as the syntactic complexity of utterances as children got older (Eaton et al., 1999; Karmiloff-Smith, 1985; Purcell-Gates, 2001), I calculated the total number of words and the mean clause length in each story. After separating each story into clauses, I calculated the mean clause length by dividing the number of clauses by the total number of words.

Combined Narrative Measure Each story was then coded along the following five dimensions: story structure, number of main events, evaluation, temporality and reference, and storybook language. This coding scheme was adapted from Willenberg and Kang (2001), whose coding checklists were based on previous Bear Story checklists from Snow et al. (1995) and Beck et al. (2003). Each category was coded for the presence or absence of each feature listed regardless of how frequently it occurred in the narrative. Story structure coding. Seven major types of narrative elements were categorized on the basis of an adaptation of high-point structure (Peterson & McCabe, 1983) and previous coding checklists used for the “Bear Story”

Results Descriptive Results In Table 2, the means and standard deviations for the total number of words, the mean clause length, and the combined narrative measure are presented for all children, and by viewing group and gender. The average for each of these three measures increased as the children approached the end of their kindergarten year. Individual variation was high for all three variables. The average total number of words increased by 43% from slightly over 20 words to 37 words, a growth of two thirds of a standard deviation. The average mean clause length went from 4 words per clause to 5 words per clause, an increase of nearly one half of a standard deviation. The average combined narrative measure score 3

Children were tested on a variety of literacy measures, but only narrative production is discussed here.

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS

469

Table 2 Means and Standard Deviations for Narrative Variables for All Children (N ⫽ 108) and for Children by Viewing Group and Gender Total Narrative measure Total no. of words October February May/June Mean clause length October February May/June Combined narrative measure October February May/June

Arthur

Between the Lions

Boys

Girls

M

SD

n

Range

M

SD

n

M

SD

n

M

SD

n

M

SD

n

21.91 26.18 37.33

21.80 20.15 32.54

99 108 102

0–96 0–108 0–253

21.92 28.29 42.66

19.10 20.42 39.44

49 51 47

21.90 24.28 32.78

22.54 19.89 24.69

50 57 55

25.85 29.13 37.79

21.74 19.63 23.47

55 61 58

16.98 22.34 36.73

18.64 20.36 41.92

44 47 44

4.41 4.70 5.05

2.54 2.28 1.90

99 108 102

0–11 0–14 0–8.43

4.72 5.35 5.26

2.55 2.18 1.67

49 51 47

4.11 4.12 4.87

2.52 2.23 2.08

50 57 55

4.63 5.00 5.51

2.05 1.72 1.35

55 61 58

4.04 4.31 4.44

3.10 2.82 2.33

44 47 44

4.90 6.32 8.07

3.76 3.77 4.14

99 108 102

0–15 0–15 0–20

5.00 7.04 9.13

3.50 3.56 4.35

49 51 47

4.82 5.68 7.16

4.02 3.87 3.75

50 57 55

6.12 7.21 8.93

3.86 3.64 3.58

55 61 58

3.52 5.17 6.93

3.15 3.66 4.57

44 47 44

Singer & Willett, 2003). Second, it allows for the spacing of waves of data to vary across individuals (Littell et al., 1996; Singer & Willett, 2003). In this data set, narrative measurements were taken at slightly different times. For some children, the time between assessments was 3 months, whereas for others it was closer to 4 months. Third, individual growth modeling can analyze data sets with varying numbers of waves of data (Littell et al., 1996; Singer & Willett, 2003). That is, unlike other approaches, individual growth modeling includes all participants in the estimation regardless of missing data. The majority of children in this study had a narrative score at all three time points, yet some had a narrative score at only two time points. To arrive at a final model that best predicted English narrative development, I built a taxonomy of theoretically motivated individual growth models. Time was denoted in number of months rather than assessment occasions, because assessments were carried out with some variation in exact timing among individuals. As most participants had three data points each, a linear model was used (Singer & Willett, 2003; Willett, Singer, & Martin, 1998). In the first stage, I fit an unconditional means model that included no predictors. This model describes variation in the outcomes (Singer & Willett, 2003). I then fit an unconditional growth model, in which I examined within-person change by

also increased from 5 points to 8 points, roughly equivalent to three quarters of a standard deviation. These three measures (total number of words, mean clause length, and the combined narrative measure) were moderately to highly correlated with each other, with correlations of about .60 at each of the three time points (see Table 3). Children who produced elaborate narratives tended to incorporate a lot of detail into their narrative clauses. The total number of words as well as the number of other features increased in the children’s narratives. Initial English vocabulary scores tended to be moderately correlated with the narrative measures at all three time points, with the median correlation of English vocabulary with the narrative measures being just over .50. Conversely, initial Spanish vocabulary scores were not correlated with English narrative scores on the “Bear Story.”

