The development of coherence and cohesion in monolingual and sequential bilingual children’s narratives: Same or different?

This is a contribution from Narrative Inquiry 24:1© 2014. John Benjamins Publishing Company

This electronic file may not be altered in any way.The author(s) of this article is/are permitted to use this PDF file to generate printed copies to be used by way of offprints, for their personal use only.Permission is granted by the publishers to post this file on a closed server which is accessible only to members (students and faculty) of the author’s/s’ institute. It is not permitted to post this PDF on the internet, or to share it on sites such as Mendeley, ResearchGate, Academia.edu. Please see our rights policy on https://benjamins.com/#authors/rightspolicyFor any other use of this material prior written permission should be obtained from the publishers or through the Copyright Clearance Center (for USA: www.copyright.com). Please contact [email protected] or consult our website: www.benjamins.com

John Benjamins Publishing Company

http://www.copyright.com

mailto:[email protected]

http://www.benjamins.com

Narrative Inquiry 24:1 (2014), 40–76. doi 10.1075/ni.24.1.03kupissn 1387–6740 / e-issn 1569–9935 © John Benjamins Publishing Company

The development of coherence and cohesion in monolingual and sequential bilingual children’s narrativesSame or different?*

Judy R. Kupersmitta, Rachel Yifatb, and Shoshana Blum Kulkac

aHadassah Academic College and Al-Qasemi Academic College / bHaifa University / cThe Hebrew University Jerusalem

Studies of monolingual narrative production have revealed interacting paths in the development of coherence and cohesion across languages but less is known about bilingual narratives, where a gap may exist between socio-cognitive and linguistic abilities. The present longitudinal study explores the relations between measures of coherence and cohesion in the picture-based oral narratives elicited from 23 sequential bilinguals at two times — 2nd and 4th years of exposure to Hebrew as L2 (ages 6 and 8, respectively), compared to those produced by age-matched Hebrew speaking monolinguals. Measures of coherence included refer-ence to story components, and to four types of causal relations: psychological, motivational, enabling and physical. These analyses served as a basis to explore cohesion in terms of (1) inter-clausal connectivity, and (2) the linguistic encod-ing of the causal chain, which in this context demanded reference to a complex motion event. We found that reference to narrative components and causal relations improved with age in both L1 and L2, but were largely delayed among bilinguals at age 6, particularly regarding the most complex scenes. While coher-ence measures reached to a parallel level among the 8 year-old children, mea-sures of cohesion showed a different path of development in L1 and L2. Thus, the constraints imposed by language use in organizing the discourse resulted in a poorer connectivity between the clauses, and in less accurate lexico-grammatical encoding of the events in the bilingual narratives. The study underscores the mutual attraction between local and global principles of narrative construction,

Requests for further information should be directed to Judy R. Kupersmitt, Department of Communication Disorders, Hadassah Academic College Jerusalem, 37 Ha-Nevi'im St., Jerusalem, 91010. E-mail: [email protected]

mailto:[email protected]

© 2014. John Benjamins Publishing CompanyAll rights reserved

The development of coherence and cohesion in children’s narratives 41

which may become dissociated in a bilingual situation, and pinpoints to vulner-able domains of L2 discourse-embedded acquisition.

Keywords: narrative coherence, causal relations, cohesive devices, L2 Hebrew, sequential bilingual children, motion

Introduction

The analysis of narratives as a basic form of extended discourse offers an ample view of the linguistic, cognitive and socio-cultural abilities of children at various stages in development, in both typical and atypical populations (Berman, 2009; Reilly, Losh, Bellugi, & Wulfeck, 2004). Through narratives, children learn par-ticular ways to talk about events beyond the sentence level, first, in a conversa-tional context with supportive input from an adult or a peer partner (Eisenberg, 1985), and later, as an autonomous, de-contextualized text produced with little verbal support, showing increasing structural and linguistic complexity towards school-age and beyond (Berman & Nir-Sagiv, 2007; Berman & Slobin, 1994). As a semi-naturalistic arena to study language changes beyond the sentence level, psycholinguistic research on narrative abilities has focused on coherence and co-hesion in producing and comprehending discourse (Hickmann, 2004; McCabe & Peterson, 1991). Coherence is widely referred to in the literature as a kind of ‘conceptual connectivity’, achieved through logical relations such as causality and knowledge about how events, actions, objects and situations are interrelated and organized (De Beaugrande, 1980; Trabasso, Suh, & Payton, 1995). Various stud-ies that focused on coherence examined the development of story-grammars as evidence of the cognitive schemata underlying the representation of events which are believed to guide the generation of well-formed stories (Mandler, 1978; Stein & Glenn, 1979). These studies have shown that stories become more complex with age in terms of number of narrative components and in their hierarchical orga-nization around a goal plan, paving the understanding of the cognitive under-pinnings of narrative development, yet leaving unanswered questions about using language under the constraints of text production. In this sense, researchers have claimed that coherence is the result of both global and local processes of discourse structuring, where global semantic representations or macro-structures (based on cognitive information processing and organization) interact with local representa-tions or micro-structures that activate lexical and grammatical systems to ensure the production of coherent texts (van Dijk, 1977; Givón, 1995). These systems are activated according to the speakers’ ability to manage and regulate the flow of in-formation (Hickmann, 2003) through use of linguistic devices such as reference,


42 Judy R. Kupersmitt, Rachel Yifat, and Shoshana Blum Kulka

substitution, ellipsis, conjunction and lexical organization referred to as ‘cohesive devices’ (Halliday & Hassan, 1976).

Studies in narrative development have adopted various approaches that ac-count for the interrelations between coherence and cohesion (Berman & Slobin, 1994; Hickmann, 2004; Peterson & McCabe, 1991), as formulated by Hickmann (2004) who claimed that “above and beyond the content of particular narrative units, their ordering into a canonical story structure partially reflects discourse principles and is linked to uses of particular linguistic devices” (p. 290). In a more narrow sense, this idea postulates the mutual attraction between global, cognitive macro-structures and local, language driven principles of discourse organization and their interaction in the process of story production. Form-function studies of narrative development in monolingual populations have looked at the connec-tions between linguistic forms (e.g., lexical and morpho-syntactic devices) and their narrative functions such as establishing temporal relations in the narrative (Aksu-Koç & von Stutterheim, 1994; Hickmann, 2003), packaging the narrative content in syntactically and semantically related clauses (Berman, 1988; Berman & Slobin, 1994; Jisa, 1984), managing the referential links throughout the narrative (Bamberg, 1987; Hickmann, 2003; Karmiloff-Smith, 1979) or expressing evalua-tive content (Bamberg & Damrad-Frye, 1991; Bamberg & Reilly, 1996; Peterson & McCabe, 1983). Most of these studies reported interacting paths in producing coherent and cohesive narratives, where children’s cognitive schemata drives the selection of particular language forms and, equally, forms are selected as to per-form new functions as discourse structuring abilities develop.

More recently, the interplay between coherence and cohesion in discourse construction has been the focus of a few studies with simultaneous bilinguals who have acquired two languages from birth. Most of these studies have used the Frog Story picture book to investigate the relationship between command of linguistic structures, narrative organization and connectivity (e.g., Aarssen, 2001 with Turkish-Dutch bilinguals; Akinçi, Jisa, & Kern, 2001 with Turkish-French bilinguals; Kupersmitt, 2004 with Spanish-Hebrew bilinguals; Minami, 2008 with Japanese-English bilinguals), the expression of evaluative content such as false be-lief and character’s thoughts (e.g., Pearson, 2001 with Spanish-English bilinguals), or the management of referential links (e.g., Chen & Pan, 2009 with Chinese-English bilinguals). Overall, these studies revealed similar patterns of structural organization and of discourse-related principles of language use in monolin-gual and bilingual narratives, which seem to be guided by universal, language-independent strategies (Berman, 2001). For instance, children mentioned more narrative components with age, or, they learned to use adequate pronouns and ellipsis for character maintenance regardless of the linguistic status (L1 versus L2). Nonetheless, bilinguals often produced less sophisticated morpho-syntactic



constructions and used fewer language specific resources (e.g., adverbials, aspec-tual markings, temporal connectors) to express various narrative functions, hint-ing at a disadvantage of narrative-embedded language use rather than sentence-level language proficiency.

Compared to the numerous studies of simultaneous bilinguals, studies of nar-rative development in sequential bilinguals are considerably scarce. Some of these studies focused on assessing narrative performance in relation to language skills in L1 and L2 with school-age children who had certain experience with both lan-guages from early childhood, though some of them showed little proficiency in at least one of the languages (Gutiérrez-Clellen, 2002; Silliman, Bahr, Brea, Hnath-Chisolm, & Mahecha, 2002). Only a few studies concerned the development of coherence and cohesion in the L2 narratives of sequential bilinguals from a form-function perspective. Viberg (1994; 2001) studied the development of narrative structure and of various expressive devices (e.g., clausal connectors, verbs, adverbs and tense forms) to map forms with functions in the narratives of Finnish speaking children from preschool to later school-age learning Swedish as L2, as compared to L1 Swedish speakers matched for age. Taking three younger Spanish-English bilinguals with different proficiency levels in both languages, Montanari (2004) examined the development of their narrative competence in terms of overall struc-turing, evaluation, use of temporal perspective and referential expressions. Both studies support the existence of an underlying language-independent cognitive ability to organize the events in the narrative into a goal-oriented framework that increases with age. Nonetheless, they emphasize that a certain linguistic thresh-old is required to elaborate on certain crucial narrative components which seem more affected by linguistic proficiency. In the domain of linguistic expression, the studies revealed particular ways of encoding narrative functions in L2, which are different from those observed in L1 at the same ages. This seems to affect narra-tive construction at both local and global levels depending on the forms analyzed and on the level of achieved language proficiency, going from an impoverished language style with neutralized lexico-grammatical contrasts to inappropriate uses of cohesive devices that might result in unclear narratives. Given the entrenched relation between narrative abilities and the development of complex ways of lin-guistic expression as well as of language academic skills (Paris & Paris, 2003; Snow & Dickinson, 1991), it seems critical to allocate narratives at the focus of language development in sequential bilinguals.

