L1 AND L2 NARRATIVE DEVELOPMENT IN EMERGENT BILINGUALS … · The present study addressed emergent...

56
L1 AND L2 NARRATIVE DEVELOPMENT IN EMERGENT BILINGUALS by Sun Hwa Baek A thesis submitted in conformity with the requirements for the degree of Master of Arts Graduate Department of Applied Psychology and Human Development Ontario Institute for Studies in Education University of Toronto © Copyright by Sun Hwa Baek (2015)

Transcript of L1 AND L2 NARRATIVE DEVELOPMENT IN EMERGENT BILINGUALS … · The present study addressed emergent...

L1 AND L2 NARRATIVE DEVELOPMENT IN EMERGENT BILINGUALS

by

Sun Hwa Baek

A thesis submitted in conformity with the requirementsfor the degree of Master of Arts

Graduate Department of Applied Psychology and Human DevelopmentOntario Institute for Studies in Education

University of Toronto

© Copyright by Sun Hwa Baek (2015)

ii

L1 AND L2 NARRATIVE DEVELOPMENT IN EMERGENT BILINGUALS

Master of Arts (2015)

Sun Hwa Baek

Department of Applied Psychology and Human Development

University of Toronto

Abstract

This study investigated the development of oral narrative production in L1 and L2 in

English-Hebrew emergent bilingual children and the associations between narrative abilities in

both languages. Participants included 33 younger cohort in JK and 24 older cohort in SK. The

children’s scripts (general descriptions about events) were elicited at two time points over one

year. The study included various measures of macro and micro dimensions of narratives that

reveal cognitive and linguistic development. Children were able to produce generalized scripts

in both languages. L1 narratives have been developing on various complexity indices of macro

and micro dimensions, while emergent L2 has been developing on narrative productivity. A

cross-linguistic association was noted at the end of grade1 with increased Hebrew language

proficiency. The findings suggest that cross-linguistic relations may be independent of

linguistic distance. Educational and clinical implications were discussed.

iii

Acknowledgements

I would like to express my sincere gratitude to my wonderful supervisor, Dr. Esther

Geva, who provided careful guidance and endless encouragement throughout the thesis

progress. Although it took a long time for me to finish the thesis, she was patient and motivated

me continuously. Without her help, this thesis would not have been possible. I would also like

to thank my second reader, Dr. Becky Xi Chen for giving me invaluable comments which

helped to clarify my findings.

I owe my deepest gratitude to Dr. Monique Herbert, and my colleagues, Emiko Koyama

and Norman Himel, who shared their expertise and gave me suggestions when I was struggling

with statistics. In addition, I am grateful to Vaunam Venkadasalam for reading through my

thesis and editing when appropriate.

My sincere thanks go to my husband, Hoon, as well as my lovely daughter Sydney, who

endured this whole process and helped me focus on my work. I also would like to thank my

parents, who pray for me diligently.

Last but not the least, I thank the children who participated in the study for giving me an

opportunity to learn and write a thesis about narrative development in children.

iv

Table of Contents

Abstract .................................................................................................................................... ii

Acknowledgements .................................................................................................................. iii

Table of Contents..................................................................................................................... iv

List of Tables ............................................................................................................................ v

Introduction .............................................................................................................................. 1

Methodology............................................................................................................................13

Participants...........................................................................................................................13

Measures ..............................................................................................................................14

Coding System .....................................................................................................................17

Results .....................................................................................................................................22

The pattern of change in L1 ..................................................................................................23

The pattern of change in L2 Narratives .................................................................................24

Qualitative investigations into the development of scripts.....................................................25

Cross-linguistic relations between narratives in L1 and L2 at Time 1 and Time 2 .................29

Discussion................................................................................................................................29

References ...............................................................................................................................38

v

List of Tables

Table 1. Descriptive statistics for participants by Age and Gender............................................43

Table 2. Variables name and an example for each measure.......................................................44

Table 3. Distributions and the McNemar test results for categorical variables in the L1 (English)

for both cohorts.................................................................................................................45

Table 4. Descriptive statistics and the Wilcoxon or the Sign test Results for continuous

variables in English for the younger cohort .......................................................................46

Table 5. Descriptive statistics and the Wilcoxon or the Sign test Results for continuous

variables in English for the older cohort ............................................................................47

Table 6. Descriptive statistics and the Wilcoxon or the sign test Results for continuous variables

in Hebrew for the younger cohort......................................................................................48

Table 7. Descriptive statistics and the Wilcoxon or the sign test Results for continuous variables

in Hebrew for the older cohort...........................................................................................49

Table 8. Intercorrelations among narrative measures in the L1 and L2 at Time 2 for the younger

cohort................................................................................................................................50

Table 9. Intercorrelations among narrative measures in the L1 and L2 at Time 2 for the older

cohort................................................................................................................................51

1

INTRODUCTION

Children’s narrative abilities not only provide rich information about cognitive and

linguistic development (Applebee, 1978; Berman, 1995, 1996; Berman & Slobin, 1994;

Bamberg, 1987), but they correlate with literacy skills, later academic success (e.g., Catts,

Hogan, & Fey, 2003; Feagans & Appelbaum, 1986; Geva & Olson, 1983; Griffin, Hemphill,

Camp, & Wolf, 2004; McCardle, Scarborough, & Catts, 2001; Scarborough, 2001; Snow,

Burns, & Griffin, 1998; Tabors, Snow, & Dickinson, 2001) and social skills (e.g., Berman &

Slobin, 1994; Richner & Nicolopolou, 2001; Snow & Dickinson, 1990).

Due to the significance of narrative abilities in children, there has been a great deal of

research into narrative development in monolingual children and there is now a growing

literature on narratives in bilingual children (e.g., Fiestas & Peña, 2004; Gutiérrez-Clellen, 2004;

Kupersmitt & Berman, 2001; Lanza, 2001; Montanari, 2004; Pearson, 2001, 2002; Pearson &

de Villiers, 2005; Schwartz & Shaul, 2013; Viberg 2001; Ucelli & Páez, 2007). Since narrative

is a universal practice that can be observed in different cultures and languages, it has been

proposed that it can be used as a parallel assessment tool across languages given the

disadvantages to standardized tests for bilingual young children (Simon-Cereijido & Gutiérrez-

Clellen, 2009).

Like the tip of an iceberg, underneath the narratives told by individuals, many kinds of

knowledge must be supported. A narrator is required to draw not only on higher-order thinking

skills such as planning discourse, organizing and monitoring it for coherence at a macro level,

but also linguistic knowledge such as vocabulary, morphology, and semantic and syntactic

skills at a micro level. Therefore, narratives are often analyzed in terms of the two levels: macro

and micro levels. Researchers address the importance of considering these two complementary

2

dimensions of narratives together in order to capture full understanding of narratives (Berman,

2001; Hickman, 2004).

The macro dimension of narratives refers to overall structure of narratives mapped on

narrative schema. The overall structure concerns the coherence and ‘orderly flow of

information’ (Hudson and Shapiro, 1991, p. 93). The macrostructure of a story can be captured

in story grammar, which consists of a setting, initiating events, a problem, a resolution, and an

ending (Geva & Olson, 1983; Mandler, 1983; Stein, 1988; Stein & Glenn, 1982). The micro

dimension of narratives refers to linguistic structure at the word, sentence and paragraph levels.

There is no ‘gold standard’ to characterize the microstructure of narratives (Justice et al., 2006),

but it is often analyzed in terms of language productivity (narrative length) and language

complexity (syntactic complexity) (Justice et al., 2006).

The present study addressed emergent bilinguals’ narrative development in their two

languages at the two dimensions by focusing on the script, which is one type of narrative genre.

A script is a general description about what typically occurs in 1an event” and ‘organized as

event schema’ (Hudson & Shapiro, 1991, p. 91).

A compelling reason for studying this genre is that event schema, which children’s

scripts are mapped on, has considerable importance for cognitive and linguistic development of

young children (Nelson, 1986). Event schema is constructed from experiences we have in our

daily lives (Schank & Abelson, 1977). For very young children before they learn to speak,

everyday experiences are almost an exclusive source to gain world knowledge (Nelson, 1986).

Thus, event schema in children is the earliest form of representation of the world around them

and it provides a crucial context for cognitive and linguistic development (Nelson, 1986;

French, 1986). Children’s cognitive systems develop through cognitive analysis such as pattern

1The term “event” used here is concerned with an event that involves people and acting on objects and interacting

with each other to achieve a certain goal. (Nelson, 1986, p.11)

3

analysis and categorization on event representation of repeated experiences over time (Nelson,

1986). For example, children learn to understand and appreciate the concepts of temporal,

clausal, conditional relationships that happen repeatedly within familiar events. Such everyday

routines provide children with a context to express those concepts in language (Farrar et al.,

1993; French, 1986).

These event schemas provide a basis for recalling and understanding personal narratives

and story constructions (Hudson & Nelson, 1983; Nelson, 1986, p. 40). It has been argued that

event schema functions as a framework to interpret and anticipate behaviour in a familiar

situation and they assist with recalling an episodic memory (e.g., Schank & Abelson, 1977) and

help comprehend other people’s narratives or organize their narratives (Slackman, Hudson &

Fivush, 1986, p. 49). That is, they influence the understanding of stories and the ability to

produce discourse (Nelson, 1986).

Because of the importance of event schema in young children, most previous studies

have investigated mainly event schema exhibited in children’s spoken scripts. In addition, little

is known about the development of bilingual children’s scripts relative to the development of a

story in bilingual children. Motivated by these limitations in past research, the present study

aimed to capture a full picture of the development of scripts in bilingual children by

investigating macro and micro dimensions of scripts. The second objective of this study was to

explore whether and how narrative abilities within the bilingual’s two languages are related.

Literature review

The script model theory

Schank and Abelson (1977) defined script as spatially and temporally organized event

representation. Children form scripts from every day real life experiences such as going grocery

shopping, getting dressed, or having dinner at a restaurant. The representation is a generalized

4

structure made of slots for variables such as actors, actions, and props that likely occur in a

particular event. For example, in the restaurant event, the actors are waiters and customers,

actions include entering the restaurant, ordering, eating and paying, and props include menus

and food. These variables can be filled with a range of probable alternatives, referred to as slot

fillers.

Scripts have two major features – it has “inferential power” and a causal structure

(Nelson & Gruendel, 1981). When we encounter a familiar event or situation, an event schema

is activated and one can predict all requirements and optional components even when only

partial information is explicitly provided. For example, when a Western person is talking about

scrambling eggs and cooking bacon, we can infer that they are talking about breakfast and we

can predict other components of the narrative within the breakfast context. This is the feature

that gives the scripts “inferential power”. The other characteristic of a script is its temporal and

causal structure, which distinguishes it from other schemta. Since events happen over time, acts

within an event are linked temporally or causally. For example, in the restaurant script, the food

ordering action precedes the eating or paying actions.