Individual Growth Modeling: Effect of Arthur? To examine differences in the level and rate of change among individuals, I used individual growth modeling to analyze the combined narrative measure. Individual growth modeling was the appropriate analysis tool for this data set for several reasons. First, it is designed for exploring longitudinal data on individuals over time (Littell, Milliken, Stroup, & Wolfinger, 1996; Singer, 1998;

Table 3 Correlation Matrix for Vocabulary and Narrative Measures (N ⫽ 108) Variable 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Spanish initial vocabulary English initial vocabulary Total no. of words in October Total no. of words in February Total no. of words in May/June Mean clause length in October Mean clause length in February Mean clause length in May/June Combined narrative measure in October Combined narrative measure in February Combined narrative measure in May/June

† p ⬍ .10.

* p ⬍ .05.

** p ⬍ .01.

1

2

3

4

5

6

7

8

9

10

11

—

.07 —

.09 .53** —

⫺.09 .33** .39** —

⫺.26** .24* .13 .31** —

.06 .51** .61** .36** .02 —

⫺.02 .45** .32** .51** .18† .54** —

⫺.25* .44** .31** .35** .54** .30** .50** —

.03 .67** .77** .34** .07 .68** .42** .37** —

⫺.02 .57** .57** .77** .32** .53** .58** .55** .61** —

⫺.16 .54** .38** .48** .63** .30** .45** .60** .47** .69** —

UCHIKOSHI

470

fitting growth trajectories for each child over time. The growth trajectories of each individual child varied. Some children showed steady growth, whereas others showed no growth. In light of the variation across children, it is important to understand the general growth patterns in ELL children’s ability to narrate stories. Hence, I looked at between-person variation and added predictors to investigate whether they affected individual changes in the combined narrative measure. I first added the predictor show to investigate whether individual changes in the combined narrative measure (CNM) were related to viewing of Arthur. Combining the within-person and between-person models yielded the following model: CNM ti ⫽ 关 ␤ 00 ⫹ ␤ 01 SHOW i ⫹ ␤ 10 TIME ti ⫹ ␤ 11 SHOW i TIME ti兴 ⫹ 关u0i ⫹ u1i TIME ti ⫹ rti兴. The parameters in the above model represent the effect of show on the initial level of the CNM (␤01) and the effect of show on the rate of change in the CNM (␤11). Because I was comparing models that differ in their fixed effects but not their variance components, I used full maximum-likelihood estimation (see Singer, 1998). As a general modeling strategy, I first evaluated the above full-model equation for significance. SHOW was kept in the model even if it was not significant, as it was a key predictor. The CLASS variable was kept in the model to control for classroom difference. Indicators of home viewing—that is, watching Arthur at home and watching Between the Lions at home—were also kept in the model to control for home viewing. Past research has used mother’s education to control for socioeconomic status (SES). Although parental education was not significant, it was included in the final model to control for SES. Subsequent analyses investigated whether other variables such as gender, prekindergarten experience, home educational TV viewing, number of older siblings, number of years the child had been in the United States, the child’s initial vocabulary levels in Spanish and English, the number of children’s books in the home, and library exposure4 were significant variables.

Individual Growth Modeling Results After fitting a baseline unconditional means model (Model 1) and baseline unconditional growth model (Model 2) for the combined narrative measure, I built a taxonomy of theoretically motivated individual growth models, as shown in Table 4. The predictors that were not statistically significant were kept out of the model. Control variables (show, Arthur home viewing, Between the Lions home viewing, class, parental education) were kept in the model even if they were not significant. The variance components in Model 1 (unconditional means model) indicate that the average child’s English combined narrative measure varied over time and that the children differed from each other. Using the results of this model, I calculated the intraclass correlation coefficient to be .50. The intraclass correlation coefficient is the intercept divided by the sum of the intercept and the residual (8.53/[8.53 ⫹ 8.38] in this case), and it “describes the proportion of the total outcome variation that lies between people” (Singer & Willett, 2003, p. 96). Thus, this correlation indicated that half of the total variation in the combined narrative measure was attributable to differences among children.