Thus, the present longitudinal study investigates the narratives produced by se-quential bilinguals, learners of Hebrew (L2 narratives), as compared to those pro-duced by age-matched Hebrew speaking monolinguals (L1 narratives). Three aims motivate the study: (a) to understand the ability of the children to represent the epi-sodic structure of the narrative; (b) to evaluate the types of causal relations expressed



throughout the episodic categories; and (c) to reveal the particular linguistic forms selected by narrators to encode the events in the story and relate them in specific ways. These aims are considered along two axes: the development of both L1 and L2 narratives with age, and their comparison at each age. This will allow to identifying the strengths and the challenges en route to the acquisition of discourse embedded language skills in L2, from an integrative perspective of coherence and cohesion.

On the grounds of the studies mentioned above, we expect to find an increas-ing ability to integrate the events in the narratives into an episodic structure in L1 and L2, with more causal links expressed with age, both between and within the episodes (Benazzo, 2004; Trabasso & Rodkin, 1994). From a closer examination of the linguistic structures used, it is expected that these become more specific with age in both groups, to serve the purposes of episodic representation and con-nectivity. However, we assume that the constraints imposed by language use in organizing the discourse will be more demanding for language learners, where a significant gap may exist between already established — though still developing — socio-cognitive abilities and their incipient linguistic skills in L2.

Method

Subjects

The subjects for this study were 23 sequential bilinguals (16 girls and 7 boys) and 41 Hebrew-speaking monolinguals aged 4 to 9. The bilingual children were drawn from a group of immigrant children entering preschool or kindergarten in Israel for the first time, interviewed at their 2nd and 4th year of exposure to Hebrew, ren-dering two groups of 6 and 8 years of age. Criteria for inclusion in the study were as follows: (a) nearly exclusive exposure to their L1’s and no significant exposure to Hebrew as L2 until entry to preschool or kindergarten; (b) no obvious language impairment detected in L1; (c) non-verbal performance within the normal range as determined by the WIPPSI Animal Pegs subtest (all participating children had standard scores > 9). The children came from different countries and spoke dif-ferent languages including Amharic, Russian, French, Korean, American English and German. Parental level of education ranged from a few years of schooling to higher education and parents’ professions included housewives, teachers, com-puter engineers, farmers and students. The monolingual children were recruited from the urban area of Jerusalem. All the children were typically developing and belonged to middle to middle-high socio-economic backgrounds. Demographic information of both study and control groups at the different production times is presented in Table 1.



Table 1. Demographic information of childrenHebrew — L2 Hebrew — L1 (control group)

Year of exposure 2nd 4th NR NRAge group 6 8 6 8N 20 17 24 18Mean age (in months) 74 98.18 69.62 98.11SD 7.38 8.12 5.52 8.84Gender 13 (F); 7 (M) 12 (F); 5 (M) 11 (F); 13 (M) 11 (F); 7 (M)

Procedure and materials

Narratives were elicited from bilingual and monolingual children using a picture sequence (Goralnik, 1995) depicting a story about an intruding bee that flies into the kitchen to interrupt a family at mealtime and their sleeping cat who tries to get rid of the bee, with various mishaps following this event (see Appendix II for the complete description of the story content). Picture sequences have proved effec-tive to examine the development of narrative abilities (Bamberg, 1987; Berman & Slobin, 1994; Hickmann, 1995; Vion & Colas, 2005), since narrators have first to interpret the events as temporally and causally connected and recruit the required linguistic forms to reconstruct the events in a narrative frame. Although this may sound more demanding than producing spontaneous stories, picture sequences provide a common content to compare the productions from both cognitive and linguistic perspectives. In order to elicit the narratives, the children looked first through the pictures (arranged one by one in different pages) to become familiar with the story, and after they have finished, they were asked to tell the story to the experimenter while looking at the pictures. Once children began their sto-ries, the examiner provided only neutral listening and continuing prompts (such as “Uh-huh”, “Yeah”, “Really?”). When the child had finished or paused for more than a few seconds, the examiner asked “Is that the end?” The children’s stories were tape-recorded and transcribed according to the CHAT format of the Child Language Data Exchange System (CHILDES; MacWhinney, 2000). A sample of 40% of the narratives coded by two additional raters exceeded 90% agreement for all the measures analyzed.

The analytical approach

The approach undertaken in this study underscores the need to examine language forms in relation to their functions in discourse (Berman & Slobin, 1994). For this purpose, we proposed an analysis that reflects a continuum from coherence



to cohesion, starting with the macro-level of story coherence, going top-down to explore the causal relations between the narrative components, and reaching to a micro-level that explores the particular language structures selected by narrators to depict the narrative events and relate between them on their way to achieve a coherent whole.

Coherence: Episodic structure and causal relations

In the course of narrating a picture-based story, the narrator needs to interpret the visual static data as a series of events and states interconnected in tempo-ral and causal terms. This is constrained by the sequence of pictures on the one hand, which provides the perceptual cues, and by knowledge of intentional action, which drives the narrator to infer about the interconnections between the data, eventually represented through language and organized into an episodic structure. Following the model proposed by Stein and Glenn (1979), a set of components triggered by goal-directed action were defined, including a setting [S] where the characters and their actions are introduced with optional mention of place and time; an initiating event [E] which triggers the episode; an attempt [A] to solve the problem; followed by an outcome [O] which leads or does not lead to other actions. The original model includes two other (optional) components: internal responses [IR] triggered by the events (further considered in the analysis of causal relations) and internal plans or goals, which is merged in this analysis with the attempt to compose a ‘purposeful attempt’ rendering goal and action together (Trabasso & Nickels, 1992). According to the causal network approach (Trabasso, van den Broek, & Suh, 1989), the unfolding of connected components of the epi-sode constitutes a GAO unit (goal-attempt-outcome) (Trabasso & Nickels, 1992) connected by a series of inferred causal relations. Thus, goals motivate actions or attempts, attempts can physically cause an outcome or enable other actions or states which may subsequently result in another outcome. Outcomes, similarly to events, can physically cause other events or bring about cognitive or emotional internal responses. As stated, these relations are implied by the categories and they can remain implicit or be explicitly encoded.

In the proposed analysis, we consider four types of causal relations between the core episodic components: motivational, enabling, physical and psychological (Trabasso et al., 1989; Trabasso & Nickels, 1992), starting from the initiating event until the final outcome. For this purpose, we focused on the relations between clauses or spans of clauses, in case those were assigned the same category, for ex-ample, an attempt or an internal response expressed through a number of clauses (Trabasso, 2005). In analyzing the clauses and their causal relations, focus is on the point of view of the main protagonist (the cat) and the secondary characters



acting upon it (the mother and her children), ignoring the perspective of the bee, as detailed below following their representation in Figure 1.

– A motivational relation between the goal generated from the Initiating event leading to Attempt 1 [Mot-1], and from Outcome 2 leading to Attempt 2 [Mot-2].

– An enabling relation between Attempt 1 and Outcome 2, where mention of the attempt generates the necessary condition [Enb-1] which subsequently enables the Outcome [Enb-2].

– A physical relation between the events that comprise Outcome 2 [Phys-1], and between Attempt 2 and Outcome 3 [Phys-2].

– A psychological relation between the Internal Responses (emotional, behav-ioral, and cognitive) optionally triggered by the core episodic events (Initiating event, Attempt 1, Outcome 2, Attempt 2) leading to the goals in some cases [Psych-1 to Psych-4, respectively]

Figure 1 below depicts the episodic representation of the story and the causal rela-tions between the episodic components (see also Appendix II). Unlike canonical multi-episodic stories where a recurrent link exists between the protagonist’s goal and the attempts creating a series of subordinate goal plans of actions (such as, for instance, in the widely studied Frog Story), in this short but complex story the two episodes are independent in the sense that no link exists between the first goal and the final outcome, and thus not many causal relations are stemming from the first goal (Trabasso et al., 1989).

The components were identified and coded only when explicitly mentioned, avoiding influence of possible inferences. By explicit mention we mean the use of any linguistic resource (including lexical alternatives in L2) used to describe the entities and the actions which referred to the specific structural components. For a causal relation to be identified as such it had to be easily inferable from the linguis-tic context. In some cases, a relation could be inferred between two components even if it remained unmarked, but in other, the simple mention of the components was not enough to infer the relation, unless the information was arranged in par-ticular linguistic ways.