The development of scripts in monolingual children

Research on children’s scripts has shown that even young children have well-organized

event schema about every day routines. For example, Nelson (1978) asked 3-and 4-year-olds in

a daycare center about familiar eating events in three different settings; at the daycare center, at

home and McDonalds. The children at both ages were able to provide generalized descriptions

of the events and used the general pronoun “you” as actor and timeless present tense to describe

actions. Another important finding from this research was that the children sequenced acts2 in

the order that matches the event sequence in reality (e.g., “we eat and after the main course we

have dessert, usually ice cream”). The finding that children are capable of giving a well-

2The term of an‘act’ was defined as any action or change in state (Nelson, 1986, p. 26).

5

organized description about everyday routines has been documented in subsequent studies

(Gruendel, 1981; Fivush, 1984; Fivush & Slackman, 1986; Myler-Worsley et al., 1986; Nelson,

Fivush, Hudson & Lucariello, 1983).

Furthermore, research has shown that young children are able to apply a general

structure when recounting novel experiences such as an airplane trip, having a fire drill, or their

first day at school (Fivush, 1984; Hudson &Nelson 1984; Nelson & Gruendel, 1986). For

example, Fivush (1984) asked children at 5 years of age about their general school day routines

on four different occasions during the first three months of school. Even on the second day of

school, they already differentiated between the routine of nursery school and that of

kindergarten. For instance, children mentioned a new routine in kindergarten such as “putting

your stuff in your locker” (which did not occur in nursery school). Likewise, in a cross-

sectional study, Hudson and Nelson (1984) found that children ranging from 3 to 7 years in age

were able to provide general accounts of the first experience they had such as “going to

Disneyland, airplane trip, and a special class at school. These findings suggest that scripts are

formed as a generalized structure rather than reflecting episodic memories of specific events

from the outset.

Although young children have relatively well-developed scripts, children’s scripts

become more generalized, elaborate and complex with increased age and experience (Farrar &

Goodman, 1990; Fivush & Slackman, 1986; Gruendel, 1980; McCartney & Nelson, 1981; Price

& Goodman, 1990; Slackman & Nelson, 1984). The actions or props reported in the narratives

become more inclusive and schematized as the representations of an event become

hierarchically organized (Fivush & Slackman, 1986). For example, at preschool a child may say,

“you get a strawberry candy and potato chip” when describing a birthday party, but might use

more abstract language such as “you get treats” in subsequent years.

6

On top of that, there is an age gap in the number of acts and the structure of act

sequences. For example, Gruendel (1980) examined the number of acts in an event and the

structure of act sequences with 4, 6 and 8 year olds with a cross-sectional design. Results

indicated that older children mentioned more main acts and sequenced them with more complex

structures. She categorized types of children’s act sequences into simple sequence, complex

sequences and hierarchical sequences. The complex sequences included sequences of acts

specifying preconditions for occurrence of acts, often marked by relative adverbs or

conjunctions such as “if-then” or “when” (For example, “When you go to the dentist, you have

to book an appointment first”), optional acts (e.g., “Sometimes they have three games”) and

optional orders of acts (e.g. “you open your presents before… or you can open it after…”).

Most children at age four sequenced acts in a simple way (e.g.,“ you eat cake and you go

home”). By contrast, one third of the narratives in older children were characterized as complex

sequences but the frequency of hierarchical sequences (e.g., “you get ready for the

party…putting balloons up…making a cake”) was still low. These age differences in the kind of

structures produced were also found when the children produced a script about a novel

experience. In an experiment designed in a laboratory setting (e.g., asking children about ‘a trip

to the wizard room’), older children reported more acts with more complex structure than

younger children (Farrar & Goodman, 1990; Price & Goodman, 1990). However, it should be

noted that age and experience are often confounded in research, as older children obviously

have more experience with various events (Slackman, Hudson & Fivush, 1986).

It is useful to distinguish between “weak” and “strong” scripts (Abelson, 1981). A

weak script specifies actions in an event but does not specify the order in which they can occur,

whereas a strong script specifies the order as well as the actions. For example, a‘ birthday party’

event generally includes actions such as ‘eating a cake’ and ‘playing games’, but these actions

can happen in any order. In contrast, ‘making cookies’ has a strong script in the sense that

7

actions such as ‘making dough’ or ‘putting the dough in the oven’ are expected to occur in a

specified order. Studies found that children and adults recall better logically ordered activities

(e.g., making cookies) than arbitrarily ordered activities such as playing with clay, or a birthday

party (Brown, 1975; Fivush, Kuebli & Clubb, 1992; Hudson & Nelson, 1983; Nelson &

Gruendel, 1986; Price & Goodman 1990; Ratner et al., 1986). Moreover, 5 year-old children

reported more acts and sequenced the acts more accurately in a weak script than 3 year-old

children (Fivush, Kuebli & Clubb, 1992; Price & Goodman, 1990). Fivush and his colleagues

(1992) suggested that young children might have more difficulty in organizing and recalling an

arbitrarily ordered weak script.

To summarize, English speaking monolingual children are able to produce a script at an

early age. Children’s script narratives are more generalized, elaborated and complex as they get

older. Furthermore, older children are more likely to report on weak sequences than younger

children.

The development of scripts in bilingual children

Whereas much is known about the development of scripts in monolingual children,

research on the development of scripts in bilinguals is in its initial stage. Two recent

longitudinal studies by Schwartz and Shaul (2013, 2014) examined the role of bilingual versus

monolingual preschool education in the development of scripts in two typologically different

languages: Russian and Hebrew. They investigated the developmental trajectory of knowledge

about slot fillers in Russian (L1) and Hebrew (L2), comparing 3 to 5 year-old bilingual children

attending bilingual (Hebrew-Russian) preschools, where language and early literacy are

fostered in both languages, and bilingual children attending monolingual (Hebrew) preschools

whose L1 literacy was not supported at school. The children’s knowledge about slot fillers was

analyzed by counting items about actors, actions and props in children’s narratives.

8

The two studies found that bilingual education does not hinder the acquisition of slot

fillers in both languages. In a study with 3-year-olds (Schwartz & Shaul, 2013), the children

were tested at three time points during an academic year. The bilingual children attending

bilingual preschools outperformed significantly on three items except for actors at Time 3 in L1

although at Time 1, both groups showed similar performance on three items except for actors in

L1. In L2, the bilingual children attending bilingual preschools showed an increase in the actors

and props categories such that significant gaps with the comparison group disappeared by

already at Time 2. Similar findings were reported in their other study (Schwartz & Shaul, 2014).

The other study (2014) found that at age four, the bilingual children attending bilingual

preschools had an advantage over bilingual children attending monolingual (Hebrew)

preschools on knowledge about slot fillers in their L1 (Russian). In addition, the bilingual

children attending bilingual preschools showed similar levels of the knowledge about slot fillers

in L2 to bilingual children attending monolingual (Hebrew) preschools just after two years of

L2 instruction.

In addition, the authors (2013) examined “interdependence” of knowledge about slot

fillers between L1 and L2 by counting the overlapping items between two languages (e.g.,

‘cake’ in L1 and L2). According to Linguistic Interdependence Hypothesis proposed by

Cummins (1981), underlying cognitive proficiency can transfer from one language to another

provided one has achieved a certain threshold of proficiency in the second language. The results

revealed that there was a significant difference between the two groups at Time 3 showing

superiority of the bilingual children attending bilingual preschools over the bilingual children

attending monolingual preschools, supporting the interdependence of knowledge about slot

fillers between two languages. These findings are interesting because interdependence between

L1 and L2 knowledge about slot fillers was found on the onset of its development and for the

9

bilingual children attending bilingual preschools in an educational setting, where instruction of

both languages is supported.

Of high relevance to the present study is the finding that continuous development of L1

language and literacy within a bilingual education context did not impede development of

knowledge about slot fillers in L2. On the other hand, lack of systematic L1 in the monolingual

school context seemed to hinder acquisition of knowledge about slot fillers in L1.

Development of stories in bilingual children

Research on L2 stories in bilingual children is limited. Thus, it is too early to draw a

conclusion about how different children’s L2 stories are from those produced by monolingual

children. However, the existing studies suggest that bilingual children show comparable

performance on story grammar (story schema) in L2 to monolingual children. The main

differences between L1 and L2 are noted on language indices such as morpho-syntactic

constructions and vocabulary (Berman, 2001; Gutiérrez-Clellen, 2002; Hipfner-Boucher, 2011;

Pearson, 2002). Berman (2001) argued that children rely on similar strategies for

conceptualization, planning, and organization of their narration regardless of whether they are

bilinguals or monolinguals, although bilingual or low SES children may be less proficient in

syntactic organization or vocabulary knowledge.

Research on stories within bilingual children’s two languages has focused largely on

three lines of research. The first line concerns differences and similarities in the development of

narratives in L1 and L2. For example, a one-year longitudinal study investigated developmental

change in L1 (Spanish) and L2 (English) narratives among 24 low SES bilingual kindergarten

children who received instruction in English (Uccelli & Páez, 2007). The mean age of the

participants at the onset, when children were in kindergarten was 5.58, and the second data

10

collection was done at the end of first grade. The authors examined narrative productivity (the

number of words and the number of different words) and narrative quality (story score and

language score) based on the measures Pearson (2002) devised. While the story score is a set of

measures to capture children’s ability to include the elements of story grammar, the language

score examined linguistic aspects of children’s narratives such as complex syntax, target nouns

of the story and morpho-syntactic accuracy. They found significant gains on all narrative

measures in English (i.e., lexical diversity, story score and language score) but not on narrative

length. In contrast, in Spanish, story score was the only measure that showed significant

improvement. In other words, children improved their story telling primarily in the language of

instruction. In a cross-sectional study, Kupersmitt and Berman (2001) studied the narrative

development of bilingual children ranging from 4 to 12 in age. The participants attended

Hebrew (L2) monolingual school and came from middle class homes where Spanish (L1) was

dominantly spoken. The authors found that the children showed similar story structure

development in Spanish and Hebrew while there were some differences in language expression

such as tense and aspect. In addition, they demonstrated that story grammar develops as a

function of age rather than the degree of mastery of the language. These findings suggest that

story structure is a general cognitive process that is less dependent on proficiency in L1 or L2.

This argument was also made by Viberg (2001) who investigated narrative development in

bilingual children from age 5 to 12 in Sweden on basis of data provided by various projects

studying the narratives of bilingual children. She also suggested that story grammar reflects

general domain knowledge which is available in L2 as well, and that can be expressed once a

certain linguistic threshold has been passed in L2.

A second line of research has examined cross-linguistic prediction. For instance, in a

study of Spanish English bilinguals in grades 2 and 5, Pearson (2002) examined whether

11

narrative abilities in one language predict children’s narrative abilities in the other language.

She found that story score and complex syntax correlated significantly across the two languages.