Comparing the variance components in Model 2 (unconditional growth model) with those of Model 1 shows that 48.6% (from 8.38 to 4.31) of the within-person variation in the combined narrative measure was systematically associated with linear time. Furthermore, as there was nonzero variability in both true initial status ( p ⬍ .01) and true rate of change ( p ⬍ .01), Model 2 suggested adding more predictors into the model to explain heterogeneity in each parameter. Preschool experience, number of older siblings, total number of children’s books in the home, and library experience were not significant when included as predictors, and they were not included in subsequent models. Model 9 was chosen as the final model and is interpreted in the following sections. Effect of show. As Model 9 in Table 4 shows, the estimated coefficient for show was not statistically significant, indicating that the two groups did not significantly differ from each other at the start of kindergarten, after I controlled for the other variables in the model. However, the estimated coefficient for the Show ⫻ Time interaction was positive and significant after the other variables in the model were controlled. This finding indicates that viewing Arthur during class hours improved the narrative outcome at a faster pace than did viewing Between the Lions, after I controlled for initial show differences, classroom differences, gender, home viewings, initial English vocabulary, and parental education, as shown in Figure 1. The standard deviation for the combined narrative measure pooled across all occasions was 4.08 points. Thus, the coefficient of .26 for the interaction of show (Arthur) and time corresponds to an effect size of slightly over one twentieth of a standard deviation per month, or an effect of slightly over one half of a standard deviation for the entire school year. Classroom differences. The estimated coefficients for most classrooms tended to be around 0, indicating that most classrooms had similar estimated average initial levels of the combined narrative measure. Moreover, differences in growth in the combined narrative measure among classrooms were also not significant. Home viewing. Home viewing was kept in the model to control for extra viewings of Arthur or Between the Lions. Model 9 indicates that home viewing of neither show was significant. Gender. After I controlled for show, class, home viewing, initial English vocabulary, and parental education, gender had a significant effect on the estimated average initial level of the combined narrative measure. However, the interaction between gender and time was not significant. Boys began with a 1.40-point (effect size of roughly one third of a standard deviation) advantage on the combined narrative measure at the start of kindergarten, and this difference remained throughout the school year, as shown in Figure 1. Initial English vocabulary. After I controlled for the other variables in the model, initial English vocabulary had a significant effect on the estimated average initial level of the combined narrative measure. Every 20-point (one standard deviation) increase in initial English vocabulary was associated with a 2.2-point (effect size of slightly over one half of a standard deviation) increase in the combined narrative measure. That is, children who started kindergarten with higher initial English vocabulary scores also started with higher initial combined narrative measures. How4

Library exposure indicates whether or not the child was taken to the library on a frequent basis.

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS

471

Table 4 Estimates of Fixed and Random Effects From a Series of Fitted Individual Growth Models in Which Show, Class, Arthur Home Viewing, Between the Lions (BTL) Home Viewing, Gender, Initial English Vocabulary Scores, and Library Experience Predict the Average Combined Narrative Measure at the Start of Kindergarten and Rate of Change in the Combined Narrative Measure During the Kindergarten Year for All Children (N ⫽ 108) Model 1: Unconditional means model

Effect

Model 2: Unconditional growth model

Model 3: Show

Model 4: Show ⫻ Time

Model 5: Class

Model 6: Home TV

Model 7: Gender

Model 8: English vocabulary

Model 9: Parental education

4.27** 0.52

4.46** 0.78

2.87** 0.90

1.51† 0.90

⫺0.73 0.77

⫺0.78 1.02

0.36** 0.07

0.36** 0.07

0.36** 0.07

0.36** 0.07

0.21 0.76

0.22 0.74

0.11 0.73

0.16 0.69

⫺0.09 0.58

⫺0.25 0.61

⫺1.08 1.10

⫺0.93 1.04

⫺0.52 0.97

⫺0.41 0.77

⫺0.37 0.81

⫺0.90 1.04

⫺0.30 1.01

⫺0.29 0.94

0.19 0.74

0.42 0.78

1.17 0.97

0.97 0.92

0.81 0.86

⫺0.94 0.72

⫺0.51 0.76

⫺1.41 0.94

⫺1.19 0.88

⫺0.89 0.69

⫺1.15 0.74

Fixed effects Intercept ␤00 SE Time (in months) ␤10 SE Show: Arthur ␤01 SE Class 1 ␤02 SE Class 2 ␤02 SE Class 3 ␤02 SE Class 4 ␤02 SE Class 5 ␤02 SE Home TV: Arthur ␤03 SE Home TV: BTL ␤04 SE Gender: Boys ␤05 SE Initial English vocabulary ␤06 SE Parental education ␤07 SE

6.34** 0.33

4.37** 0.38 0.46** 0.05

⫺0.74 0.87 0.32** 0.08 ⫺0.03 0.63

⫺1.78† 0.98 0.77 0.91

0.35** 0.07

0.33** 0.07

1.38 0.92

1.46† 0.86

0.57 0.68

1.08 0.77

1.62* 0.68

1.69** 0.63

0.76 0.51

0.46 0.55

1.05 0.63

0.96 0.59

0.47 0.47

0.45 0.50

2.19** 0.53

1.31** 0.44

1.40** 0.47

0.10** 0.01

0.09** 0.01 0.12 0.14

Show ⫻ Time ␤11 SE

0.22* 0.11

0.22* 0.11

0.22* 0.11

0.22* 0.11

0.23* 0.11

0.26* 0.11

Random effects (variance components) Intercept Estimate SE Slope Estimate SE Residual Estimate SE Proportional reduction in variance from Model 2: Intercept Slope Akaike’s information criterion † p ⬍ .10.

* p ⬍ .05.