Cohesion: Inter-clausal connectivity

The choice of inter-clausal connectors provides a window to how narrators ar-range the flow of information and to their capacity of expressing different kinds of event-relationships at the local and global levels of the narrative (Berman, 1988; Berman & Slobin, 1994). While these links shape the organization of the text at the ‘surface’ level and are not mandatory in text production (McCutchen & Perfetti,



1982) they clearly contribute to discourse coherence, among other cohesive devic-es (Halliday & Hassan, 1976). Based on a number of studies on the development of inter-clausal connectivity (Berman, 1996; Berman & Nir-Sagiv, 2009; Berman & Slobin, 1994), the present analysis adopts the clause as the minimal unit of lin-guistic analysis where a clause refers to ‘any unit that contains a predicate which expresses a single situation, i.e. an activity, an event or a state’ (Berman & Slobin, 1994, pp. 660–663), exploring the types of clauses encompassed in three main groups, in order to capture the strategies of inter-clausal connectivity:

Linear: a single and independent main clause (a) with no overt syntactic con-nection with the surrounding clauses [MC]; or (b) headed by an overt marker of temporal sequentiality (e.g., axar-kax ‘afterwards’, axrey-ze ‘after that’) or by the conjunction ve ‘and’ either alone or followed by az ‘then’, where the function is not necessarily syntactic or semantic but rather marking the linear progression of the narrative [MC-OM] (Berman, 1996; Jisa, 1987).Coordinated: a clause linked to the preceding one by means of the conjunction ve ‘and’, o ‘or’, or aval ‘but’, both clauses being semantically compatible within a range of semantic relations, such as sequence, cause, location, or contrast (Halliday & Hassan, 1976). Three sub-types of coordinated clauses were considered: (a) co-ordinated clause with a different subject from the anterior main clause [CO-DS]; (b) coordinated clause with same subject [CO-SS]; and (c) coordinated clause with subject elision [CO-SE]. Coordination is early acquired in the context of the single sentence, but it is not until age 5 that children use it as a connectivity strat-egy in narratives, since it requires a certain degree of planning and organization (Berman, 1988; 2009). The analysis thus poses particular attention on the uses of and — as one of the most prevalent and early acquired connectors in various dis-

Bee

Cat

Family(Mother, Boy & Girl)

G Am e

IRIE

IE

psy

psy psy

psyG

IR IR

IR

G

m

m

e

e

e

e

psypsypsy

psy

psy

psy

m phiO

O

O

AIR

2 3 4 5 6

G OA

A

(+)

(0)1

(-)2a

O(-)

2 3

2b

1

(-)

EE

Figure 1. Causal network of the story showing the episodic components and the causal re-lations between them, in pictures 2 to 6 (causal relations analyzed in this study are in bold)



course contexts including narratives (Jisa, 1987; Berman, 1996; Berman & Slobin, 1994; Peterson & McCabe, 1991).Subordinated: a clause that is connected to another to form a tighter ‘package’ that renders a unified idea (Berman & Slobin, 1994), including: (a) finite and non-finite complement clauses [COMP] as well as complement clauses in the form of direct speech [COMP-DIR]; and (b) finite and non-finite relative [REL] and adverbial clauses [ADV].

Cohesion: Linguistic structures in encoding the episodes

In this particular story, the main first episodic string comprises a series of motion events varying in complexity of ground, path and motion (Talmy, 1991; 2000). Encoding the first attempt, for instance, demands that narrators refer to a complex path with a goal en-route (the bee) and an endpoint goal (the table) where the first is connected to the initiating event and presupposes a motivational causation and the second relates to the subsequent events and presupposes an enabling causation (see Appendix II). Focus is on how children manage to express intentional action and motion together, monitoring the selection of particular linguistic structures that contribute both to encoding the structural components and the inference of causal relations. Consider, for instance, the difference in encoding the locative tra-jectory in a text by a Hebrew speaking adult (taken from a corpus collected and analyzed by the first author) and an 8 year old girl in L1.

(1) ha-xatul hitorer, ra’a et ha-dvora ve kafac le-kivun ha-dvora heysher le-shulxan-ha-oxel. ‘The cat woke-up, saw the bee and jumped towards the bee straight onto the dinning-table.’ [Adult]

(2) ve axrey ze ha-xatul radaf axrey ha-dvora ve kulam nivhalu ve az ha-xatul kafac al ha-oxel shel ha-yeladim. ‘And afterwards the cat chased after the bee and everyone got-scared and then the cat jumped on-(to) the children’s meal.’ [L1, 8, TAL]

In (1) a compact single-clause construction is used with the verb kafac ‘jumped’ followed by two prepositional phrases to indicate the complete trajectory — goal in path and endpoint. In contrast, the child in (2) encodes a split trajectory using two verbs radaf ‘chased’ and kafac ‘jumped’, the first referring to the goal in path (the bee) and the second to the endpoint (the table), but these are not specific enough to integrate the whole path, rendering a loose causal relation between the events.

For the analysis, we adopted the notion of event conflation, referring to “the dis-tribution of the information across the verb and its associated elements in the clause” (Aksu-Koç, 1994, p. 345) encompassing a series of motion events from the initiating



event up to the first outcome. The target verbs are dynamic and represent prototypi-cal motion events consisting of a figure moving through a path with respect to an-other reference object or ground, varying along certain features attributed to motion verbs, such as manner, path, agency, or volition (Hickmann, 2003). As Hebrew be-haves more like a verb-framed language (Slobin, 2004; Talmy, 2000), focus is on the lexicalized expression of motion in the verb, but considering also the elaboration of the event by means of prepositional phrases and other elements beyond the clause.

Results

As the basis for a comparative analysis of the narrative measures and in order to examine the amount of language production in a narrative context, the stories were analyzed by number of clauses, presented in Table 2.

Table 2. Number of clauses in L1 and L2 narratives by ageL1 L2

Age 6yrs 8yrs 6yrs 8yrsRange 6–29 8–25 6–62 8–39Mean (SD) 15.08 (5.05) 16.89 (4.96) 17.95 (12.59) 18.8 (9.07)

Unexpectedly, there were no significant differences in number of clauses by lan-guage background or age, partly explained by the wide range in story length within the groups, as shown in the table. The L2 group shows larger standard deviations (particularly at age 6), which probably point to a higher inter-group variability among language learners. The next sections present the results on the measures of coherence and cohesion as described above.

Episodic structure and causal relations

To evaluate the children’s ability to build an episodic representation of the story (see Figure 1), we calculated first the proportion of subjects mentioning the main structural components in the L1 and L2 stories (excluding IR as an optional com-ponent), as presented in Table 3.

As expected, taking L1 and L2 narratives together results show a significant increase in the number of episodic components mentioned with age. The compo-nents that increased more markedly were the setting from .29 to .69 (χ2 = 11.924, df = 1, p < .01), the attempt 1 from .50 to .86 (χ2 = 11.053, df = 1, p < 0.01), and the outcome 1 (Bee) from .20 to .54 (χ2 = 9.751, df = 1, p < .05), while the initiating event increased moderately though significantly from .86 to 1 (χ2 = 5.165, df = 1, p < .05).



The remaining components did not change with age, indicating that certain com-ponents may be more representative of a developmental change than others. For example, reference to the bee or fly leaving the kitchen or escaping — mentioned by around a half of the 8 year-olds (e.g., ve ha-dvora afa la me-ever la-xalon ‘and the bee flew itself [ = flew away] through the window’ [L2, 8, DAS]) is indicative of an ability to sustain a hierarchical goal plan and its subsequent outcome, without being distracted by the events in between. In contrast, the second attempt was at ceiling probably due to the prototypical, script-like features of the scene showing a mother and their children washing their cat (Nelson, 1986; Shank & Abelson, 1977).

From a comparative perspective between L1 and L2, we calculated the effect of language background on the frequency of each structural component by age. Results of 2x2 chi-square tests showed significantly less mention of the initiating event (χ2 = 4.020, df = 1, p < .05), the attempt 1 (χ2 = 5.867, df = 1, p < .05), and the outcome 2 (χ2 = 4.768, df = 1, p < .05) in the L2 narratives at age 6, but differences between L1 and L2 disappeared at age 8 for all the structural components.

The next aim was to evaluate the types of causal relations expressed between the episodic components in L1 and L2. Figure 2 presents the proportion of sub-jects that expressed each of the causal relations analyzed across the narrative (ren-dering a total seven) in L1 and L2.

In developmental terms, our predictions were partly confirmed. Psychological causation at four different sites throughout the episodes (see Figure 1) was most prevalent overall, showing no difference by age or language background. More than .80 of the narratives in all groups contained at least one reference to psycho-logical causation related to a previous event or state. In order to examine whether this overwhelming tendency to express psychological causation was reflected in the content of the narratives, we compared the proportion of clauses coded as internal reactions as compared to clauses that represented events (including at-tempts and outcomes, even if expressed as outcome states such as hu haya naqi ‘he was clean’), starting from the initiating event. We found that regardless of language background or age, psychological reactions reached to a quarter of total clauses in all groups without exception — with a vast majority of emotional and behavioral responses, while the other three-quarters were event clauses. This tendency is il-lustrated in the following (translated) examples from two 6 year-old children in L1 and L2, respectively, with psychological reactions marked in bold.