However, subcomponent measures of language score such as morpho-syntactic accuracy and

lexical diversity were not correlated across languages. She explained that because story score

and complex syntax represent relatively general capabilities, they show cross-linguistic

correlations (i.e., transfer), whereas morpho-syntactic accuracy and vocabulary are language

specific and depend on proficiency in each. Similar cross-linguistic associations were found in

the study by Uccelli and Páez (2007). The authors examined the correlations among narrative

skills and vocabulary skills in English-Spanish bilinguals. Kindergarten Spanish story score

predicted English narrative quality (story score and language score) in grade1, even after

controlling for English vocabulary and English narrative productivity. However, cross-

linguistic relations in predicting Spanish narrative quality were not found.

Other researchers have investigated what can affect discrepancy in narrative production

in bilingual children’s stronger and weaker language. For instance, Fiestas and Peña (2004)

studied the effect of language on 4 to 6 year old children’s narratives elicited in two different

task conditions. The participants had similar proficiency in Spanish (L1) and English (L2).

Children’s narratives were generated in response to two different stimuli; a wordless book

“Frog, where are you?”(Mayer, 1969) and a picture depicting a family birthday party. Children

included more attempts and initiating events in Spanish but more consequences in English,

even though they told equally complex stories in both languages. In addition, task effects were

found in both languages. Children used more Spanish-influenced utterances in terms of

grammar in the book condition but for the picture condition, more English-influenced utterance.

The authors explained that children might be able to tell a complex story, which requires

12

greater cognitive linguistic load than a photo task by using their first language, which is the

stronger language.

To summarize, within the context where L2 instruction or exposure is supported, L2

narratives in bilingual children may be comparable to narratives in monolinguals. Less

exposure to the L1 appears to hinder L1 narrative development in bilingual children. A few

studies suggest cross-linguistic relations concerning macro components such as knowledge of

slot fillers, story grammar and complex syntax between L1 and L2 narratives. By contrast,

language specific features were not found for cross-linguistic relations.

The present study

Previous studies have focused primarily on the macro level of scripts such as the acts

sequences in an event and knowledge about slot fillers. Most research has investigated bilingual

children who have a similar proficiency in L2 with their L1 or are exposed to L2 as a societal

language. Less is known about the development of narratives in emergent bilingual who are just

starting to learn L2. Studying emergent bilinguals who have not had exposure to L2 outside

school provides a unique opportunity to observe the development of narrative skills in the home

language and the emergence of narrative skills in L2. With the exception of two studies

(Montanari, 2004; Ucelli & Páez, 2007), most of the research in this area has been based on

cross-sectional designs. The current study utilized a cross sequential design, which the oral

narrative ability of 4 year-old cohort and 5 year-old cohort whose home language is English and

who attended a bilingual English-Hebrew program was tested twice over a one year period.

With all of this in mind, the first objective of the present study was to characterize the

development of scripts in two languages in emergent bilingual children and compare the

development in the two languages at the macro and micro levels.

13

The second objective of the study was to examine cross-linguistic relationships between

L1 (English) and L2 (Hebrew) narrative abilities. It is not known whether cross-linguistic

relation would be found in narratives in the two languages in emergent bilingual children.

Furthermore, available research into bilinguals’ narratives is limited to Spanish-English

bilinguals. Spanish and English are typologically similar as compared to Hebrew and English.

They share the Latin alphabet, many cognates (i.e., words with a common Latin root such as

information (English) and información (Spanish), and principles of word derivation (e.g.,

inform - information; informar - información). By comparison, English and Hebrew are more

dissimilar in terms of lexical structures, morphology, syntax, and orthography. Therefore, this

current study will shed light on whether cross-linguistic relations would be dependent on

language distance.

The following research questions guided the present study:

1) How do scripts in L1 (English) and L2 (Hebrew) develop in children from 4 to 7 years old in

terms of macrostructure and microstructure of scripts? How does the developmental pattern of

children’s narratives compare across English and Hebrew?

2) Are narrative abilities in the stronger language (L1) associated with those in the weaker

language (L2)?

METHODOLOGY

Participants

This thesis is part of a larger longitudinal study designed to investigate the development

of emergent language and literacy skills in young bilingual children. The participants were

recruited from a private English Hebrew bilingual school in a metropolitan area in Canada.

14

Participants included fifty-seven children in Junior Kindergarten (JK) and Senior Kindergarten

(SK) at the beginning of the study. They were followed over one year and tested at two time

points: May of 2011, (i.e., end of JK and SK) (Time 1) and May 2012, (i.e., end of SK and Gr1)

(Time 2). The mean age of the 33 children in the younger cohort (Female 16 and male 17) was

4 years 10 months and the mean age of the 24 children in the older cohort (14 female 10 male)

was 5 years and 9 months at Time 1. It should be noted that the participants in the sample came

from middle to upper middle class families, with parents who had at least an undergraduate

degree, and spoke English at home. See the table 1 for distribution by gender and age for two

cohorts.

The children were first exposed to Hebrew in JK. In JK and SK, children at this school

learn Hebrew for 40 minutes per week, in two, 20-minute sessions. However, by Grade 1, they

spend half of their school day learning Hebrew, taught by native speakers of Hebrew, and the

other half learning English, taught by native speakers of English. In other words, by Time 2,

children in the older cohort were finishing Grade 1 and have had much more intensive exposure

to Hebrew than they had for the previous 2 years. The Hebrew program for children in SK

emphasized the acquisition of hidden person pronouns and more verbs, nouns and adjectives,

and, aimed for children in Grade 1 to have some reading and writing fluency in Hebrew.

Measures

In order to examine possible differences in general abilities between two cohorts, two

standardized tests were administered in English at Time 1: Receptive vocabulary and Rapid

Automized Naming. Children’s narratives were assessed in English and Hebrew at Time1 and

Time 2.

Standardized Tests

15

Receptive Vocabulary. The Peabody Picture Vocabulary Test 4(PPVT-IV; Dunn &

Dunn, 2007) Form B was used to assess children’s receptive vocabulary. Children are required

to point to the picture that best illustrates the meaning of a word read aloud by the examiner.

The PPVT-4 consists of 228 items and items cover 20 categories of content (e.g., actions,

vegetables, tools) and part of speech (e.g., nouns, verbs, and adjectives) across all levels of

difficulty. The PPVT-4 is considered to be a reliable and valid test of receptive vocabulary.

Rapid automatized naming. Rapid automatized naming (RAN) was assessed using the

Objects Rapid Naming subtest from the Comprehensive Test of Phonological Processing

(CTOPP; Wagner et al., 1999). This task measures the speed with which children can name a

series of objects on two pages. Each page contains pictures of 36 objects which are arranged

((i.e., pencil, star, fish, chair, boat, key). The pictures are presented to the child and the child

starts to name them as soon as possible until all of the objects have been named. The total

number of seconds to name all of the objects on both pages is taken. The test manual reports a

coefficient alpha of .82 for five year-old children, .81 for 6-year-old children and .79 for 7-year-

old children.

Narrative Assessment

Procedure and transcription

The children were individually tested in a quiet room at the children’s school. Oral

narratives in each language were elicited in two separate sessions that were 2-4 weeks apart.

The children were tested by trained research assistants who were either undergraduate or

graduate students. Each session was conducted in one language by a research assistant who was

a native speaker of the language. Children were tested first in English, their home language so

16

that they could familiarize themselves with the task and feel comfortable about all the tasks. All

sessions were audio recorded.

The task of describing a birthday party was chosen for eliciting narratives for several

reasons. First, birthday parties are a familiar and interesting event to all children. Second, it is

linguistically and culturally appropriate for children. Third, it was expected that because

birthday parties can be characterized as a weak script, the topic would provide a rich context in

which children can sequence acts in various ways.

Following the elicitation procedure described by Nelson and Gruendel (1986), the

introductory statement, “I know you know a lot about birthday parties. Can you tell me what

happens when there is a birthday party?” was first presented. If the child was silent or paused,

the examiner provided encouraging prompts such as “Can you tell me more?”. If a child needed

further prompting, the examiner would provide more specific prompts such as “‘what is the first

thing that happens when …”, “What happens next?”, and “What happens at the end?”. The

same procedure was followed in the Hebrew session and repeated a year later, at Time 2.

However, the instructions for the Hebrew task were provided in English because children’s

command of Hebrew was minimal and it was important to ensure that children understood what

was expected to them.

The children’s narrations in each language were transcribed verbatim and they were

then segmented into “Communication units” (C-Units; Loban, 1976) by research assistants who

were native speakers of each language. A C-Unit is defined as an independent clause plus

modifiers attached to it. Following Craig, Washington, and Thompson-Porter (1998),

successive main clauses linked by simple coordinating conjunctions (e.g., ‘and’, ‘but’ and ‘or’)

were segmented into separate C-Units if the second clause included a subject. For example,

17

“You have a cake and play a game” was segmented into one C-Unit because the subject “you”

was omitted in the second clause.

In order to ensure accuracy of C-Unit segmentation and transcription in the two

languages, the second researcher assistant coded randomly selected 30% of the narrative

samples (n= 34, each language) and coded them independently. Item-by-item comparisons

were made to determine agreements. Inter-rater agreement scores were calculated by dividing

the number of agreements by the sum of the number of agreements and the number of

disagreements. Reliability of C-Unit segments between the first coder and the second coder was

96% for English samples and 95% for Hebrew samples. Reliability of transcription was 99%

for English and 97% for Hebrew. Any disagreements were resolved through discussion between

the two research assistants.

Acknowledging that a couple of prompts given to children in this study can structure

children’s narratives, the narratives produced before prompts such as “’What is the first thing

that happens when …’, ‘What happens next?’, and ‘What happens at the end?’ were analyzed.

In addition, if the structure of a more complex act sequence is present in a narrative, it was

coded as more complex structure.

Coding System

Children’s narratives were assessed in terms of both macrostructure and microstructure.

At the macro level, four measures were used: (1) type of output, (2) slot fillers, (3) act

sequences”, and (4) object grouping. Narrative micro level was coded on indices of language

productivity and complexity. Since the children were in the process of starting to learn to read

and speak Hebrew, their Hebrew narrative production was minimal. As a result, it was not

18

possible to assess ‘Type of Output’ and ‘Act Sequences’ for macrostructure in Hebrew. All

measures included were listed in Table 2 and were described more in detail below.

Coding for Macrostructure

Type of Output. Following Gruendel (1980), children’s narratives were coded as ‘script’,

‘script with episodic information’ or ‘total episode’. A narrative was rated as a script when it

was a general description marked with general pronouns (e.g. “You”, “They” and “It”) and

using the present tense. If episodic information is reported as the background to general account

whereas the foreground of the narrative is general, it is classified as a script (Hudson & Shapiro,

1991, p. 94). When a child talked about general birthday parties although he or she used

personal pronouns (e.g., “I” and “We”), it was coded as a script.