8.53** 1.61

8.38** 0.84

1,691.8

** p ⬍ .01.

10.74** 2.22

10.96** 1.50

10.79** 2.22

10.03** 2.14

9.30** 2.05

7.77** 1.85

4.07** 1.39

3.69** 1.45

0.11** 0.05

0.11* 0.05

0.11** 0.05

0.11** 0.05

0.11** 0.05

0.11** 0.05

0.11* 0.05

0.09* 0.05

4.31** 0.60

4.42** 0.66

4.28** 0.60

4.29** 0.60

4.31** 0.60

4.29** 0.60

4.28** 0.60

4.28** 0.64

1,611.1

N/A N/A 1,338.8

N/A N/A 1,607.9

6.6% N/A 1,606.1

13.4% N/A 1,598.5

27.7% N/A 1,584.7

62.10% N/A 1,535.8

65.6% 18.2% 1,346.5

UCHIKOSHI

472

Figure 1. Average fitted growth trajectories, after controlling for parental education, that describe the effect of show and gender on the change in the combined narrative measure for English-language learner children in an average class with average initial vocabulary scores of 40 points who watch both Arthur and Between the Lions (BTL) at home (N ⫽ 108).

ever, initial English vocabulary was not associated with rate of growth on the narrative measures. Parental education. Parental education was included in the model to control for SES. Model 9 indicates that parental education was not significant.

Five Narrative Measures The fitted growth trajectories for each individual narrative measure (story structure, events, evaluations, temporality and reference, and storybook language) were further examined using the growth modeling perspective. The estimated coefficients for the interaction between show and time were positive and significant for story structure (␤11 ⫽ .09; p ⫽ .04) and evaluation (␤11 ⫽ .09; p ⫽ .03), after class, home viewing, gender, initial English receptive vocabulary, and parental education were controlled. This finding indicates that viewing Arthur during class hours improved the narrative outcome for story structure and evaluation at a faster pace than did viewing Between the Lions. The interaction between show and time was not significant either for events or for temporality and reference. For the storybook language measure, many of the children scored 0 in October and stayed at 0. The low occurrence of storybook language suggests that it might be difficult for kindergarten-age ELL children to provide such language. In addition, gender was significant for story structure, events, and temporality and reference. That is, boys started kindergarten with higher scores in these three areas than did girls, and these differences remained constant throughout kindergarten. Figure 2 shows the average growth trajectories describing the effect of show on the change in story structure, evaluation, events, and temporality and reference for ELL boys who watched Arthur during the intervention period. On average, the most gains were seen in the story structure measure.

change. The average growth trajectories (see Figure 1) show a gradual increase in narrative skills. In this section, the stories provided by two boys are presented in order to demonstrate the average growth pattern. These two boys, Miguel and Raymond, were in the Arthur group. Miguel started the school year with minimal English abilities. Raymond was relatively fluent and comfortable in speaking English. At Time 1, in October, for many children, their lack of English vocabulary kept them from creating long, detailed narratives. Many children with limited English, such as Miguel, tended to stay silent or to produce narratives in Spanish. These Spanish stories were not used in the evaluation, because the objective of this study was to measure English narrative growth. Children with greater English abilities, such as Raymond, were able to connect several sentences together to make a narrative, and their narratives included more events to move the plot forward. Under Peterson and McCabe’s (1983) categorization, many of these narratives would be classified as chronological narratives, in which the child connects sentences with and and then, or leap-frog narratives, in which the child jumps from one event to the next. Excerpt 1: Miguel at Time 1 [Said nothing] Excerpt 2: Raymond at Time 1 Last morning the bear was getting kite out of tree. And it was stuck. Then last morning the kite fell down. Last morning he went to take out the kite. Then it fell. Then it broke. Now it pull pull harder.

At Time 2, in February, children began to include more evaluative words in their stories. For example, Miguel said, “One [bear] was tired.” He described the bears as being little. In addition, the words bear and kite were mentioned more in the stories.

Qualitative Results

Excerpt 3: Miguel at Time 2 The three little bear was walking. One was tired. And one was in the tree. That is it.

Growth trajectories for each individual child varied. Some children displayed steady development, whereas others exhibited little

Excerpt 4: Raymond at Time 2 One day a bear was running a kite. And it was a sunny day. Then the

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS

473

Figure 2. Average fitted growth trajectories, after controlling for parental education, that describe the effect of show on the change in story structure, evaluation, events, and temporality and reference for English-language learner boys who watched Arthur during the intervention period, who were in an average class with average initial vocabulary scores of 40 points, and who watched both Arthur and Between the Lions at home (N ⫽ 108).

kite was stuck in the tree. Then the bear was climbing to the tree. Then he was trying to get it. Then the kite ripped.