(3) a. L1, 6, DAN ‘[1] And then the boy shouted [2] and got-scared [3] because (there)

came (a) fly. [4] And the cat jumped [5] and the girl shouted [6] and the boy shouted too. [7] Then the soup spilt, [8] the cat was sad [9] and the mother said “oy va’avoy”! [10] And then (the) mother quickly, quickly



washed the cat in the bathtub [11] because he was dirty [12] and she didn’t want [13] that he stain [=didn’t want him to stain].’

b. L2, 6, DIO ‘[1] And then (there) was (a) fly [2] and *cat [MM, cf. the cat] got-up [3]

and (the) mother still washed (the) dishes. [4] And to-the boy (it) was not nice [=the boy didn’t like it] [5] because it [=the fly] *was bother [SYN, cf. was bothering] him, [6] and then to-the cat (it) was also not nice [WO, cf. also was not nice] [=the cat didn’t like it either] [7] and then he jumped straight on(to) the fly, [8] and then everyone said “oy”! [9] and then (the) cat jumped on(to) the dishes [10] and the fly flew up [11] and the girl laughed [12] that the cat was sad [13] and (the) mother was very sad. [14] And then (they) washed the cat [14] and to-the cat (it) was not nice at all [=he didn’t like it at all] [15] that (they) wash him.’

Table 3. Proportion of subjects mentioning each component by age, in L1 and L2 stories

Group

Story components

Age 6 Age 8L1(N=24)

L2(N=20)

L1&L2(N=44)

L1(N=18)

L2(N=17)

L1&L2(N=35)

Setting .29 .30 .29 .61 .76 .69Initiating event .96 .75 .86 1 1 1Attempt 1 .67 .25 .50 .94 .76 .86Outcome 1 Bee .17 .25 .20 .61 .47 .54Outcome 2 Cat .92 .65 .79 .89 .71 .80

a. Spilt .79 .60 .69 .83 .70 .76b. Dirty .42 .35 .38 .22 .47 .34

Attempt 2 1 1 1 1 1 1Outcome 3 .50 .56 .56 .65 .82 .69Note: For the purpose of statistical analyses, a combined measure of Outcome 2 ‘a’ and ‘b’ is presented.

0

0.2

0.4

0.6

0.8

1

1.2

Age 6 Age 8 Age 6 Age 8

L1 L2

MOT-1

MOT-2

ENB-1

ENB-2

PHYS-1

PHYS-2

PSYCH

Figure 2. Proportion of children producing inferable causal relations by age in L1 and L2



In both examples, the main story events are either preceded or followed by psy-chological reactions, most of them emotional (clauses 2, 8 in L1, and clauses 4, 6, 12, 13, 14 in L2) or behavioral (clauses 1, 5, 6, 9 in L1 and clauses 8, 11 in L2), expressed through emotion verbs or adjectives and by use of direct speech, with a single example of a cognitive reaction in the form of an explanation expressed in clause [5] in the L2 excerpt.

Motivational causation increased with age, showing interesting interactions by age and language background. Thus, MOT-1 significantly increased in L1 (χ2 = 4.714, df = 1, p < .005) and in L2 (χ2 = 7.943, df = 1, p < .005), but MOT-2 increased in L2 only (χ2 = 10.117, df = 1, p < .005). Both types were significantly less frequent in L2 than in L1 narratives at age 6 (χ2 = 5.867, df = 1, p < .05 for MOT-1; χ2 = 7.134, df = 1, p < .05 for MOT-2). MOT-1 involves an internal motivation, which may delay its ex-pression through language, particularly among the younger children, while MOT-2 is triggered by a physical condition, leading to a more familiar, script-like event.

Enabling relations were the less frequent overall. ENB-1 significantly in-creased with age from .29 and .05 at age 6 to .61 and .35 at age 8 in L1 (c2 = 4.286, df = 1, p < .05) and L2 (c2 = 5.498, df = 1, p < .05), respectively. ENB-2, however, significantly increased in L2 from .05 to .35 (c2 = 5.498, df = 1, p < .05) but not in L1 where it remained stable at around .50. This portrays a significantly different pattern of development in L1 and L2 at age 6, similarly to motivational relations (c2 = 4.283, df = 1, p < .05 for ENB-1 and c2 = 10.612, df = 1, p < .05 for ENB-2).

Unexpectedly, physical relations were low in all the groups, and did not change by age or language background. PHYS-1 occurred in around .20 of the narratives between the two events that constitute the outcome 2, which represent two dif-ferent perspectives toward the events depicted in the pictures: the cat ‘spilling the soup’ is a perceived event while the cat ‘getting dirty’ is inferred (see Appendix II). Going back to Table 3, this difference is clearly reflected in the relatively frequent reference to the first event — around three quarters in all groups — as compared to around one third referring to the cat getting dirty. The physical relation was identified only when the two events were mentioned. The second physical relation PHYS-2 — ‘washing the cat’ caused its state of ‘being clean’ — occurred in only .50 of the narratives, since many of the children referred to a psychological rather than a physical outcome as shown in example 3(b) above.

The next two sections focus on the particular linguistic configurations selected by narrators and their contribution to the inference of causal relations in local and global terms of discourse organization. Particular emphasis is given to the expression of motivational and enabling relations, as these show more inter-group variability, starting with an analysis of inter-clausal connectivity subsequently fol-lowed by a qualitative analysis of the linguistic encoding of the ‘motion chain’ within the episodes.



Inter-clausal connectivity

Figure 3 displays the proportion of linear, coordinated and subordinated clauses by age and language background. Linear strategies of clause combining [MC and MC-OM] were the most frequent overall at both ages but significantly decreased with age from .72 to .57 (F(75,1) = 14.830; p < .01), particularly the use of overt markers of sequentiality used for chaining clauses.

Coordination of all types almost doubled with age from .17 to .28, with a larg-er gap in coordination with subject ellipsis (F(75,1) = 9.800; p < .05), reflecting the increasing ability of children to arrange the information in more compact con-structions maintaining the topic. Subordination, however, remained below .15 of total clauses at both ages (F(75,1) = 2.478; p = .120), in line with results of previous studies in Hebrew, reporting a sharp development in narrative-embedded use of complex syntax after age 9 (Berman, 2009; Berman & Nir-Sagiv, 2007; Berman & Neeman, 1994), an age that only a few of the older subjects have reached in the present study. As expected, this developmental trend in strategies of connectivity differed in L1 and L2, replicating the developmental effect. Thus, linear strategies were significantly favored among language learners reaching to .73 as compared to .57 of total clauses in L2 and L1, respectively (F(75,1) = 15.970; p < .01); coordina-tion, in contrast, was more than double in L1 narratives — .30 compared to .14 in L2 (F(75,1) = 20.607; p < .01), and subordination reached to only .12 in both groups (F(75,1) = .79; p = .780). Further analyses by age and language show that this prefer-ence for linear patterns of connectivity persisted in L2 even at age 8, where, for instance, linear strategies reached as high as .65 and coordination to only .19 of total clauses as compared to .49 and .36 in L1, respectively.

It was interesting to examine the relationship between causal relations and the strategies of connectivity selected by the children. Overall, clauses were less linear in the first episodic string, where diverse forms of coordination and to a lesser extent subordination emerge to encode motivational and psychological relations. This is particularly noted at age 8 in both L1 and L2, as illustrated in the excerpts below, starting at the initiating event and ending with the first attempt.

(4) a. L1, 8, ALM Axarkax ba zvuv ve Ø icben et ha-yeladim ba-zman she axlu. Axarkax

ha-xatul kafac ki hu ra’a et ha-zvuv ve nisa litfos Ø oto. ‘Afterwards (there) came (a) fly [LINEAR] and Ø (=it) bothered the

children [CO-SE] in-the time that [=while] (they) ate [=were eating] [ADV]. Afterwards the cat jumped [LINEAR] because he saw the fly [ADV] and Ø (=he) tried to-catch it [CO-SE].’



b. L2, 8, MAS Axarkax ba’a dvora ve Ø *hecika [COL, cf. hifri’a] la-yeled le’exol. Ve

ha-xatul ra’a et ze, Ø kafac al ha-shulxan ve Ø raca litfos et ha-dvora. ‘Afterwards (there) came a bee [LINEAR] and Ø (=it) *annoyed [COL,

cf. bothered] the boy [CO-SE] to-eat [ADV] [= while eating]. And the cat saw this [LINEAR], Ø jumped on-to the table [CO-SE] and Ø wanted to-catch the bee [CO-SE].’

The excerpts in (4) show an emerging complexity in causal packaging achieved through use of coordinated clauses with subject ellipsis which develop with age to serve other than additive functions in multi-clausal packages, as well as some use of subordination. Connectivity strategies developed in both form and func-tion. For instance, at age 8, causal subordination with ki ‘because’ served more global functions such as motivating the unfolding episode as in example (4a) (e.g., ‘afterwards the cat jumped because he saw the fly’). In contrast, younger narrators in both groups used this connector mostly for local purposes to con-nect internal states and emotions (e.g., ‘the cat got very angry because they both-ered him the nap’).

The expression of enabling relations within the episodic string starting right after the first attempt and ending with the second outcome portrayed a more lin-ear pattern of clause connectivity, as illustrated in these examples taken from the narratives by two 8 year-old children in L1 (a) and L2 (b).

(5) a. L1, 8, YOS axrey ze ha-xatul kofec al ha-dvora ve az kulam mistaklim ve az ha-xatul

kafac al ha-oxel ve ha-oxel nishpax ve ha-dvora halxa ve ha-ima hayta acuva ve ha-yalda caxaka.

‘Afterwards the cat jumps on-to the bee [LINEAR] and then everyone looks [=is looking] [LINEAR] and then the cat jumped on-to the food [LINEAR] and the food spilt out [CO-DS] and the bee went away [LINEAR] and the mother was sad [LINEAR] and the girl laughed [LINEAR].’

b. L2, 8, DAS ve ha-xatul kafac me ha-makom shelo el ha-zvuv ve ha-zvuv naxat al

ha-marak ve az ha-xatul naxat al ha-marak ve ha-dvora afa la me-ever la-xalon ve az kulam hayu acuvim.