“They’re fun. You get to eat cake, you get to eat marshmallows and you get lots of

presents and you get to hand out loot bag.” (Child age 4; 10)

A narrative was rated as a script with episodic information when a child included particular

episodes representing personal experiences about birthday parties, and where a personal

pronoun “I” or “We” was used in the past tense:

“People come in. We play a little, then I blow out the candles. Then after that we could

play a little more. And when everyone is gone, I go back to my home. When it was my

birthday party, I went to a bowling place. I had so much bowling. Me then Lela then

Axel.”(Child age 4;9)

When a child talked about the event that happened in the past, the narrative was rated as

‘total episode’. For example, “I went to my friend’s birthday party. It was a trampoline party. I

gave my friend a small gift”.

Slot fillers. Proposed by Schank and Abelson (1977), a script is made of slots for

variables such as actors, actions, and props in an event. These variables can be filled with a

19

range of probable alternatives, which are slot fillers. That is, slot fillers are basic elements that

form macrostructure of a script.

Following Schwartz and Shaul (2013), slot fillers provided in the narrative were

classified into the following categories: actors, actions and props. The total frequency of each

category of slot fillers was analyzed. In the following example, two actors were mentioned: the

birthday girl and boy, as well as three actions: eating, singing, and getting a present. In addition,

four props were mentioned: present, cake, junk, and happy birthday song.

“When it’s your birthday, you get a present and then you have cake. You have junk and

play. They sing happy birthday to the birthday girl or boy.”

Object grouping. Nelson (1985) proposed that children develop knowledge of

taxonomic categories though abstraction of patterns of relationships in events. As noted by

Rosch et al. (1976), the category of object can be classified into superordinate, basic and

subordinate nouns according to inclusiveness of a category. Superordinate nouns (e.g.,

‘furniture’) reflect a high level of inclusiveness and represent a high level of abstraction

whereas subordinate nouns (e.g., ‘bunker bed’) represent a low level of inclusiveness and a low

level of abstraction. Based on the classification, we tracked how children’s use of object

category changed over time in the narratives. Slot fillers of actors and props in the narratives

were classified into superordinate nouns (e.g., ‘food’), basic nouns (e.g., ‘cake’) and

subordinate nouns (e.g., ‘chocolate cake’) and counted for the analyses.

Act Sequences. As Schank and Abelson (1977) defined, scripts are spatially and

temporally organized. The structure of a script is reflected in how a child sequences acts in an

event (Nelson & Gruendel, 1986, p.35). Act sequences in children’s narratives were coded into

one of four categories, which reflect complexity in act sequences based on Gruendel (1980)’s

classification. : 1) no sequences, 2) simple sequence, 3) complex sequence and 4) hierarchical

20

act sequences.

Either when a child describes one act such as “You have a cake, juice and snack” or when

a child doesn’t sequence acts “You have a cake. You play. You go”, it would be coded as ‘no

sequences’. When a child links acts simply through temporal order, for example, “You eat a

cake and open the presents and then you get a loot bag and go home”, the narrative was coded

as ‘simple sequences’. A narrative was coded as complex acts sequence when (a) the child

employs subordinating conjunctions such as “if”, or “when” to qualify acts or 2) specifies

alternative acts or 3) describes acts can occur in an alternative order. In hierarchical act

sequences, which are the most complex, one or more acts are embedded in head acts that

subsume other acts. Two examples of complex, hierarchical acts sequences, produced by the

participants are presented below.

“You’re celebrating how old you turn, and if you didn’t have a birthday then you would

stay like a baby. People who you know can come to your birthday party, but if you don’t

want people to come to your birthday it’s okay because you can spend time with your

parents. Some people don’t really like want people to come to their birthdays but people

could come to their birthdays” (Child age 6; 3)

“When there's a birthday, you can play with your friends. There's lot of things to do at

birthday parties. In some birthday parties there is soccer. Some birthday parties are

sleepovers. Some birthday parties are playplaces at the playplaces. A birthday party is to

celebrate the day that someone was born. And they wanna celebrate it by having a lot a

lots a lot of fun.” (Child age 7; 3)

Coding for microstructure

Language Productivity. Language productivity is often measured by the number of

words (or morphemes) or utterance units used in the narratives. However, this metric is

problematic because Hebrew is a highly inflected language. This means that the same

21

information is presented in fewer words in Hebrew than in English. For example, in English,

the phrase “in the park” has three words but the Hebrew parallel, “בפארק” (ba/park) is one

highly inflected word. For this reason, the number of morphemes not the number of words was

calculated to enable comparison across the two languages. Thus, two measures were included:

Total number of morphemes and the total number of C-Units in the narratives.

Syntactic complexity. Syntactic complexity can be measured by various syntactic

indices such as ‘mean length of C-Units in morphemes’ and cohesion. Cohesion concerns

linguistic devices such as conjunctions indicating the semantic relationship between

prepositions (Cain, 2003; Geva & Ryan, 1985).

The narratives were analyzed in terms of four measures of syntactic complexity (Justice

et al., 2006): mean length of C-Units in morphemes, total number of complex C-Units, total

number of subordinate conjunctions, and proportion of complex C-Units. A complex C-Unit

refers to a C-Unit containing an independent clause and at least one dependent clause. For

instance, “You have a cake, which your mom bakes for you” is a complex C-Unit.

Inter-rater Reliability

In order to establish the level of agreement between two independent raters on all

measures, a doctoral student and a research assistant completed training for the coding systems

for English and Hebrew narratives respectively. Following the training, 30% (n=34, each

language) of narrative samples at the two time points were randomly selected for independent

coding by the two raters. The Kappa value for categorical measures was performed and the

Cronbach’s alpha score for continuous measures were calculated. The Kappa value was .892 for

Type of Output and .740 for Act Sequences for English narratives suggesting that the

agreements are considered substantial (Landis & Koch, 1977). Cronbach’s alpha for continuous

22

variables for English and Hebrew samples ranged from .93 to 1.00 and .98 to 1.00 respectively,

which indicate a high level of inter-rater agreement. All disagreements were resolved and

resulted in 100% agreement after discussion.

RESULTS

A series of independent sample t-tests were conducted to evaluate whether there were

differences in PPVT and RAN between the two cohorts. The results revealed no significant

difference in the standard scores of the PPVT, (t (55)=. 98, ns) and the standard scores (t (53)=.

69, ns) and raw scores (t (53)=1.20, ns) of RAN between two cohorts. It should also be noted

that the PPVT and RAN scores were within the average range.

Initial exploratory analyses revealed that the distributions of the majority of the

continuous dependant variables under study were positively skewed. Therefore, it was not

possible to conduct repeated measures ANOVA (analysis of variance) since ANOVA assumes

normal distributions of dependent variables in each cohort. As a result, non-parametric

procedures, which do not require a normal distribution, were performed for continuous

dependant variables as well as for categorical dependent variables. In order to test the effect of

time for the continuous dependent variables in each cohort, either the Wilcoxon test or the sign

test was used. Since the Wilcoxon test assumes that the distribution of the difference scores is

continuous and symmetrical in the population, for the variables that have not met the

assumption, sign tests were performed. Since it is not possible to perform mixed analyses in

nonparametric procedures in SPSS at a time, a series of analyses were run separately for each

cohort and each language.

The first purpose of this study was to characterize the development of scripts in L1 and

L2 in children from 4 to 7 years old. The results are divided into macrostructure and

23

microstructure of scripts in each language and are followed by qualitative observations of the

first research question.

The pattern of change in L1

Macrostructure of scripts in L1

As can be seen in Table 3, frequency distribution on type of output revealed that

children’s scripts in L1 were all either script or script with personal episode and there was no

‘total episode’ for either of the cohorts. As a result, the type of output variable was

dichotomized. Therefore, in order to determine if there are differences on type of output

between Time 1 and Time 2, the McNemar test was performed. The results suggested that there

is no significant change between the proportion of types of output in Time 1 and the proportion

of types of output in Time 2 for both cohorts.

The analyses on all four measures of slot fillers (i.e., actors, actions and props) showed

no significant improvement on any measures for the younger cohort. But, it is worth noting that

the change on props from Time 1 to Time 2 approached significance (z= 1.93, p=.054). The

results for the older cohort indicated no improvement on any of the slot filler categories. With

regard to the object grouping variable, both cohorts showed significant improvements from

Time 1 to Time 2 only on the superordinate class (see Table 4 and 5 for p values).

As the widely accepted rule of thumb, the frequency of one category should be greater

than five for adequate test approximation when calculating non-parametric statistics. As the

frequency in both a single act category and hierarchical sequences category was less than 5 for

both cohorts, a single act category was collapsed into simple sequence category and

hierarchical sequences into complex sequences category respectively. Using the McNemar test,

a significant tendency was found for the structure of act sequences to more likely be changed to

24

complex sequences over one year for the younger cohort (p= .001). In contrast, no significant

change was found for the older cohort (see Table 3).

Microstructure of scripts in L1

Descriptive statistics, test type and p value for the measures of language productivity

and complexity for each cohort are summarized in Table 4 and Table 5. In terms of productivity,

a significant improvement was found for the younger cohort in total number of C-Units (TNC-

U), total number of complex C-Units (TCC-U), and mean length of C-Units in morphemes

(MLC-UM) of complexity: TNC-U, z=2.25, p<.05; TCC-U z=2.45, p<.05; MLCU-M z=5.57,

p<.01. However, the older cohort improved significantly only in MLC-UM of complexity.

z=4.70, p<.01.

The pattern of change in L2 Narratives

At Time 1, there was very little variation on most of the continuous measures for both

cohorts. Most children scored 0 on the continuous variables and there were a few outliers in

each cohort. At Time 2, performance of both cohorts was still low on superordinate,

subordinate class measures of object grouping variable and all of the language complex

measures except for mean length of C-Units in morphemes. Therefore, it was impossible to

apply inferential statistical analyses on these measures. Descriptive statistics, and test type used

and p value for continuous variable for both cohorts are presented in Table 6 and Table 7.

Macrostructure of scripts in L2

The results showed that both cohorts improved significantly on slot filler categories: For

the younger cohort, Actors z=5.004, p=. 000; Actions z= 3.75, p=. 000 ; Props z= 3.064, p

=.001;Total sum of slot fillers z=4.619, p =.000. For the older cohort, Actors z=3.928, p =. 000;

25

Actions z=2.598, p =. 006; Props z=3.25, p =. 001; Total sum of slot fillers z=3.624, p =. 000.

In addition, significant improvement from Time 1 to Time 2 was found for both cohorts, with

both cohorts improving on basic class. (The younger cohort, z=3.32, p=. 001; the older cohort,

z=2.025, p=. 043)

Microstructure of scripts in L2

The results revealed significant gains for the younger cohort on TNM (z=3.336, p=. 000)

and on MLC-UM (z =3.73, p =. 000). Although no significant differences were found on the

two measures for the older cohort, the differences were almost significant on both measures for

the older cohort (z =1.75, p=. 078).

Qualitative investigations into the development of scripts

Qualitative examination of children’s narratives provides an in-depth understanding of

L1 narrative development by looking at some examples of narratives produced by children.

Given that statistical analyses on a lot of measures in Hebrew could not be conducted,

qualitative observation of Hebrew narratives provides complimentary insights into L2 narrative

development.