At Time 3, in May or June, many of the advanced ELL children provided more events and advanced story structures. Some advanced ELL children used direct speech, painting a more vivid verbal picture of their narratives. Some children started their narratives with titles, such as Miguel and his title “The Book.” He then continued by beginning his story with a traditional beginning such as “Once upon a time” and ended his story with a formal ending such as “The End.” According to Peterson and McCabe’s (1983) categorization, some of the stories would be classified as ending-at-the-high-point narratives, in which the child builds up to the high point and then abruptly ends the story. For example, Miguel ended his story at a climactic moment with “he fall off.” Some children even produced stories of the classic pattern. It is interesting to note that in May, Raymond gave the same story as he did in February yet he added a resolution and coda. Excerpt 5: Miguel at Time 3 The Book. Once upon a time, they was flying kites. The kite then hit the tree. Then a bear try to get it. Now he fall off. And the end. Excerpt 6: Raymond at Time 3 One day a bear was running a kite. And it was a sunny day. Then the kite was stuck in the tree. Then the bear was climbing to the tree. Then he was trying to get it. Then the kite ripped. So they got the kite out of the tree. And it was so much fun. That was the end.

Discussion The results from this year-long study show intervention effects; children who viewed Arthur during class hours had steeper trajectories on the combined narrative measure than did those who

viewed Between the Lions during class hours. The study was designed so that half of the children in each classroom watched one of the two shows. Because of time constraints, it was not possible for the teacher or the researchers to follow up with exercises to reinforce learning from the educational TV shows. Yet even though no reinforcement followed the viewing sessions, growth in narrative skill was greater for children who watched Arthur, after any initial show differences, classroom differences, gender, home viewings, initial English vocabulary, and parental education were controlled for, as shown in Figure 1. Each half-hour Arthur episode presented two stories, each with a plot, a conflict, and a resolution. Hence, children who viewed Arthur had a lot of exposure to mainstream storytelling techniques. This effect was also demonstrated in the separate growth modeling analysis for story structure, in which the interaction between show (Arthur) and time was significant. The interaction between Arthur and time was also significant in the separate growth modeling analysis for evaluation. Coviewing in a classroom with peers may have assisted in comprehension and understanding of the meaning of the events. Researcher observations and teacher reports indicated that there appeared to be more conversations concerning plot, storyline, and characters after the viewing sessions among the children who viewed Arthur than among those who viewed Between the Lions. Arthur may have had more memorable characters and impressionable storylines than Between the Lions, which focused more on phonics and rhythm. Children may have understood the emotions of the Arthur characters through these casual discussions with classmates. Conversations among the Arthur viewers may have reinforced some vocabulary items, expanding the children’s lexicons. Although each episode of Between the Lions begins with a story that the Lion family reads together, the program spends consider-

474

UCHIKOSHI

able time directing viewers’ attention to phonological sensitivity, the alphabetic principle, letter–sound correspondence, word meanings, punctuation, and other conventions of written English (Rath, 2000). Thus children’s attention may be diverted from the narrative structure of the extended discourse. After the viewings, the researchers and teachers noted that the children sang the songs and chanted the rhymes from the show, but they did not discuss the contents of the show, including plot and character development, like the Arthur groups did.5 This study also suggests that children can develop mainstream narrative styles prior to first grade. Past research shows that narrative styles are influenced by early socialization experiences and may differ depending on cultural background (Dart, 1992; Heath, 1982, 1986; Minami & McCabe, 1991; Ninio, 1980; Shiro, 1995; Silva & McCabe, 1996; Tannen, 1980; Wang & Leichtman, 2000) as well as SES (Burger & Miller, 1999; Heath, 1982; Shiro, 2003; Whitehurst et al., 1994). Children from nonmainstream cultural backgrounds or with lower SES enter school with different narrative styles than those valued in schools (Dickinson & McCabe, 1991; Heath, 1982; Minami & McCabe, 1991). The combined narrative measure examined features typical of school norms (such as having a title, an introduction, and a closing as part of the story structure) and of the stories shown in Arthur. This study shows that routine and attentive viewing of Arthur can assist nonmainstream bilingual children to develop English narrative styles that match the English-speaking school norms faster, even prior to formal literacy instruction. It is interesting that there was a gender effect; boys started with and maintained higher English narrative scores than girls. This gender gap was also seen in the children’s initial English vocabulary scores. Yet for Spanish vocabulary, there was no gender difference. Past research with monolingual English children has shown that, on average, girls have higher literacy skills than boys (Bornstein, Haynes, & Painter, 1998; Gambell & Hunter, 1999; Karmiloff & Karmiloff-Smith, 2001). Although only a few studies have looked at gender differences in second-language acquisition, mainly in the areas of reading, these studies have provided contradictory results. Medrano (1986) found no significant gender differences in English reading among sixth-grade Mexican American students whose first language was Spanish. Yet L. Reese, Garnier, Gallimore, and Goldenberg (2000) showed that second-generation seventh-grade Latina students outscored their male counterparts in English reading achievement tests. Although further research is necessary to verify the patterns identified here, these results are suggestive of several hypotheses. Boys and girls may have different exposure histories; for example, girls may be more likely to be taken care of by Spanish-speaking relatives and may have fewer interactions with their English environments than boys. Gender effects may also depend on the age of acquisition, the age of testing, as well as on the skill being assessed. In addition, differences in cultural values may play a role. Parents may emphasize academic language attainment more to boys than to girls. Future research should involve a more in-depth study of children’s home environments. Furthermore, and similar to past findings by Tabors, Roach, and Snow (2001) that kindergarten assessments of oral narrative production and receptive vocabulary were moderately correlated, a relationship between English narrative skills and English vocabu-