‘And the cat jumped from his place toward the fly [LINEAR] and the fly landed on the soup [CO-DS] and then the cat landed on the soup [LINEAR] and the bee flew away through the window [LINEAR] and then everyone was sad [LINEAR].’



In both examples, all the clauses except one are linear, introduced by ve ‘and’ or ve az ‘and then’ with no clear semantic link other than continuing the narrative, even when connecting internal responses (e.g., ‘and then everyone was sad’) with the precipitating events. This is also manifested by the frequent change of topic, which creates a more discontinuous, picture-by-picture style of organization. These two strategies together hinder the inference of the enabling relations relevant to this scene (see Figure 1) which usually remain unclear if the clauses are not properly connected unless the relation is achieved by other means (as further shown in the section below). Interestingly, the occurrence of tense shifts from present to past in (5a) and the lexical shifts when referring to the same referent (e.g., ‘bee’ replacing ‘fly’) in (5b) are also indicative of more linear strategies in organizing the infor-mation (Berman, 1988; Berman & Slobin, 1994; Hickman, 2003). An emergent development toward more hierarchically connected clauses can be noticed in L1 starting already at age 6, as illustrated in the following examples from a narrative by a 6 year-old in L1 as compared to an 8 year-old in L2.

00.10.20.30.40.50.60.70.8

Linear SubordCoord

00.10.20.30.40.50.60.70.8

Linear SubordCoord

Age 6

Age 8

L1

L2

Figure 3. Proportion of connectivity strategies out of total clauses, by age and language backgroundNote: LINEAR = Main clauses [MC] and Main clauses with overt marker [MC-OM]; COORD = Coordinated clauses with same and different subject [CO-SS/DS] and coordinated clauses with subject ellipsis [CO-SE]; SUBORD = Finite and non-finite complement [COMP] as well as direct speech complement clauses [COMP-DIR], relative [REL] and adverbial clauses [ADV].



(6) L1, 6, IDA ve ha-xatul kafac litfos et ha-zvuv aval bimkom ze hu kafac al ha-shulxan ve

kol ha-marak nishpax alav. ‘And the cat jumped [LINEAR] to-catch the fly [ADV] but instead he

jumped on-to the table [CO-SS] and all the soup spilt-out on-him [CO-DS].’

(7) L2, 8, ANG ve pit’om ha-xatul qafac al ha-zvuv. ve ha-xatul be-ta’ut qafac al shulxan-ha-

oxel ve shapax [cf. shafax] et kol ha-oxel ve ha-marak. ‘And suddenly the cat jumped on-to the fly [LINEAR]. And the cat by

mistake jumped on-to the dinning-table [LINEAR] and Ø (=he) spilt all the food and the soup [CO-SE].’

These examples are similar in that both contain the necessary information to en-code the enabling relations. They also show a similar strategy in connecting the second enabling relation [ENB-2] by use of coordination with different subject in (6) and coordination with subject ellipsis in (7). However, the difference emerges in connecting the clauses that encode the attempt which enabled the condition ‘being on the table’. The younger L1 narrator skillfully used same subject coordi-nation with the marker ‘but’, reinforcing the semantic contrast using the adverb ‘instead’. In example (7), however, the adverb ‘by mistake’ hints a semantic connec-tion between the clauses but the causal relation remains questionable due to the linear connectivity and the repetition of the lexical subject ‘the cat’.

The use of linear strategies rendered a ‘listing’ of actions followed by internal states or responses, as shown in (8) from an 8-year-old narrative in L2.

(8) L2, 8, NAD axar kax hem asu lo ambatya. Ha-yeled maxzik oto, ima shotefet oto, ha-yalda

maxzika et ha-magevet ve axar-kax hem niku oto. Ve axar-kax talu lo min ka-ze…{shaxaxti. INT: “seret”} seret. Ve ha-yalda moxet kapayim ve ha-yeled meruce.

‘Then they make him a bath [=gave him a bath] [LINEAR]. The boy holds him [LINEAR], (the) mother washes him [LINEAR], the girl holds the towel [LINEAR] and then they cleaned him [LINEAR]. And afterwards (they) put on him a kind of… {I forgot. INT: ‘ribbon’} ribbon [LINEAR]. And the girl claps hands [LINEAR] and the boy is happy [LINEAR].’

Less frequently, the clauses were arranged in more compact and shorter packages that contained less topic shifts, as shown in the next example taken from a narra-tive by an 8 year-old in L1.



(9) L1, 8, MOR ima niqta et ha-xatul ve sama mayim karim al ha-kviot me-ha-marak. Axrey

she ha-xatul yaca me-ha-miklaxat kulam hayu merucim. ‘(the) mother cleaned the cat [LINEAR] and Ø [=she] put cold water on the

wounds from the soup [CO-SE]. After the cat got-out of the shower [ADV] they were all pleased [LINEAR].’

These examples above show that the same actions or states as encoded on the basis of pictorial stimuli can be organized as discrete units or as unified events which become integrated into hierarchical levels. This is the case in (9), where clauses are combined in tighter pairs with some kind of logical relation. Undoubtedly, there is more than clause connectivity contributing to the expression of causal relations in the flow of discourse organization.

In the following section we show a qualitative analysis of the linguistic struc-tures used to mention the core episodic components throughout the first sub-ep-isode.

Linguistic structures in encoding the episodes

The descriptive analyses presented in this section show the linguistic structures used to express the events that comprise the motion chain, focusing on the use of verbs and other related elements within and beyond the clause in order to reveal possible differences as well as similarities between L1 and L2 at both ages. Table 4 provides the breakdown of forms for each event, starting with the use of specific — ‘target’ verbs (indicating the motion features between parentheses) and followed by other verbal alternatives as well as by additional elements that elaborate on the event. The table presents the linguistic encoding of the events, independently of the linguistic criteria defined for considering the structural components (see Table 3).

Reference to the incoming bee to encode the initiating event that triggers the plot becomes more specific with age in L1 but not so in L2. Around a quarter of the children in L1 mentioned the event by means of a specific verb as nixnesa ‘entered/came in’ or ba’a ‘came’ (.66 at age 6 and .88 at age 8) as compared to only a half in L2 at both ages. The alternative forms used at age 6 in L1 and at both ages in L2 were copular verbs (e.g., ve haya zvuv she hecik la-yeled ‘and (there) was (a) fly that annoyed the boy’ [L1, 6, IDA]) or presenting the bee or fly from the perspective of the boy (e.g., ha-yeled ra’a dvora ‘the boy saw (a) bee’ [L2, 8, YEV]. Less con-ventional uses found among L2 speakers were direct speech (e.g., ha-yeled amar: hine dvora ‘the boy said: here’s (a) bee’ [L2, 8, DAS]), or use of semantically related but inappropriate motion verbs (e.g., *nigash [LEX, AGR, cf. nigsha] dvora ‘(there)



approached-MASC (a) bee-FEM’ [L2, 6, SHI]). Location was expressed in around .20 of the narratives (e.g., ve az ba zvuv ha-bayta ‘and then came (a) bee to-the-home’ [L1, 6, LIO]), remaining presupposed in most cases. A very first emergent expression of a spatial ground is evident at age 8 in L1 (e.g., ba zvuv she hitkarev el ha-marak ‘(there) came a fly that approached to-the soup’ [L1, 8, MAY]) serving as a spatial anchor that allows the inference of causal relations at a more global level of discourse organization.

Next in the motion chain comes the ‘waking up’ event, mentioned by around one third of the children except for the L2 group at age 8 where it reached .47, mostly by means of specific verbs such as kam or hit’orer ‘get/wake-up’ (e.g., ha-xatula kama ‘the cat got-up’ [L2, 6, SHU]), by occasional use of more literary style lexical expressions in L1 (e.g., ha-xatul amad al arba raglayim ‘the cat stood on (its) four legs’ [L1, 8, YOH]), or by much less conventional expressions in L2 (e.g., ha-xatul me’od ka’as ki hem hefri’u lo et ha-shena [COL, cf. hefri’u lo lishon] ‘the cat (got) very angry because they bothered him the-sleeping [COL, cf. bothered him to-sleep]’ [L2, 6, DAS]). Interestingly, more than half of the children at age 8 mentioned the cat sleeping as part of the setting component. The rather infrequent reference to the cat waking up as a response to the intruder bee may indicate the local organization of the contents, where each picture is still treated in isolation. It is worth comparing here the expression of the same content in the oral produc-tions of the Frog Story (Berman, 1988; Berman & Slobin, 1994) where almost none of the children mentioned the boy and the dog waking up. The authors concluded that this information was inferable from the previous context and by frequent ref-erence to the temporal anchoring ‘in the morning’. In the present story, the cat’s waking up is motivated by the initiating event, and thus more significant for story coherence than merely enabling the subsequent events.