As shown in Table 3, although there was no significant improvement over time for both

cohorts on ‘Type of output’, most of children (75% of the younger cohort and 79.2% of the

older cohort) who were at age 4 and 5 at the first data collection were able to generate a script

quite well using neutral subjects “ you”, “they” and “it” and present tense verbs. A few children

in both cohorts used their episodic information as a background to their general account about a

26

birthday party, which is one of the characteristics of a script. For instance, Sydney in the

younger cohort reported her personal experience to support the foreground of her narrative.

“You can play, and you have pizza and you can play with your friends, you sometimes

fight… you take off your shoes sometimes, and sometimes you can hurt yourself, like

one time I hurt myself when it was one of my friends birthday parties, and whosever

birthday party you had you get, they get a lot of presents...”

The children who included a partial episode tended to report their personal experience

when a tester asked them more narration after they described a birthday party in general. The

pattern was found in 18 narrative samples out of 22 narrative samples with a partial episode.

Thus, it might not be the case that children who included a partial episode haven’t established a

script about a birthday party in mind. Rather, they seemed to lack genre awareness about a

script.

“There’s balloons, sometimes there’s piñatas. you have to put the presents somewhere.

… (Anything else you want to tell Ms. Bear about birthday parties?) My birthday party

was actually at the bowling club in front of this. Its this birthday party at the top that

you play games, and then on the two bottom floors.”

“you can have, you can have, loop bags, birthday presents, birthday cakes, sing songs,

people, people from Disney world. And you can have invitations, … (Can you tell me

anymore?) You say goodbye and then you have lots of stuff to do, then you go home,

you go home after birthday party. I couldn’t go to one, to some birthday parties for my

friends so I had skiing on one, and my dance recital at the same time. I missed three

birthday parties. I had a swimming party at the same day at the same time. At here at the

school. ”

With respect to the slot fillers variable, although no significant improvement was found

for either cohort, children’s knowledge about schema components has been developing between

Time 1 and Time 2. Slot fillers produced by children on the whole increased except for the

action category for the older cohort. Specifically, the knowledge about the props category was

improved for the both cohorts and these changes were almost significant, the younger cohort

27

p= .054; the older cohort p=. 078 (see Table 4). For instance, Andrew who was in the younger

cohort mentioned a few actions such as ‘eating’, ‘playing’ and ‘getting picked up’ but the

information about props was not produced at Time 1. However, at Time 2 Andrew included

more props which can be seen at a birthday party such as balloon, cake, pizza, game,

gymnastics, laser tag, a birthday present and so on.

“It’s a kinda party. You play. (Yes. And what happens next?). You eat. (Yes. And

then?). You go to play again. (Yes, and what happens at the end?). You get picked up.”

(Time 1)

“Birthday parties have balloons, cake, lots of people, sometimes it’s gymnastics or laser

tag, there’s a birthday present. You go into the room and you do what you have for your

birthday. Then after you do all the games and stuff and everything, then you go and

have cake, and pizza. (You get to do whatever you want sometimes, or sometimes you

play a little bit more.” (Time 2)

The significant development of object grouping over time was found for both cohorts

only on superordinate class category. At Time 2, children mentioned more superordinate class

nouns such as activities, food, people, vegetables, fruit, sports, place, staff, guest, family, arts

and crafts and drinks.

The significant change in the structure of act sequences was made for the younger

cohort. The structure of act sequences became more complex. In other words, linear temporal

structure of act sequences became complex by employing complex temporal structure or

conditional and clausal structure. As can be seen in Laura’s narrative below, at time 1 she

connected acts with ‘and’ or ‘and then’ reflecting a linear temporal structure. At time 2, she

mentioned optional paths of acts, e.g., “you can have one at your house or you can go

somewhere for it” and employed conditional structure in connecting acts with the use of ‘if’ or

‘when’ subordinating conjunctions.

28

“You sing songs, then you blow out the candles, and you get a present and you eat cake

and you play party games, you can make stuff .you let everyone come in...You play

games… You eat the cake and then go back to your house.” (At Time 1)

“You can have any kind of birthday party, you can have one at your house or you can go

somewhere for it.…… it's not really fun when you don't invite anybody but if you invite

people it's more funnier because it's nice and you get to see your friends and you can be

nice and you have to be nice and you have to not, like if someone does something bad

you don't say that's not how you do it.” (Time 2)

Subordinating conjunctions produced in children’s L1 narratives were further analyzed

to identify their subtypes and categorized into temporal (e.g. before, after), causal (e.g. because,

as) and conditional (e.g. if, unless). Each category was coded as either present (1) or absent (0).

The McNemar test revealed no significant change on any subtype over time but improvement

for the older cohort on conditional subordinating conjunctions was almost significant (p=. 062).

At Time1, 18.2% of the younger cohort and 25% of the older cohort were able to include/use

temporal subordinating conjunctions (e.g. “when” “before”) but the use of other types of

conjunctions either had not appeared yet or were very rare in the narratives. At Time 2, both

cohorts started to use conditional and causal conjunctions. While the use of causal conjunctions

was still rare in both cohorts, conditional conjunctions were used often such that the difference

was almost significant for the older cohort as noted earlier.

The focus of the development of children’s emergent narrative abilities in Hebrew was

on building up slot fillers. As mentioned above, at time 1, both cohorts either scored 1 or 0 on

all categories of Slot fillers but at Time 2, they reported actors and props (e.g., “birthday” “cake”

“pizza” “candle”, “grandma”, “boys”, “girls” , “mom” and “dad”). At Time 2, children started

to mention main actions such as “eating” and “playing”. Although a few actions were

mentioned in children’s narratives, the structure of act sequences was still a single act. In other

words, children focused on describing each act rather than sequencing the acts, e.g. “The

29

children eat the cake. Children sing songs. Children play” or “I eat pizza. I open birthday

presents”. With respect to object grouping, children mentioned basic nouns such as ‘cake,

birthday pizza, song, balloon, yogurt, apple and present’ but only three children at Time 2 used

superordinate nouns such as ‘people’ and ‘food’ and two children reported subordinate nouns

such as ‘chocolate cake’ and ‘Tapuz’.

Cross-linguistic relations between narratives in L1 and L2 at Time 1 and Time 2

The second objective of the present study was to investigate cross-linguistic relations

among narrative measures across English and Hebrew. Since Hebrew narratives of both cohorts

were minimal at Time 1, narrative measures produced at Time 2 were chosen to examine the

relations. Spearman’s rho coefficients were estimated because most of the continuous variables

under study were not normally distributed and there were some outliers. For correlations

between categorical dichotomous variables, Phi coefficients were reported. Table 8 and Table 9

display cross-language correlations among narrative measures in English and Hebrew for each

cohort at Time 2. A cross language relation was found for the older cohort. The English

measure of the number of subordinating conjunctions was moderately associated with the

Hebrew measure of the total sum of slot fillers. (r=.405, p<.05). No other correlations were

significant.

DISCUSSION

L1 and L2 narrative development in emergent bilinguals

The first objective of this study was to investigate the development of scripts in L1 and

L2 at the macro and micro levels in 4 to 7 year-old emergent bilingual children. Previous

research has only targeted the macro level in studying a script. This study incorporated two

30

dimensions of narrative to provide a global picture about the development of scripts. The results

from this study indicated that children at age 4 have established a generalized macrostructure of

script in L1. L1 narratives have been developing specifically in terms of complexity in macro

and microstructure of the narratives. In their emergent L2, children have been developing both

macro and microstructure of scripts.

Macrostructure of scripts in L1 and L2

Generalized and abstracted macrostructure of scripts in L1

The findings from this study suggest that the children as young as 4 years old have the

ability to generalize their narratives in L1 about birthday parties based on their past experiences,

as suggested in previous studies that involved monolingual middle class children (e.g.,

Gruendel, 1981; Nelson, 1978; Nelson & Gruendel, 1986). The children at this age do not

merely “rely on their perception of the here and now, but do create an abstract generalisation

about the event” (Schwartz & Shaul, 2013, p. 37). This was evident by the findings that no one

in this study produced ‘total episodes’ about personal occurrences and the children in this study

were able to use a particular episode which occurred to them in the past as the background or

commentary information to back up their general foreground case of the event.

Furthermore, the results suggest that children’s knowledge about slot fillers related to

the event become abstract over a year. In L1, the children’s knowledge about objects at basic

class was well established, so not much improvement was noted. An area in which one noted

improvement was in their familiarity with superordinate categories. Producing more terms in

superordinate class at Time 2 may indicate that as children learn more basic classes over

different occurrences of the event, children are involved in abstraction of alternative fillers for

the slots (Lucariello & Rifkin, 1986). These results confirm that objects at basic class are

31

generally acquired before objects at other classes of hierarchy are acquired as suggested in

other studies (Anglin, 1977; Mervis & Crisafi, 1982; Rosch et al., 1976).

In their emerging L2, although it was not possible to carry out parallel analysis with L1

narrative, qualitative observation showed that at Time 2 when the children obtained some

proficiency in Hebrew, they provided a general description about the event but not particular

episodes that happened in the past. One can assume that as the children have established a

generalized event representation in L1, emergent L2 narratives might be created in general

forms rather than particular episodes. Familiarity with basic objects was developing while

familiarity with superordinate and subordinate objects has not developed yet.

Knowledge about slot fillers and act sequences in L1 and L2

The results indicate that the younger cohort showed a significant improvement on the

structure of act sequences in L1 and on slot fillers in L2. In L1, when they were 4 years old, 30%

of the children were able to display complex act sequences. However, this percentage increased

to 82% by the time they reached the end of SK. Although there was no significant change for

the older cohort, the children’s ability to represent acts hierarchically at Time 2 emerged in the

older cohort. These results are consistent with the findings by Gruendel (1981) that major

change in the structure of act sequences occurred between four and six years with the

appearance of more complex structure.

Qualitative observations revealed that even though the children were not able yet to

produce full grammatical sentences in their L2, they attempted to mention relevant actions after

being exposed to more extensive Hebrew instruction at school for a year or two. Explicit

sequencing was not evident yet and instead listing concrete actions was apparent. A possible

32

explanation for this tendency in L2 is that without an array of language devices at their disposal,

it makes a cognitive load when they sequence the acts (Viberg, 2001).

On the other hand, with respect to the development of knowledge about slot fillers,

while the knowledge in L1 has plateaued, the knowledge in L2 has progressed significantly on

all the categories of slot fillers. This indicates that knowledge about slot filler in L1 has

developed well enough given that slot fillers about birthday parties are, to some degree, fixed

(Schwartz & Shaul, 2013), while the significant development in acquiring basic class words of

slot fillers has been made in L2 narratives.