lary was found in this study. English vocabulary was predictive of initial levels of English narrative skills. An investigation of the children’s word choice from the narratives may be useful for further understanding the relationship between second-language English vocabulary and narrative abilities. I have demonstrated the positive impact of Arthur on children’s narrative skills. It would also be of value to explore further the degree to which characteristics of the language used in Arthur, such as syntactic complexity and lexical repetition, contributed to this effect. Research has shown that TV programs designed for children use characteristics of child-directed speech (HoffGinsberg, 1986; Snow, 1984; Wells, 1985), but a more targeted analysis of what ELL children actually understand after one viewing of the various episodes would be informative. Further, this study focused on only one bilingual population; it would be of value to replicate the study with other groups of young bilinguals, particularly as limited research shows that ELL children are likely to begin schooling in English less prepared for literacy instruction than children from middle-class English-speaking homes (Snow, Burns, & Griffin, 1998). In addition, these children were followed only for their kindergarten year; although considerable other data suggest that narrative skills in kindergarten relate to later narrative and literacy skills, a longer term longitudinal study collecting data on those later outcomes would be of great value.

Conclusion This study showed how ELL children progressed in their narrative skills as a function of exposure to two educational TV shows that are easily available on PBS to all children. For the population of children in this study, viewing Arthur assisted in the development of their extended discourse. Further studies should examine the influence of other children’s TV shows on ELL children’s literacy outcomes. A more detailed understanding of ELL children’s literacy processes as well as of the effects of educational TV is important to the design of better language education practices, assessments, and interventions.

5 Intervention effects on phonological awareness skills were seen for the Between the Lions viewers. This finding will be discussed in a separate article.

References Adams, M. J. (1990). Learning to read: Thinking and learning about print. Cambridge, MA: MIT Press. August, D., & Hakuta, K. (Eds.). (1997). Improving schooling for language-minority children: A research agenda. Washington, DC: National Academy Press. Bakeman, R., & Gottman, J. (1997). Observing interaction (2nd ed.). Cambridge, England: Cambridge University Press. Bamberg, M., & Damrad-Frye, R. (1991). On the ability to provide evaluative comments—further explanations of children’s narrative competencies. Journal of Child Language, 18, 689 –710. Beck, S., Coker, D., Hemphill, L., & Bellinger, D. (2003). Literacy skills of children with a history of early corrective heart surgery. In C. M. Fairbanks, J. Worthy, B. Maloch, J. Hoffman, & D. Schallert (Eds.),

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS 52nd yearbook of the National Reading Conference. Oak Creek, WI: National Reading Conference. Bornstein, M. H., Haynes, O. M., & Painter, K. M. (1998). Sources of child vocabulary competence: A multivariate model. Journal of Child Language, 25, 367–393. Burger, L. K., & Miller, P. J. (1999). Early talk about the past revisited: Affect in working-class and middle-class children’s co-narrations. Journal of Child Language, 26, 133–162. Chang, C. (2004). Telling stories of experiences: Narrative development of young Chinese children. Applied Psycholinguistics, 25, 83–104. Dart, S. (1992). Narrative style in the two languages of a bilingual child. Journal of Child Language, 19, 367–387. Dickinson, D. K., & McCabe, A. (1991). The acquisition and development of language: A social interactionist account of language and literacy development. In J. F. Kavanagh (Ed.), The language continuum from infancy to literacy (pp. 1– 40). Parkton, MD: York Press. Dickinson, D. K., & Tabors, P. (Eds.). (2001). Beginning literacy with language. Baltimore: Brookes. Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test (3rd ed.). Circle Pines, MN: American Guidance Service. Dunn, L. M., Padilla, E. R., Lugo, D. E., & Dunn, L. M. (1986). Test de Vocabulario en Ima´genes Peabody. Circle Pines, MN: American Guidance Service. Eaton, J. H., Collis, G. M., & Lewis, V. A. (1999). Evaluative explanations in children’s narratives of a video sequence without dialogue. Journal of Child Language, 26, 699 –720. Galdwell, M. (2000). The tipping point: How little things can make a big difference. Boston: Little, Brown and Company. Gambell, T. J., & Hunter, D. M. (1999). Rethinking gender differences in literacy. Canadian Journal of Education, 24, 1–16. Griffin, T. M., Hemphill, L., Camp, L., & Wolf, D. P. (2004). Oral discourse in the preschool years and later literacy skills. First Language, 24, 123–147. Gutie´rrez-Clellen, V. F. (2002). Narratives in two languages: Assessing performance of bilingual children. Linguistics and Education, 13, 175– 197. Heath, S. B. (1982). What no bedtime story means: Narrative skills at home and school. Language in Society, 2, 49 –76. Heath, S. B. (1986). Taking a cross-cultural look at narratives. Topics of Language Disorders, 7, 84 –94. Hill-Scott, K. (2001). Industry standards and practices. In D. G. Singer & J. L. Singer (Eds.), Handbook of children and the media (pp. 605– 620). Thousand Oaks, CA: Sage. Hoff-Ginsberg, E. (1986). Function and structure in maternal speech: Their relation to the child’s development of syntax. Developmental Psychology, 22, 155–163. Huston, A. C., Wright, J. C., Rice, M. L., Kerkman, D., & St. Peters, M. (1990). Early television viewing. Developmental Psychology, 26, 409 – 420. Karmiloff, K., & Karmiloff-Smith, A. (2001). Pathways to language: From fetus to adolescent. Cambridge, MA: Harvard University Press. Karmiloff-Smith, A. (1985). Language and cognitive processes from a developmental perspective. Language and Cognitive Processes, 1, 61– 85. Labov, W. (1972). Language in the inner city: Studies in the Black English vernacular. Philadelphia: University of Pennsylvania Press. Lemish, D., & Rice, M. (1986). Television as a talking picture book: A prop for language acquisition. Journal of Child Language, 13, 251–274. Linebarger, D. L. (2000). Summative evaluation of “Between the Lions”: A final report to WGBH Educational Foundation. Boston: WGBH. Littell, R. C., Milliken, G. A., Stroup, W. W., & Wolfinger, R. D. (1996). SAS system for mixed models. Cary, NC: SAS Institute. Mason, J., Stewart, J., Peterman, C., & Dunning, D. (1992). Toward an