Perhaps the most significant scene in the motion chain is the sequence of the cat’s jumping toward the bee and ending on the table. As stated, this scene de-picted in two consecutive pictures links the first and the second sub-episodes and is critical for story coherence. Table 4 shows the structures to encode this scene decomposing the complex path. The younger children in L2 are behind the rest in expressing the first phase of the motion (‘jumping towards the bee’), both in explicitly mentioning the action and in using a specific verb. Only half use the verb ‘jump’ while around a third use less specific verbs of motion (e.g., hu *raca [LEX, AGR, cf. rac] la-dvora ‘he ran-FEM to-the-bee’ [L2, 6, SIN]) or a non-motion verb (e.g., ve axal oto sof sof ‘and (he) ate it at the end’ [L2, 6, CHA]), which in many cas-es is irrelevant to the story. These less specific forms occur in L1 to a lesser extent, and in more sophisticated constructions (e.g., ve az ha-xatul ba litrof et ha-dvora ‘and then the cat comes to-devour the bee’ [L1, 6, YUV]). On the other extreme, 8-year-olds in L1 are almost at ceiling in use of specific verbs, showing emergent



use in around a third of their narratives of motion verbs with an additional inten-tional value (e.g., ha-xatul hitnapel al ha-dvora ‘the cat pounced on the bee’ [L1, 8, HAD]). Beyond the verbal nucleus, the L1 narratives show an advantage over those by language learners in mention of the goal (bee) in various structures, as either a PP in the same clause (around a quarter at both ages versus only 0.12 at age 8, respectively) or in a separate clause occurring twice more often in L1 than in L2. The former used more compact structures with non-finite adverbial clauses (e.g., ve ha-xatul kafac litfos et ha-zvuv ‘and the cat jumped to-catch the fly’ [L1, 6, IDA]) whereas language learners were confined to more linear strategies with the goal mentioned in a separate clause either before or after the motion event (e.g., axrey ze hu raca likfoc al ha-zvuv ve hu kafac ‘afterwards he wanted to jump on-(to) the bee and he jumped’ [L2, 8, DIO]). Among the 8 year-olds, only .22 in L1 and .12 in L2 mentioned both the goal in path and the endpoint, allowing a smoother link to the second episode (e.g., ha-xatul hitacben ve kafac al ha-shulxan litfos et ha-zvuv ‘the cat got-nervous and jumped on-(to) the table to-catch the fly’ [L2, 8, MAR]). At age 6, children occasionally mentioned the endpoint without refer-ence to the goal in path, hinting about the challenge they faced in L1 and L2 when referring to the whole locative trajectory to encode the attempt. This was reflected in the overwhelming absence of motion verbs indicating the change of location to an endpoint (e.g., ve az ha-xatul naxat al ha-marak ‘and then the cat landed on the soup’ [L2, 8, DAS]). Instead, around a quarter of the children (but only .10 in L2 at age 6) repeated the verb ‘jump’ either with or without a prepositional phrase to indicate the motion to the table or the food as an endpoint, while a few mentioned only the endpoint (e.g., axrey ze ha-xatul kafac aleha [=the bee] ve-az le-tox ha-calaxot ‘afterwards the cat jumped on-her [=the bee] and then into the dishes’ [L2, 8, YEV]). Only .22 of the children at age 8 in L1 and even less in L2 (.11) were able to provide a global spatial anchoring, as illustrated in the following excerpt from an 8 year-old in L1.

(10) ha-zvuv ba le’exol et ha-marak az ha-yeled giresh oto. Axar-kax ba ha-xatul ve raca likfoc al ha-zvuv. Ha-xatul kafac, aval ha-zvuv barax ve ha-xatul shafax et kol ha-marak.

‘The fly came to eat the soup so the boy got-rid of it. Afterwards came the cat and Ø wanted to jump on-(to) the fly. The cat jumped, but the fly escaped and the cat spilt all the soup’ [L1, 8, GIL].’

In (10), after having mentioned the soup as a spatial anchor, the narrator skill-fully creates a referential link throughout the narrative so that later reference to the cat ‘spilling the soup’ allows to inferring the enabling causal relation between the attempt and the outcome. It is possible that dealing with such a complex scene blurs the reference to the fate of the bee, and renders it as a secondary, non-salient



Table 4. Proportion of different verbs and their related elements used to express the mo-tion chain, by age, in L1 and L2.

Age 6 Age 8L1 (N=24) L2 (N=20) L1 (N=18) L2 (N=17)

BEE — enter, comeV Specific (+change of location; +deixis) .66 .50 .88 .53

Other — .05 — .06No motion* .29 .30 .11 .41

Location (PP) .21 .15 .22 .23Spatial ground (PP) .04 — .16 .06CAT — wake up, stand upV Specific (+change of posture) .25 .30 .27 .47

Other .16 .05 .05 — CAT — jumpV Specific (+manner) .71 .55 .61 .70

Specific (+intentionality) .08 — .33 .06General (e.g., go, come) .08 .15 .05 .06No motion .12 .20 .05 .18

Goal

In Path (PP) .25 — .22 .12In Path (Clause) .25 .05 .44 .23Endpoint (PP) .12 .10 — — Endpoint (PP) + Path (Clause) .04 .05 .22 .12

CAT — land, fallV Specific (+path, +endpoint) — — — .17

Other (+manner; -endpoint) .29 .10 .22 .29

End-point

PP/NP .25 .05 .33 .29Other (spatial reference, adverbial link) .12 — .22 .11

BEE — fly, go awayV Specific (+path, +change of location) .04 .05 .27 .12

Specific (+manner) .12 .20 .22 .29No motion .08 .05 .11 .05

Note: PP’s may occur within or beyond the clause (e.g., ba zvuv she hitkarev el ha-marak ‘(there) came a fly that approached to-the soup’), or as PP complements of copular verbs (e.g., ha-xatul haya al ha-shulxan ‘the cat was on the table’). Goal in path and endpoint were analyzed in relation to a motion verb only.* In her seminal study about the expression of motion and location in the narratives of English, German, Chinese and French speakers, Hickmann (2003) concludes that French speaking children tend to focus on static rather than dynamic predicates. In the present study, some of the French speaking children indeed used static verbs to refer to the entrance of the bee, but the sample was too small to perform a typologically based analysis of the selection of forms in L2. This is a challenging direction for future research dealing with young L2 learners.



event in the eyes of the young narrators. Thus, only around a quarter and half of the children in L1 and L2 respectively, refer to the bee leaving the kitchen either by a manner of motion verb (e.g., ve ha-dvora afa la me-ever la-xalon ‘and the bee flew itself [=flew away] through the window’ [L2, 8, DAS]), or by a path verb either specific (e.g., ve ha-zvuv barax ‘and the fly escaped’ [L1, 6, GAL]) or general (e.g., ve ha-zvuv halax lo ha-bayta ‘and the fly went itself [=went back] home’ [L2, 8, CHA]). Only two children in L1 referred to the event as a negative outcome from the cat’s perspective (e.g., ha-xatul kafac ve lo hicliax litfos et ha-zvuv ‘the cat jumped and (he) failed to catch the fly’ [L1, 8, GIL]), and two children referred to the bee as dead or scared.

Discussion

Storytelling practices of sequential bilinguals offer a unique platform to reconsider the relation between coherence and cohesion in discourse organization. The first is guided by more universal, top-down principles of discourse production while the latter is manifested in the ability to recruit an ample repertoire of language forms in a bottom-up fashion in order to connect and regulate the flow of information across the discourse. In this study, the interaction between both domains across development allowed for a more comprehensive examination of the children’s narrative embedded language skills, following a psycholinguistic functional ap-proach (Berman & Slobin, 1994; Hickmann, 2003; 2004; Karmiloff-Smith, 1979). The study thus contributed to understanding the effects of language experience on the patterns of narrative development in both domains of coherence and cohesion.

Going on a continuum from coherence to cohesion, our first aim was relat-ed to the most rudimentary level of structuring the narrative, by looking at how the static, visual stimuli throughout the pictures were “translated” into the verbal structural components of the story (Berman, 1995). We found that narrative com-ponents increased with age in L1 and L2, confirming the developmental results reported elsewhere for monolinguals and bilinguals (Berman, 1988; Berman & Slobin, 1994; Peterson & McCabe, 1983). However, 6-year-olds at their second year of exposure to Hebrew as L2 lagged behind monolinguals in mention of some of the narrative components, reaching to a similar level only at age 8. This clearly points to the critical effect of language experience, meaning that global-level nar-rative skills are not enough to guide the production of coherent stories but hav-ing reached a minimal linguistic threshold seems a vital requisite (Viberg, 2001). Yet, since the linguistic criteria defined in this study for coding the components demanded explicit reference to the events that represent the episode, probably, the gap between L1 and L2 narratives at age 6 would be reduced adopting more flexible



linguistic criteria. Interestingly, while certain components were delayed even at age 8 other components were close to or even at ceiling in both groups across age. For instance, results showed no difference between L1 and L2 narratives in setting the scene or in the components throughout the second episode which represent-ed more familiar, daily routines, though significant differences were found in the first episodic string. This corroborates that the strategies undertaken by younger children when organizing the narrative largely depend on their knowledge of the events and the language used for representing them (Nelson, 1986), and on the complexity of story structure, as reported in previous studies of monolinguals and bilinguals (Akinçi, Jisa, & Kern, 2001; Berman, 1995; Pearson, 2001).