Microstructure of scripts in L1 and L2

Microstructure of narratives in both languages was tracked in terms of both language

productivity and syntactic complexity. The findings suggest that in both L1 and L2, mean

length of C-Units in morphemes can be a robust measure that captures growth in syntactic

complexity in both children’s L1 and L2 narrative. Moreover, the results suggest that in

emergent L2 narratives, productivity indices such as total number of morphemes can be a

sensitive developmental measure. Past research (Muñoz et al., 2003; Ucelli & Paez, 2007)

proposed that the total number of words or morphemes (TNW or TNM) has been found to be

less robust as an indicator of developmental change in L2 narrative in bilingual children

compared to complexity measures. However, given that past studies investigated children with

proficiency in L2 as a societal language, their findings may not be applicable to emergent

bilingual children. Therefore, a sensitive measure to capture developmental growth in L2

narrative would be dependent on level of language proficiency in L2.

Although the use of subordinating conjunctions in L1 such as when, because and if in

the utterances of the both cohorts didn’t change significantly over one year, different types of

33

subordinating conjunctions increased from ‘temporal’ at Time 1 to ‘causal and conditional

subordinations’ at Time2. However, children in this study tended to use temporal coordinating

conjunctions such as ‘and’ and ‘and then’ as a main means of connecting two prepositions

(Berman, 2009; Geva, 2007; Geva & Olson 1983; Shapiro & Hudson, 1991). Clearly,

subordinating conjunctions are more challenging for young children to master than coordinating

conjunctions (McClure & Geva, 1983). In L2, the children were not yet able to make use of

coordinating conjunctions (e.g.,‘and’) or connect two propositions.

To summarize, the overall picture emerging from this study is that the L1 narratives, of

4 to 7 year olds have developed in terms of complexity at both the macro level and micro level.

Given that the children have well established event representation about birthday parties in L1,

the development of both dimensions in children’s L2 script is dependent on L2 language

proficiency. According to Slackman, Hudson and Fivush (1986), there are three levels of

representation for scripts. The first level refers to the actual event in the real world, and the

second level is event representation in memory. In the third level, verbal description of the

event knowledge takes place. Accordingly, the ‘birthday party’ script is stably established in

their mental representation through their L1, and the third level, that is, verbal expression in L2

has been developing through exposure to L2 at school.

The Association between narrative abilities in emergent bilinguals

The second aim of this study was to examine cross-language relationships among

narrative abilities in the L1 and L2. A cross-language relation was observed at Time 2 when the

children were a year older. Interestingly, the children who could use more subordinating

conjunctions in their L1 were more likely to construct slot fillers in L2 for the event than were

children who used fewer subordinating conjunctions in their L1. It could be said that the ability

to understand semantic relations between two different propositions and the ability to construct

34

basic elements of event representation reflect general domain abilities across languages. This

observation is in line with previous studies (Pearson, 2002; Ucelli & Páez, 2007). However,

given that unlike the children in the previous studies, the children in this sample had much less

proficiency in L2, the knowledge of slot fillers in L2 would be more dependent on language

proficiency. In other words, it would be difficult to say that the knowledge of slot fillers was

expressed as a function of general cognitive ability. Instead, this relationship suggests that

individual differences in the ability to use subordinate conjunctions in the stronger language is a

reflection of language ability which is related to familiarity with vocabulary, and in particular,

nouns in the emerging language.

Furthermore, it is noteworthy that despite differences between English and Hebrew in

language typology, the cross-linguistic association between narrative abilities was detected. As

previous studies have focused on English and Spanish, which have similarity in common, this

finding suggests that correlations between narrative abilities may be independent of linguistic

distance.

Interestingly, the cross-language association was found only in the older cohort at the

end of Grade 1, at a time when they were exposed to 2.5 hours of instruction in Hebrew for a

whole year, but not at an earlier time when exposure was limited to 30 minutes a week. With

this more extensive exposure to Hebrew, the children in the older cohort gained more language

proficiency in Hebrew. It was only at that time that cross-linguistic relation began to emerge

between narratives in L1 and L2. In other words, proficiency in the L2 has to pass a certain

threshold before it is possible to note more consistently cross-language associations. One may

expect more cross-language correlations later on as children continue receive more L2

instruction at school (Pearson, 2002; Uccelli & Páez, 2007).

Limitations of the study and suggestions for future research

35

Although this study provides insights into development of a script in emerging bilingual

children, it is important to address several limitations for future research directions. First, the

present study has a small sample size for each cohort. Since statistical testing results are

dependent on sample sizes, the limited sample may have obscured some significant results. A

larger sample is required to confirm the pattern found in this study.

The participants in this study were recruited within one private school and came from

upper middle SES families with parents who had at least an undergraduate degree. Given that

school curriculum and SES are factors impacting narrative development, it is possible that the

population in this study might not be generalized to other populations though they may be to

other day schools. Therefore, it is necessary to take these factors into account when comparing

the results in this study to those of other studies. Research has indicated that SES is one of

important factors impacting narrative development. That is, children from low SES homes have

difficulty using decontextualized language (i.e., narrative) because they are less exposed to this

type of language with their parents compared to middle SES homes (Hart & Risley, 1992;

Heath, 1994). Thus, as the population in this study might be advantaged in language

development, resulting in better narrative abilities in L1 than broader populations. Studies with

multiple cohorts from various schools are needed for more powerful results that can be

generalized to other populations.

The present study was conducted on a cross sequential design following two cohorts for

one year to characterize development of a script in children from 4 to 7 years old. Although

there were no differences in their PPVT and RAN between two cohorts, the findings in this

study need to be interpreted with caution having possible cohort effects in mind. Although it is

true there are difficulties such as time constraints and administrative restrictions in conducting a

36

longitudinal study, in the future research, more efforts should be made to follow the same

children from 4 to 7 for powerful findings.

Further limitation relates to the tool of assessing children’s script narrative at macro

level. The present study attempted to provide a suggestion of accessing macrostructure of a

script. As noted earlier, various assessment tools for a story retelling has been developed at both

levels. One can apply the tools for microstructure developed for a story retelling to a script but

not for macrostructure of a script. Thus, more efforts should be made to devise and refine the

scalable tool for macrostructure of a script.

The present study investigated narrative development in emerging English Hebrew

bilinguals who just began to learn Hebrew as their second language. Since proficiency in the

second language is minimal, there were limitations in statistical analyses. For example,

correlation analyses couldn’t be estimated because their L2 production was so minimal at Time

1. However, at Time 2 when they had more time to develop their L2, a significant cross-

linguistic relation for older cohort was found at this early stage of L2 development. Thus, in the

future research, it would be interesting to see how cross language relationships change as

children progress in L2 proficiency. At the same time, it would be meaningful to see if there is

an interaction between the two languages at both levels as a function of L2 proficiency.

Implications

The findings of this study could include key messages delivered to educators and

clinicians. First, scripts can be a good clinical tool to measure cognition and language

development in young children. As noted earlier, scripts in children are of great importance in

their cognitive and language development. This suggests that clinicians should make use of

scripts as well as story telling or retelling as an assessment tool. Second, the indices of language

37

complexity can capture more nuanced and sensitive growth in 4 to 7 years old in L1 than

language productivity measures. When assessing narratives in children ranging from 4 to 7 in

age, clinicians should put more attention to complexity in the language than length of narratives.

Third, lack of oral proficiency in L2 might prevent bilingual children from constructing

a coherent and cohesive narrative although they may possess the cognitive underlying abilities

for producing well-formed narratives. Therefore, in particular, in the case of bilingual children

who moved to other countries and are learning a societal language as a second language, it is

possible that educators might view them ‘at risk for language disability’ or ‘incapable of

organizing planning a narrative discourse’ (e.g., Limbos & Geva, 2001; Montanari, 2007), due

to their failure to successfully produce a narrative discourse in a classroom. The findings from

this study suggest that educators first need to look at a narrative in their stronger language (L1)

with other cognitive and linguistic tests before identifying them as cognitively deficient through

L2 narrative.

Finally, the cross-linguistic association found in the study suggests that cross-linguistic

relationships can be detected already at an early stage of L2 development. From this finding,

the key message delivered to parents or educators would be that at the early stage of learning

L2, exposure to two languages could be mutually facilitating.

38

REFERENCES

Abelson, R. P. (1981). Psychological status of the script concept, American Psychologist, 36(7),715-729.

Anglin, J. (1977) Word, object and conceptual development. New York: Norton

Applebee, A.N. (1978). The child’s concept of story. Chicago: The University of Chicago Press.

Berman, R. (2001). Crosslinguistic perspectives on narrative development. Epilogue. In L.Verhoeven & S. Strömqvist, eds. Narrative development in a multilingual context.Amsterdam: John Benjamins, 419-428.

Berman, R. (2001). Crosslinguistic perspectives on narrative development. Epilogue. In L.Verhoeven & S. Strömqvist, eds. Narrative development in a multilingual context.Amsterdam: John Benjamins, 419-428.

Berman, R.A. & Slobin, D.I. (1994). Relating events in narrative: A crosslinguisticdevelopmental study. Hillsdale, NJ: Lawrence Erlbaum Associates.

Brown, A. L. (1975), "The Development of Memory: Knowing, Knowing About Knowing, andKnowing How to Know," in Advances in Child Development and Behavior, (ed.), H. W.Reese, New York: Academic Press.

Craig, H. K., Washington, J. A. & Thompson-Porter, C. (1998). Average c-unit lengths in thediscourse of African American children from low income, urban homes. Journal ofSpeech, Language, and Hearing Research, 41, 433-444.

Cummins, J. (1978). Bilingualism and the development of metalinguistic awareness. Journal ofCross-Cultural Psychology, 9(2), 131–149.

Cummins, J. (1981). The role of primary language development in promotingeducationalsuccess for language minority students. In California State Department ofEducation (Ed.), Schooling and Language Minority Students: A Theoretical Framework(pp. 3-49). Los Angeles:Evaluation, Dissemination and Assessment Center CaliforniaState University.

Davies, P., Shanks, B., & Davies, K. (2004). Improving narrative skills in young children withdelayed language development. Educational Review, 56, 271-286.

Dunn, L.M., & Dunn, D.M. (2007). Peabody Picture Vocabulary Test-IV. Circle Pines, MN:American Guidance Service.

Farrar, M. J., & Goodman, G. S. (1992). Development changes in event memory. ChildDevelop ment, 63, 173–187.

Fivush, R., Kuebli, J. and Clubb, P. (1992). The structure of events and event representations: adevelopmental analysis. Child Development, 63, 188-201.

39

Fivush, R., & Slackman, E. (1986). The acquisition and development of scripts. In K. Nelson(Ed.), Event knowledge: Structure and function in development (pp. 71–96). Hillsdale,NJ: Erlbaum.

French, L. A. (1986). The language of events. In K. Nelson (Ed.), Event knowledge: Structureand function in development (pp. 119-136). Hillsdale, NJ: Erlbaum.

Genesee, F., Geva, E., Dressler, D., & Kamil, M. (2006). Synthesis: Cross-linguisticrelationships. In D. August & T. Shanahan (Eds.), Developing literacy in secondlanguage learners: Report of the national literacy panel on language-minority childrenand youth (pp.153-174). Mahwah, NJ: Lawrence Erlbaum.

Geva, E., & Olson, D. (1983). Children’s story retelling. First Language, 4, 85-110.