475

integrated model of early reading development (Tech. Rep. No. 566). Champaign, IL: Center for the Study of Reading. McCabe, A., & Peterson, C. (1991). Getting the story: A longitudinal study of parental styles in eliciting oral personal narratives and developing narrative skill. In A. McCabe & C. Peterson (Eds.), Developing narrative structure (pp. 217–253). Hillsdale, NJ: Erlbaum. Medrano, M. F. (1986). Evaluating the long-term effects of a bilingual education program: A study of Mexican American students. Journal of Educational Equity and Leadership, 6, 129 –138. Minami, M., & McCabe, A. (1991). Haiku as a discourse regulation device: A stanza analysis of Japanese children’s personal narratives. Language in Society, 20, 577–599. Ninio, A. (1980). Picture-book reading in mother–infant dyads belonging to two subgroups in Israel. Child Development, 51, 587–590. Nord, C. W., Lennon, J., Liu, B., & Chandler, K. (1999). Home literacy activities and signs of children’s emerging literacy, 1993 and 1999 (NCES 2000 – 026rev). Washington DC: U. S. Department of Education, Office of Educational Research and Improvement, National Center for Education Statistics. Retrieved November 2, 2003, from http://nces.ed .gov/pubs2000/2000026.pdf Office of English Language Acquisition, Language Enhancement, and Academic Achievement for Limited English Proficient Students. (n.d.). United States rate of LEP student growth. Retrieved May 1, 2002, from http://www.ncbe.gwu.edu/ncbepubs/reports/state-data/2000/usar.pdf Paul, R., & Smith, R. L. (1993). Narrative skills in 4-year-olds with normal, impaired, and late developing language. Journal of Speech & Hearing Research, 36, 592–598. Peterson, C., Jesso, B., & McCabe, A. (1999). Encouraging narratives in preschoolers: An intervention study. Journal of Child Language, 26, 49 – 67. Peterson, C., & McCabe, A. (1983). Developmental psycholinguistics: Three ways of looking at a child’s narrative. New York: Plenum Press. Peterson, C., & McCabe, A. (1991). Linking children’s connective use and narrative macrostructure. In A. McCabe & C. Peterson (Eds.), Developing narrative structure (pp. 29 –53). Hillsdale, NJ: Erlbaum. Peterson, C., & McCabe, A. (1992). Parental styles of narrative elicitation: Effect on children’s narrative structure and content. First Language, 12, 299 –321. Purcell-Gates, V. (1988). Lexical and syntactic knowledge of written narrative held by well-read-to kindergarteners and second graders. Research in the Teaching of English, 22, 128 –160. Purcell-Gates, V. (1991). Ability of well-read-to kindergarteners to decontexualize/recontextualize experience into written-narrative register. Language and Education, 5, 177–188. Purcell-Gates, V. (1992). Roots of response. Journal of Narrative and Life History, 2, 151–161. Purcell-Gates, V. (1996). Stories, coupons, and the TV guide: Relationships between home literacy experiences and emergent literacy knowledge. Reading Research Quarterly, 31, 406 – 428. Purcell-Gates, V. (2001). Emergent literacy is emerging knowledge of written, not oral, language. New Directions for Child and Adolescent Development, 92, 7–22. Purcell-Gates, V., & Dahl, K. L. (1991). Low-SES children’s success and failures at early literacy learning in skills-based classrooms. Journal of Reading Behavior, 23, 1–34. Rath, L. (2000, March). Between the Lions: Using television to promote literacy. Book Links, pp. 41– 45. Reese, E., Haden, C. A., & Fivush, R. (1993). Mother– child conversations about the past: Relationships of style and memory over time. Cognitive Development, 8, 403– 430. Reese, L., Garnier, H., Gallimore, R., & Goldenberg, C. (2000). Longitudinal analysis of the antecedents of emergent Spanish literacy and middle-school English reading achievement of Spanish-speaking students. American Educational Research Journal, 37, 633– 662.