The story used in this study is quite complex, since it lacks a unified episodic structure in the sense that the chain of events in the first sub-episode is not fully related to the chain of events in the second sub-episode, as shown in Figure 1 above. Besides, the final resolution is not at all related to the initiating event, un-like in stories with more canonical structures. This complexity was challenging even for some of the children at age 8 in both groups, who failed to achieve a fully-fledged story structure. Undoubtedly, this has a language cost, too, particularly for second language learners. Complex scenes may demand more efforts to recruit the appropriate linguistic forms resulting in an impoverished account of the events. Thus, even if the children could provide a basic description of the events depicted across the pictures, interpreting these events as being part of a unified plot was not always possible, particularly at those instances where more subtle but still complex relations were involved between them. This suggests that coherence relies on a more abstract level of representation, where the units of the story schema (or story grammar) become integrated or “chunked” in causal chains leading to an inter-connected episodic structure (Trabasso & Rodkin, 1994). Our second aim was to tap this level of story coherence.

Similarly to the findings for story components, causal relations tended to increase with age in L1 and L2 as reported in previous studies of monolingual narratives (Benazzo, 2004; Trabasso & Nickels, 1992), but there is no single path of development for the four types of relations explored here. Psychological rela-tions between internal reactions and their precipitating events seemed more resis-tant to the effects of language background or age, as corroborated by their quite pervasive expression in L1 and L2 at both ages. Together, these findings are not in line with previous studies on the development of evaluation in picture-based narratives (as a partially similar category to psychological reactions, see footnote 4), where the overall proportion of evaluative devices was reported to be scarce at these developmental stages (Bamberg & Damrad-Frye, 1991; Bavin, 2009; Nakamura, 2009; Pearson, 2001). We believe that perceptual saliency represent-ed in facial expressions and body gestures throughout the pictures triggered the



consistent expression of psychological causation, where narrators mostly referred to the emotional reactions of the protagonists. In their study about the develop-ment of evaluative devices in the ‘frog stories’ of English speaking children and adults, Bamberg and Damrad-Frye (1991) claimed that “it seems to be the facial expression in agreement with the immediately precipitating event that motivates the evaluative device” (p. 701). This seems the case in the present study, too, and it may explain why this type of causation is easily inferable even when more rudi-mentary linguistic forms such as exclamations, direct speech and common emo-tional verbs and adjectives (e.g., ka’as ‘got-angry’ or hayu acuvim ‘were sad’) were used by L2 children.

This finding has interesting implications for the role of emotions in second language development, echoing previous claims about the importance of under-standing and expressing emotion in a first language (Tomasello, 2000). In the con-text of narrative discourse, identifying emotions seems a first step in constructing the meaning of the story and may be regarded as a primitive type of causal relation which brings about more complex causal links between the events in the story later in development (Makdissi & Bosclair, 2006; Trabasso & Stein, 1997). The findings also point to the difference between the relatively early reference to feelings and emotions through language, as compared to more complex forms of evaluations such as reference to causality or cognitive states of mind. The first seem indepen-dent of language proficiency but are rather driven by the particular features of the narrative — or in some cases by individual preferences (Montanari, 2004), while the latter, emerging by age 9, are highly influenced by the development of cogni-tive abilities such as integrating the event in a more hierarchical schema, as well as of language-specific forms.

The expression of physical relations was stable across groups. Generally, when-ever physical causation was involved, children mostly referred to the cause which is both necessary and sufficient to infer the outcome (Trabasso et al., 1989). In this story, the cause involving the first of the two physical relations was perceivable in the pictorial stimuli while the outcome had to be inferred, what could further explain the narrators’ choice. The two remaining types of relations — motivational and enabling — showed developmental as well as L1-L2 differences. Motivational causation can be triggered either by an internal or an external cause. It was the first type that increased in monolingual and bilingual narratives, being it marked explicitly in a separate clause or implicitly stated as the goal of the action (i.e. a ‘purposeful attempt’) by means of a prepositional phrase (e.g., jumped at the bee). This is not to say that goals should be explicitly marked in order to infer the causal relation. Trabasso and Nickels (1992), for instance, claimed that pur-poses are not always informative and about half of the adults tend to omit them in their narratives, in case there is sufficient information to infer them. In our



stories, explicit mention of motivational causation was more likely to be triggered by an internal response following the initiating event rather than by a physical outcome. Motivational causation was tightly related to enablement. If the children missed the intentional nature of the action they generally omitted the goal and the enabling relation between the attempt and the unsuccessful outcome could not be inferred. This explains why enablement was delayed as compared to other causal relations, especially in the L2 narratives. In this story, expression of the enabling relations served as a link between the first and second sub-episodes, and story coherence was compromised when this link was missing. Stories differ from each other regarding the types of causal relations that need to be mentioned to achieve coherence. For instance, the enabling relation in the well-known Frog Story — where ‘waking up’ enables the main protagonists to discover that the frog had escaped — is not critical for story coherence, since this relationship can be easily inferred from world knowledge. Previous studies of story comprehension have found that enablement is a weaker relation as compared to others (Trabasso et al., 1989; Tapiero, van den Broek, & Quintana, 2002), a claim that is worth examin-ing in the development of narrative production. From these findings, one may suggest a gradual progression from locally to more globally driven expression of causal relations in L1 and L2, although some types of causality relations may be predominantly data-driven and thus more reliant on accessible language forms (Karmiloff-Smith, 1985; Nakamura, 2009).

The third aim of this study was to explore the children’s selection of language forms to encode the story components and re-construct the relations between them so as to establish causal relations. Taking only part of the children’s nar-rations (excluding the setting), the analysis considered the ability of the narra-tors to use inter-clausal strategies for connecting between the events they narrate. Overall, results showed a preference for marking linear connectivity using ‘and’, ‘then’, or ‘and then’ at both ages in both groups, with little subordination overall, replicating the findings in previous studies of monolingual narratives using pic-tures (Berman, 1988; Berman & Slobin, 1994; McCabe & Peterson, 1991; Vion & Colas, 2005). With age, the information was more tightly packaged presuppos-ing the narrators’ increasing ability to engage in top-down processes of discourse production in both L1 and L2 (Karmiloff-Smith, 1985), which becomes evident as the events are organized on a macro-level independently of perceptual contiguity (see Berman & Slobin, 1994, pp. 543–544). Quantitative and qualitative analyses of the use of ve ‘and’ reveal that while native speakers show an increasing ability to “narrow” its use to create more cohesive links already from age 6 (Jisa, 1984), it takes a longer time for language learners to quit the pattern of sequential con-nectivity, even when the causal relation is incompatible with the default sequential interpretation. In general, the frequency of ‘and’ in discourse and the range of



semantic as well as pragmatic functions it performs in different linguistic con-texts prompts its overuse as the most accessible unmarked form for connecting propositions (Schiffrin, 1986). Studies of monolingual narratives revealed that the functions of ‘and’ change with development. It firstly serves a chaining function at the local level of the isolated pictures (even in those cases where the semantic relation is not temporal per se) gradually acquiring more global functions to con-nect denser packages of clauses in temporal or causal relations of different types (Berman, 1996; Berman & Nir-Sagiv, 2009; Vion & Colas, 2004). Studies about the development of narrative abilities in L2 have reported on the ‘vulnerability’ of inter-clausal connectivity, showing simplification of native-like patterns, use of less complex syntactic constructions or most basic uses of a form (Akinçi, Jisa, & Kern, 2001; Kupersmitt, 2004; Kupersmitt & Berman, 2001) as well as an over-representation of sequential connectors, even after four years of exposure to the language (Viberg, 2001). The present study showed that these trends were more accentuated in the more complex episodic strings, where flexibility of linguistic expression was required to render discrete content units into cohesive linguistic configurations. One such example was the ability to connect the motivational rela-tion in the first episode with the attempt performed by the cat as the main pro-tagonist. Most of the children in L2 could express these content units as discrete information in two separate simple clauses: ha-xatul raca litfos et ha-dvora ‘the cat wanted to catch the bee’ followed by ha-xatul kafac ‘the cat jumped’, but less could portray a unified representation of the scene using either coordination (e.g., hu raca litfos et ha-dvora ve hu kafac aleha ‘he wanted to catch the bee and he jumped on it’) or subordination (e.g., hu kafac kedey litfos et ha-dvora ‘he jumped in order to catch the bee’).

The development of cohesion was further addressed by looking at the use of verbs and other forms inside and outside the clause for expressing the ‘motion chain’ that constituted the backbone of the episodes. Results showed an emergent phase in the ability of children in L1 and L2 to both mention and elaborate on motion events in specific ways so as to infer the causal links within the episode, as found in previous studies about the expression of motion in narratives (Guo & Chen, 2009; Hickmann, 2003; Özçalişkan, 2009). Language learners tended to en-code less information about paths, thus obscuring the integration of the events into an episode such as for example, when referring to the incoming bee to encode the initiating event by means of the static predicate hayta ‘(there) was-FEM’ instead of using the more specific nixnesa ‘entered-FEM’ or ba’a ‘came-FEM’. Besides, lan-guage learners seemed particularly disadvantaged when narrating more complex scenes which demanded reference to multiple perspectives such as motion and cause together. The low accessibility to the linguistic resources in L2 may distract the youngest narrators from the main goal of the narrative, but rather confine their



description to a series of actions which motivation remains unclear (Trabasso & Nickels, 1992). ‘Low accessibility’ does not necessarily mean lack of vocabulary, since, for instance, most of the children could mention the bee at the onset of the episode but not as the goal in the path of the cat’s motion.