Geva, E., & Ryan, E. B. (1985). Use of conjunctions in expository texts by skilled and lessskilled readers. Journal of Reading Behavior, 17, 331-346.

Green, S. B., & Salkind, N. J. (2008). Using SPSS for Window and Macintosh: Analyzing andunderstanding data (5th ed.). Upper Saddle River, NJ: Pearson Prentice Hall.

Griffin, R., Hemphill, L., Camp, L., & Wolf, D. (2004). Oral discourse in the preschool yearsand later literacy skills. First Language, 24(2), 123-147.

Gruendel, J. M. (1980). Scripts and stories: a study of children’s event narratives, Dissertations

Gutiérrez-Clellen, V. F, Simon-Cereijido, G., & Erickson Leone, A. (2009). Code-switching inbilingual children with specific language impairment. International Journal ofBilingualism, 13, 91–109.

Hart, B., & Risley, T. (1992). American parenting of language-learning children: Persistingdifferences in family-child interactions observed in natural home environments.Developmental Psychology, 28,1096–1105.

Heath, S. B. (1994). What no bedtime story means: Narrative skills at home and school. In B.B.Schieffelin & E. Ochs (Eds.), Language socialization across cultures (pp. 97–124).Cambridge, England: Cambridge University Press.

Hickman, M. (2004). Coherence, cohesion and context. Some comparative perspectives innarrative development. In S. Strömqvist & L. Verhoeven (Eds.), Relating events innarrative. Typological and contextual perspective. London: Lawrence Erlbaum.

Hipfner-Boucher, K. (2011). A comparative study of ELL and EL1 narrative competenceduring the kindergarten years. Unpublished doctoral dissertation, University of Toronto,Toronto.

Hudson, J., & Nelson, K. (1983). Effects of script structure on children's story recall.Developmental Psychology 19, 625-635.

Hudson, J., & Shapiro, L. (1991). From knowing to telling: The development ofchildren’s scripts, stories and personal narratives. In A. McCabe & C. Peterson (Eds.),Developing narrative structure (pp. 89-132). Hillsdale, N.J.: Lawrence Erlbaum Assoc.

40

Justice, L. M., Bowles, R. P., Kaderavek, J. N., Ukrainetz, T. A., Eisenberg, S. L., & Gillam, R.B. (2006). The index of narrative microstructure: A clinical tool for analyzing school-age children’s narrative performances. American Journal of Speech-LanguagePathology, 15(2), 177–191.

Kaufman, D. (2001). Narrative development in Hebrew and English. In L.Verhoeven & S.Stromqvist (Eds.) Narrative Development in a Multilingual Context. Amsterdam: JohnBenjamins Publishing.

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categoricaldata. Biometrics, 33(1), 159-174.

Lanza, E. (2001). Temporality and language contact in narratives by children bilingual inNorwegian and English. In L.Verhoeven & S. Stromqvist (eds.), Narrative developmentin a multilingual context, pp. 15–50. Amsterdam: John Benjamins.

Limbos, M., & Geva, E. (2001). Accuracy of teacher assessments of second-languagestudents at risk for reading disability. Journal of Learning Disabilities, 34(2), 136-151.

Loban, W. (1976). Language Development: Kindergarten through Grade Twelve. Urbana, IL:National Council of Teachers of English.

Lucariello, J. & Rifki, A. (1986). Event Representations as the Basis for CategoricalKnowledge. In K. Nelson (Ed.), Event knowledge: Structure and function indevelopment (pp.189-203). Hillsdale - NJ, USA: Lawrence Erlbaum Associates.

Mandler, J. M. (1979). Categorical and schematic organization in memory. In C.Richard Puff(Eds.), Memory organization and structure (pp.259-296). London: Academic Press.

McCabe, A. & Peterson, C. (Eds) (1991). Developing narrative structure. Hillsdale, NJ:Erlbaum.

McCardle, P., Scarborough, H. S., & Catts. H. W. (2001). Predicting, explaining, andpreventing

reading difficulties. Learning Disabilities Research and Practice, 16, 230-239.

McCartney, K.A., & Nelson, K. (1981). Children’s use of scripts in story recall. DiscourseProcesses, 4, 59-70.

McClure, E. & Geva, E. (1983). The development of the cohesive use of adversativeconjunctions in discourse. Discourse Processes, 6, 411-432.

Mervis, C. B, & Crisafi, M. A. (1982). Order of acquisition of subordinate-, basic-, andsuperordinate level categories. Child Development, 53, 258-266.

Michael J. F., Margaret J. F., James N. F. (1993). Event knowledge and early languageAcquisition, Journal of Child Language, 20, 591- 606.

Myles-Worsley, M., Cromer, C. C., & Dodd, D. H. (1986). Children's preschool scriptreconstruction: Reliance on general knowledge as memory fades. DevelopmentalPsychology, 22, 22-30.

41

Nelson, K. (1978). How young children represent knowledge of their world in and out oflanguage. In R. Seigler (Ed.), Children’s thinking: What develops? Hillsdale, NJ:Erlbaum.

Nelson, K. (1986). Event knowledge: Structures and function in development. Hillsdale, NJ:Erlbaum.

Nelson, K. (1986). Event knowledge and cognitive development. In K. Nelson (Ed.), Eventknowledge: Structure and function in development (pp. 231-247). Hillsdale, NJ:Erlbaum.

Nelson, K., Gruendel, J. M. (1981), "Generalized event representations: Basic Building Blocksof Cognitive Development," in A. Brown and M. Lamb (Eds.), Advances inDevelopmental Psychology. Hillsdale, NJ: Lawrence Erlbaum Associates.

Nelson, K. (1985). Making sense: the acquisition of shared meaning. New York: AcademicPress.

Nelson, K., Fivush, R., Hudson, J., & Lucariello, J. (1983). Scripts and the development ofmemory. In M. T. H Chi (Ed.), Trends in Memory Development (pp.52-71), New York:Karger

Pearson, B. Z., & de Villiers, P. A. (2005) Child language acquisition: Discourse, narrative andpragmatics. In K.Brown and E. Lieven (eds.), Encyclopedia of language and linguistics,2nd edn, pp. 686–693. Oxford: Oxford University Press.

Pearson, B. Z. (2001). Language and mind in the stories of bilingual children. In L. Verhoeven& S. Lundquist (eds.), Narrative development in a multilingual context (pp. 373–398),Amsterdam: John Benjamins.

Peterson, C., & McCabe, A. (1991). Linking children’s connective use and narrativemacrostructure. In A. McCabe & C. Peterson (Eds.), Developing narrative structure (pp.29–53). Hillsdale, NJ: Erlbaum.

Price, D. W. W., & Goodman, G. S. (1990). Visiting the wizard: children’s memory fora recurring event. Child Development, 61, 664-680.

Ratner, H. H., Smith, B. S, & Dion, S. A. (1986). Development of memory for events,Journal of Experimental Child Psychology, 41, 411-428.

Richner, E. S. & Nicolopoulou, A. (2001). The narrative construction of differing conceptionsof the person in the development of young children’s social understanding. EarlyEducation and Development, 12, 393–432.

Rosch, E., Mervis, C. B., Gray, W., Johnson, D., & Boyes-Braem, P. (1976). Basic objects innatural categories. Cognitive Psychology, 8, 382-439.

Scarborough, H. S. (2001). Connecting early language and literacy to later reading (dis)abilities:Evidence, theory, and practice. In S. Neuman & D. Dickinson (Eds.), Handbook forresearch in early literacy (pp. 97-110). New York: Guilford.

42

Schank, R. P. & Abelson, R. C. (1977). Scripts, plans, goals, and understanding: an inquiryinto human knowledge structures. Hillsdale, NJ: Erlbaum.

Schwartz, M., & Shaul, Y. (2013). Narrative development among language-minority childrenthe role of bilingual versus monolingual education. Language, Culture and Curriculum,26(1), 36-51.

Simon-Cereijido, G., & Gutiérrez-Clellen, V. F. (2009). A crosslinguistic and bilingualevaluation of the interdependence between lexical and grammatical domains. AppliedPsycholinguistics, 30 (2), 315–337.

Slackman, E. A., Hudson, J. A., & Fivush, R. (1986). Actions, actors, links and goals: thestructure of children's event representations In K. Nelson (Ed.), Event Knowledge (pp.47-69). Hillsdale: LEA.

Snow, C. & Dickinson, D. (1990). Social sources of narrative skills at home and school.First Language, 10, 87-103.

Snow, C., Burns, S., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in youngchildren. Washington, DC: National Academy Press (NAP).

Tabors, P.O., Snow, C.E., & Dickinson, D.K. (2001). Homes and schools together: Supportinglanguage and literacy development. In D.K. Dickinson & P.O. Tabors, Beginningliteracy with language: Young children learning at home and school (pp. 313-334).Baltimore, MD: Brookes Publishing.

Wagner, R.K., Torgesen, J.K. & Rashotte, C.A. (1999). Comprehensive Test of PhonologicalProcessing. Austin, TX: PRO-ED.

43

Table 1. Descriptive statistics for participants by Age and Gender

The younger cohort (n=33) The older cohort (n=24)

Time 1 Time 2 Time 1 Time 2

Age M (SD) Range M (SD) Range M (SD) Range M (SD) Range

57.73(3.29) 51-63 69.97(3.80) 60-75 69.33(3.58) 64-76 82.08(3.87) 77-92

Gender Female Male Female Male

16 17 14 10

44

Table 2. Variables name and an example for each measure

Domain Variable name (type) (included language) Example/Description in English narrativesMacrostructure

Type of Output (Categorical) (E)Script

Script with partial episode:

Total episode:

“You invite people to your house or some place. You can bring agift or you can make something for a birthday boy or girl:“ You have a cake and Juice. When I turned 5, I had a bigbirthday party. There was Ariel cake”“My mom baked a cake for me and gave me a bracelet”

Slot fillers (Continuous) (E, H)ActorActionPropsTotal number of Slot fillers

Object Grouping (Continuous) (E, H)Superordinate class:Basic classSubordinate class

Mom, Friends, Cousin, Crown, Birthday boyEating, Singing, DancingGift, Cake, Juice, loot bag, Trampoline, Birthday hat

Food, Family, Activity, Sports, PeopleCake, Mommy, Daddy, Friends, BaseballChocolate cake, Kids Zone, Scavenger game, Piñata

Type of act sequences (Categorical) (E)A single actSimple act sequencesComplex act sequencesHierarchical act sequences

“You play games”“You sing a song and have a cake”“ After parties are over, your daddy come to pick you up”“Birthday parties can have themes. Some birthday parties aresports parties and some are music parties”

Microstructure

Productivity (Continuous) (E, H)Total number of morphemes (NM)Total number of C-Units (NCU)

Complexity (Continuous)Mean length of C-Units in morphemes (MLC-UM) (E, H)

Total number of complex C-Units (TCC-U) (E)

Total number of subordinate conjunctions (TNSC) (E)

Proportion of complex C-units (PCC-U) (E)

The total number of words in the child’s oral narrativeThe raw number of C-Units in the child’s oral narrative.A C-Unit is defined as an independent clause with its modifiers

The average length of C-Units in morphemes in the child’s oralnarrative.The total number of C-Units containing an independent clauseand at least one dependent clause in the child’s oral narrative.The raw frequency for use of subordinating conjunctions. (e.g.,since, when, while, because, as, although, though, because, where,that and so on)Calculated by dividing the number of complex C-Units by thetotal number of C-Units.