476

UCHIKOSHI

Rice, M. L., Huston, A. C., Truglio, R., & Wright., J. C. (1990). Words from “Sesame Street”: Learning vocabulary while viewing. Developmental Psychology, 26, 421– 428. Rice, M. L., & Woodsmall, L. (1988). Lessons from television: Children’s word learning when viewing. Child Development, 59, 420 – 429. Scarborough, H. S., & Dobrich, W. (1994). On the efficacy of reading to preschoolers. Developmental Review, 14, 245–302. Scarborough, H. S., Dobrich, W., & Hager. M. (2001). Preschool literacy experience and later reading achievement. Journal of Learning Disabilities, 24, 508 –511. Shiro, M. (1995). Venezuelan preschoolers’ oral narrative abilities: Focus on research. Language Arts, 72, 528 –537. Shiro, M. (2003). Genre and evaluation in narrative development. Journal of Child Language, 30, 165–195. Silva, M. J., & McCabe, A. (1996). Vignettes of the continuous and family ties: Some Latino American traditions. In A. McCabe (Ed.), Chameleon readers: Teaching children to appreciate all kinds of good stories. New York: McGraw-Hill. Singer, J. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics, 24, 323–355. Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press. Snow, C. E. (1984). Parent– child interactions and the development of communicative ability. In R. L. Schiefelbusch & J. Pickar (Eds.), Communicative competence: Acquisition and intervention (pp. 69 –108). Baltimore: University Park Press. Snow, C. E., Burns, M. S., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in young children. Washington, DC: National Academy Press. Snow, C. E., Tabors, P., Nicholson, P., & Kurland, B. (1995). SHELL: Oral language and early literacy skills in kindergarten and first-grade children. Journal of Research in Childhood Education, 10, 37– 48. Sulzby, E. (1985). Children’s emergent abilities to read favorite story-

books: A developmental study. Reading Research Quarterly, 20, 458 – 481. Tabors, P., Beals, D., & Weizman, Z. (2001). “You know what oxygen is?”: Learning new words at home. In D. Dickinson & P. Tabors (Eds.), Beginning literacy with language (pp. 93–110). Baltimore: Brookes. Tabors, P., Roach, K., & Snow, C. E. (2001). Home language and literacy environment: Final results. In D. Dickinson & P. Tabors (Eds.), Beginning literacy with language (pp. 111–138). Baltimore: Brookes. Tannen, D. (1980). A comparative analysis of oral narrative strategies: Athenian Greek and American English. In W. Chafe (Ed.), The Pear stories: Cognitive, cultural and linguistic aspects of narrative production (pp. 51– 87). Norwood, NJ: Ablex. Van Evra, J. (1998). Television and child development. Mahwah, NJ: Erlbaum. Wang, Q., & Leichtman, M. D. (2000). Same beginnings, different stories: A comparison of American and Chinese children’s narratives. Child Development, 71, 1329 –1346. Wells, G. (1985). Language development in the pre-school years: Language at home and at school (Vol. 2). Cambridge, England: Cambridge University Press. Whitehurst, G. J., Arnold, S. A., Epstein, J. N., Angell, A. L., Smith, M., & Fischel, J. E. (1994). A picture book reading intervention in day care and home for children from low-income families. Developmental Psychology, 30, 679 – 689. Willenberg, I., & Kang, J. (2001). Bear Story coding manual. Unpublished manuscript, Harvard University. Willett, J. (1994). Measuring change more effectively by modeling individual growth over time. In T. Husen & T. N. Postlethwaite (Eds.), The international encyclopedia of education (2nd ed., pp. 671– 678). Oxford, England: Pergamon Press. Willett, J., Singer, J. D., & Martin, N. C. (1998). The design and analysis of longitudinal studies of development and psychopathology in context: Statistical models and methodological recommendations. Development and Psychopathology, 10, 395– 426.

NARRATIVE DEVELOPMENT IN BILINGUAL KINDERGARTENERS

477

Appendix Coding of the Five Dimensions in the Combined Narrative Measure Each Bear Story was coded along five dimensions: story structure features, events, evaluation, temporality and reference, and storybook language. The following coding scheme was adapted from the

“Bear Story Coding Manual” (an unpublished manuscript by I. Willenberg and J. Kang, 2001, Harvard University) with the permission of its authors:

(Appendix continues)

Received February 10, 2004 Revision received November 17, 2004 Accepted January 14, 2005 䡲