45

Table 3. Distributions and the McNemar test results for categorical variables in the L1 (English) for both cohorts

Variables The younger cohort (n=33) The older cohort (n=24)

Time 1 (%) Time 2 (%) p Time 1(%) Time 2 (%) p

Type of output

Scripts 25 (75.8) 28 (84.8).549

19 (79.2) 20 (83.3)1.00

Partial episode 8 (23.2) 5 (15.2) 5 (20.8) 4 (16.7)

Act sequences

Simple 23 (69.7) 6 (18.2)

.001

12 (50) 11 (45.8%)

1.00Complex 10 (30.3) 27 (81.8) 12 (50) 13 (54.2%)

46

Table 4. Descriptive statistics and the Wilcoxon or the Sign test Results for continuous variables in English for the younger cohort

Variables The younger cohort(n=33)

Time 1 Time 2 Significance Test

M(SD) Md Range M(SD) Md Range z pSlot fillers

ActorsActionsPropsTNSF

.6 (.93)4.40(2.15)4.88(2.72)9.97(4.50)

.004.004.009.00

0-30-110-281-25

1.12(1.47)5.88(2.87)6.30(3.35)

13.30(6.29)

1.005.005.0011.00

0-52-132-156-31

SSSS

.831.541.931.79

.405

.124

.054

.072

Object grouping

SuperordinateBasicSubordinate

.33(.65)3.36(2.46)1.58(1.12)

.003.001.00

0-30-120-4

1.39(1.12)3.67(2.00)2.03(2.01)

1.003.001.00

0-51-90-10

SSS

4.051.32.189

.000

.186.85

Productivity

TNC-UTNM

6.64(2.97)44.61(24.28)

6.0040.00

1-123-91

9.58(7.76)77.42(68.27)

8.0054.00

2-4922-369

WS

2.251.04

.02.296

Structural Complexity

TCC-UMLC-UMTNSCPCC-U

.55(.97)6.58(2.86).61(1.39).06(.11)

.006.110.000.00

0-32.33-17.00

0-6.000-.38

1.55(1.86)7.90(4.07)1.03(1.63).14(.14)

1.006.551.00.13

0-74-25.17

0-7.00-.50

WSSS

2.445.571.491.77

.01.000.134.078

Note: W = Wilcoxon test, S= Sign test, TNC-U= Total number of C-Units, TNM= Total number of morphemes, TCC-U= Total number of complex C-Units, PCC-U =

Proportion of complex C-Units, MLC-UM= Mean length of C-Units in morphemes, TNSC= Total number of subordinate conjunctions

47

Table 5. Descriptive statistics and the Wilcoxon or the Sign test Results for continuous variables in English for the older cohort

Note: W = Wilcoxon test, S= Sign test, TNC-U= Total number of C-Units, TNM= Total number of morphemes, TCC-U= Total number of complex C-Units, PCC-U =

Proportion of complex C-Units, MLC-UM= Mean length of C-Units in morphemes, TNSC= Total number of subordinate conjunctions

Variables The older cohort(n=24)

Time 1 Time 2 Significance Test

M(SD) Md Range M(SD) Md Range z pSlot fillers

ActorsActionsPropsTotal number of Slotfillers

.83(.70)6.17(1.99)6.5 (5.32)

13.50(6.47)

1.006.005.008.50

0-22-101-284-37

2(2.38)4.83(2.26)7.67(4.38)

14.50(6.04)

1.004.006.5013.00

0-102-101-195-29

SWSS

.24-2.181.75.83

.815

.018

.078

.405

Object grouping

SuperordinateBasicSubordinate

.83(1.09)4.54(3.24)1.83(2.14)

1.004.001.00

0-40-170-10

1.54(1.10)4.75(2.51)2.29(2.76)

1.005.001.00

0-41-90-11

SSS

2.07.83.00

.035

.4051.00

Productivity

TNC-UTNM

9.50(4.48)70.08(49.86)

8.5053.00

3-2418-256

9.29(4.97)86.92(52.40)

8.0075.50

3-2115-228

SS

-1.251.43

.210

.152

Structural Complexity

TCC-UMLC-UM

TNSCPCC-U

1.21(1.38)7.15(1.89)

.71(1.16).15(.17)

1.007.33

.00

.12

0-43.60-10.780-4

0-.57

1.92(1.80)9.20(3.85)

1.04(1.46).19(.19)

1.509.08

.00

.15

0-53-19

0-5.00-.67

SS

SS

.674.7

0.29.00

.503

.000

.7741.00

48

Table 6. Descriptive statistics and the Wilcoxon or the sign test Results for continuous variables in Hebrew for the younger cohort

VariablesThe Younger cohort

(n=33)Time 1 Time 2 Significance Test

M(SD) Md Range M(SD) Md Range z p

Slot fillersActorsActionsPropsTotal number of Slot fillers

.06(.24)

.03(.17)

.30(.47)

.39(.56)

.00

.00

.00

.00

0-10-10-10-2

2.67(3.29).79(.99).94(.90)

4.40(4.74)

2.00.001.004.00

0-170-30-40-23

SSSS

5.003.753.064.62

.000

.000

.001

.000

Object groupingSuperordinateBasicSubordinate

0.70(1.29)

.03

.00

.00

.00

0-00-60-1

.06(.24)1.82(2.66)

.03(.17)

.001.00.00

0-10-140-1

W 3.32 001

ProductivityTNC-UTNM

.36 (.82)2.76(7.74)

.00

.000-40-37

.58 (1.62)6.82(16.89)

.002.00

0-0-670-92

SS

.323.34

.75.000

Structural ComplexityTCC-UMLCUMTNSCPCCU

01.23(2.91)

00

.00

.00

.00

.00

0-00-13.50

0-00-0

.09 (.52)1.81(3.34)

0.05(.13)

.00

.00

.00

.00

0-30-10

00-.38

S 3.73 .000

Note: W = Wilcoxon test, S= Sign test, TNC-U= Total number of C-Units, TNM= Total number of morphemes, TCC-U= Total number of complex C-Units, PCC-U =

Proportion of complex C-Units, MLC-UM= Mean length of C-Units in morphemes, TNSC= Total number of subordinate conjunctions

49

Table 7. Descriptive statistics and the Wilcoxon or the sign test Results for continuous variables in Hebrew for the older cohort

Note: W = Wilcoxon test, S= Sign test, TNC-U= Total number of C-Units, TNM= Total number of morphemes, TCC-U= Total number of complex C-Units, PCC-U =

Proportion of complex C-Units, MLC-UM= Mean length of C-Units in morphemes, TNSC= Total number of subordinate conjunctions

Variables The Older Cohort(n=24)

Time 1 Time 2 Significance Test

M(SD) Md Range M(SD) Md Range z p

Slot fillers

ActorsActionsPropsTotal number of Slot fillers

.50(1.25).04(.20).42(.58).96(1.63)

.00

.00

.00

.00

0-40-10-20-5

3.88(3.35)1.17(1.79)1.63(1.28)6.67(5.95)

3.000.001.005.00

0-120-60-5

0-23

SSSS

3.932.603.253.62

.000

.006

.001

.000

Object grouping

SuperordinateBasicSubordinate

01.67(2.12)

0

0.001.000.00

0-00-70-0

.04(.20)2.86(2.80)

.08(.41)

.002.00.00

0-10-110-2

W 2.03 .043

Productivity

TNC-UTNM

.42(.50)2.79(3.48)

.001.50

0-10-12

.79(.98)6.71(8.14)

1.004.00

0-40-38

SS

.0001.75

1.00.078

Structural Complexity

TCC-UMLCUMTNSCPCCU

.002.38(3.60)

00

.00

.000-00-12

04.41(4.75)

00

.004.00

00

0-15 S 1.75 .078

50

Table 8. Intercorrelations among narrative measures in the L1 and L2 at Time 2 for the younger cohort

Variables 1 2 3 4 5 6 7 8 9 10 11 12

English measures1. Type of output2. TNSF .1113. Act sequences -.020 .2944. TNC-U .237 .570** .390*5. TNM .369* .683** .433* .777**6. TNCC-U .437* .410** .329 .665** .767**7. PCC-U .338 .234 .289 .373* .597** .906**8. MLC-UM .213 .383* .342 .191 .688** .559** .617**9. TNSC .301 .451** .460** .429* .544** .748** .705 .456**Hebrew measures10. TNSF -.139 -.056 -.067 -.073 -.200 -.002 -.051 -.258 .02911. TNC-U .155 .034 -.100 .101 .111 .144 .015 -.004 .158 .685**12. TNM .000 -.057 -.225 -.115 -.129 .118 .069 -.111 .105 .832** .713**13. MLC-UM .153 .071 -.104 .103 .115 .141 -.008 -.012 .149 .684** .988* 725**Note: TNSF= Total number of Slot fillers, TNC-U= Total number of C-Units, TNM= Total number of morphemes, TNCC-U= Total number of complex C-Units, PCC-U = Proportion of complex C-Units, MLC-UM= Mean length of C-Units in morphemes, TNSC= Total number of subordinate conjunctions. **p < .01, *p < .05

51

Table 9. Intercorrelations among narrative measures in the L1 and L2 at Time 2 for the older cohort

Variables 1 2 3 4 5 6 7 8 9 10 11 12

English measures1. Type of output2. TNSF .1393. Act sequences -.262 .488*4. TNC-U -.130 .631** .468*5. TNM .113 .721** .483* .734**6. TNCC-U -.116 .539** .645** .684** .759**7. PCC-U -.156 .397 .535** .409* .558** .901**8. MLC-UM .194 .347 .356 -.022 .558** .402 .487*9. TNSC -.177 .432* .575** .422* .507* .638** .624** .357Hebrew measures10. TNSF -.171 .159 .225 -.183 .012 -.079 .022 .192 .405*11. TNC-U -.302 -.018 .072 -.080 -.140 -.213 -.212 -.098 .134 .455*12. TNM -.284 .123 .134 -.029 -.031 -.070 -.035 -.044 .375 .759** .835**13. MLC-UM -.277 .170 .176 .080 .077 .020 .027 -.011 .319 .626** .888** .911**Note: TNSF= Total number of Slot fillers, TNC-U= Total number of C-Units, TNM= Total number of morphemes, TNCC-U= Total number of complex C-Units, PCC-U = Proportion of complex C-Units, MLC-UM= Mean length of C-Units in morphemes, TNSC= Total number of subordinate conjunctions. **p < .01, *p < .05