Developmentally, the study uncovered a non-obvious difficulty of children in L1 and L2 to elaborate on motion events with complex paths where a figure moves through toward a goal en-route and an endpoint goal, compromising the inference of enabling relations in the story. Similar findings were reported in previous studies that examined the development of expressive options to refer to motion and space in narratives. Berman and Slobin (1994), for example, claimed that young English-speaking children aged 3 to 5 often failed to relate both source and endpoint of a movement within a conceptual frame and even in successive clauses, and mostly opted to refer to one single facet of the movement. In a thorough account of her studies about motion expression in picture-based narratives, Hickmann (2003) reports a progression with age in the type of information encoded by the verb to express motion, path and causality across the scenes in the narrative, depending on language typology. When referring to the relation between cause and motion, she claims that younger French-speaking children aged 4–5 tend not to explicitly mention the causal links neither through the verbs nor through other causative constructions, as compared to speakers of English or German. The present study, inspired on the vast research and the careful methodologies adopted in the stud-ies mentioned above, underscores the entrenched relation between motion and the causal relations that stem from the particular language forms selected by the narrators. Taking form and function together the analysis contributed to a better understanding of the children’s performance in terms of universal as compared to specific language knowledge that guide discourse construction, and in terms of the interaction between structural or linguistic abilities that are driven by top-down principles of discourse organization (Karmiloff-Smith, 1979). To give a fur-ther example, it can be definitely assumed that 8-year-old monolingual children know the verb nafal ‘fall’. However, almost none of the children used this verb to indicate the end of the path and the displacement of the cat to another location which was crucial for advancing the plot. Thus, more emphasis should be put on the constraints imposed by discourse production in expressing motion and on the functions served by the forms. This is particularly relevant when looking at paths, which elaboration may serve to encode the narrative components and to integrate them into a coherent episodic structure, as different from encoding manner which is an optional feature of motion (Slobin, 2004). In relation to spatial anchoring, this study suggest that at this stage in development, children in all groups are more centered in the animate protagonists and rather ignore the inanimate entities that serve as spatial anchors around which the protagonists move. Hickmann (2003)



suggests an increasing ability of children to plan the discourse so as to provide “initial spatial anchors for the subsequent interpretation of locations and chang-es of location throughout the narrative” (p. 278). This is an interesting point to consider, since narratives seem to differ in the relevance of spatial anchoring for achieving coherence. Thus, the present study paves the way to understanding how children conceptualize motion events as serving the purposes of unfolding the plot and connecting between events, beyond their ability to describe the motion in terms of its features distributed in a range of lexical or syntactic ways.

To conclude, the study has shown the developmental progression of coherence and cohesion in bilingual narratives as two domains that are mutually dependent in discourse construction. The method adopted in the study allowed to carefully tracking the narrators’ abilities from both top-down and bottom-up perspectives, rendering a reliable picture of the entrenchment between the language forms and their functions en route to developing narrative embedded language skills in L2. The comparison to the monolingual data was important to disentangle the mani-fold factors affecting the attainment of narrative abilities and to understand the multifaceted path to L2 development. While the study was designed with three goals in mind, the analyses we performed on the data draw our attention to the different psycholinguistic and communicative factors that may influence discourse production. Certainly, within a semi-structured, non-experimental task, the nar-rative as a product of the task provided not only the measurable data as a function of the controlled variables of age and language background but also served as a window to raise hypotheses on the children’s abilities as a function of language knowledge, story complexity and the degree of the cognitive load imposed by the task, which are not always possible to control a priori. More research is needed to be performed on larger populations of sequential bilinguals, controlling contextual as well as typological factors so as to expand our knowledge of the role of language proficiency versus universal, cognitive driven abilities in discourse production.

Given the significance of narrative abilities in attaining decontextualized lan-guage and literacy skills, the present study has launched an innovative methodolo-gy to situate discourse-embedded skills at the core of research on L2 development in children. This has clinical as well as educational implications. Clinically, the study should contribute to a better understanding of the linguistic performance of sequential bilinguals at the level of discourse, adding to previous path-breaking studies on the domains of morpho-syntax (see, for example, Paradis, 2005, and Paradis, Rice, Crago, & Marquis, 2008 among others). This may ensure a more comprehensive view of the learner language which is critical for the assessment of possible language delays in bilingual populations (Kohnert, 2010). In educational terms, the study underscores the need to implement programs that enhance active participation in discourse practices in L2 at schools, in order to equip the children



with the range of expressive and communicative skills required to successfully perform in academic tasks, while encouraging parents and educators to sustain L1 language and literacy skills at home.

Notes

* This paper is dedicated to Prof. Shoshana Blum-Kulka, who passed away in June, 2013. We owe Shoshana for her wisdom, her immense knowledge and her unique personality.

1. The present study is part of a wider longitudinal study aimed to follow the linguistic and socio-cultural experience of sequential bilingual children from their 1st entry to Hebrew speak-ing educational institutions, at four consecutive years. The study has been conducted by Prof. Shoshana Blum-Kulka at the Hebrew University Jerusalem and her research team.

2. On the basis of language samples collected by a bilingual research assistant who spoke the child’s L1. The assistant examined the child’s expressive vocabulary, the conversational abilities and the narrative abilities asking the child to retell the same story used in this study as well as a personal story in her L1.

3. The setting of the story was disregarded from the present analyses despite that it is indicative of the ability of children to provide the adequate background for the interpretation of the plot (Berman, 2001).

4, Internal responses are emotional or affective reactions to events or explanations to their causes and effects and thus may be considered a subset of the much broader and controver-sial notion of ‘evaluation’ first introduced by Labov and Waletzky (1967). Evaluative content has been defined as that part of the narrative that suspends the temporal line of the events and focuses on emphasizing the meaning of the narrative, by means of particular semantic and structural strategies. Developmental studies on the expression of evaluation in narratives have focused on specific categories such as frames of mind (emotion, cognitive and affective terms), character speech, causal connectors, or intensifiers (Bamberg & Damrad-Frye, 1991; Nakamura, 2009; Peterson & McCabe, 1983).

5. In Hebrew, subject ellipsis is optional in coordination with same subject predicates and in certain subordinated contexts (Berman, 1990). Coordination with same-subject ellipsis has been reported as a favored strategy to tightly pack two or more events together in discourse (Berman, 1988; Nir & Berman, 2010).

6. All the examples are classified according to language background (L1 or L2), age (6 or 8) and abbreviated name of subject. Two examples of children at similar ages in L1 and L2 are classified as (a) and (b), respectively. The examples are translated verbatim with reformulated expressions provided in square brackets when necessary, followed by the sign = . In the sections that deal with cohesion, examples are presented in the original Hebrew version, followed by the translated version.



7. Sentence level errors are marked with an asterisk when possible. Types of errors are classified between square brackets followed by the correct form (cf.) both in the original and the trans-lated version. For a list of errors with their definitions and examples see Appendix I.

8. The choice of the most specific verbs as the ‘default’ verb to encode the subsequent motion events was based on a sample of more than 40 narratives told by native Hebrew speaking adults, collected and analyzed by the first author.

References

Aarssen, J. (2001). Narrative development in a multilingual context. In L. Verhoeven & S. Strömqvist (Eds.), Development of temporal relations in narratives by Turkish-Dutch bilin-gual children (pp. 209–233). Amsterdam: John Benjamins.

Akinçi, M., Jisa, H., & Kern, S. (2001). Influence of L1 Turkish on L2 French narratives. In L. Verhoeven & S. Strömqvist (Eds.), Narrative development in a multilingual context (pp. 189–208). Amsterdam: John Benjamins. DOI: 10.1075/sibil.23.07aki

Aksu-Koç, A., & von Stutterheim, C. (1994). Temporal relations in narrative: Simultaneity. In R. A. Berman & D. I. Slobin (Eds.), Relating events in narrative: A crosslinguistic developmental study (pp. 393–455). Hillsdale, NJ: Lawrence Erlbaum.

Aksu-Koҫ, A. (1994). Development of linguistic forms: Turkish. In R. A. Berman & D. I. Slobin (Eds.), Relating events in narrative: A crosslinguistic developmental study (pp. 329–388). Hillsdale, NJ: Lawrence Erlbaum.

Bamberg, M. (1987). The acquisition of narratives: Learning to use language. Berlin: Mouton de Gruyter. DOI: 10.1515/9783110854190

Bamberg, M., & Damrad-Frye, R. (1991). On the ability to provide evaluative comments: Further explorations of children’s narrative competencies. Journal of Child Language, 18(3), 689–710. DOI: 10.1017/S0305000900011314

Bamberg, M., & Reilly, J. (1996). Emotion, narrative, and affect: How children discover the re-lationship between what to say and how to say it. In D. Slobin, J. Gerhardt, A. Kyratzis & J. Guo (Eds.), Social interaction, social context, and language (pp. 329–342). Mahwah, NJ: Lawrence Erlbaum.

Bavin, E. (2009). Plot and evaluation: Walpiri children’s frog stories. In J. Guo, E. Lieven, S. Ervin-Tripp, N. Budwig, S. Özçalişkan & K. Nakamura (Eds.), Crosslinguistic approach-es to the psychology of language: Research in the tradition of Dan I. Slobin (pp. 137–148). Mahwah, NJ: Lawrence Erlbaum.

Benazzo, S. (2004). L’expression de la causalité dans le discours narrative en Français L1 et L2. Langages, 3, 35–51.

Berman, R. A. (1988). On the ability to relate events in narrative. Discourse Processes, 11, 468–497. DOI: 10.1080/01638538809544714

Berman, R. A. (1990). On acquiring an (S)VO language: Subjectless sentences in children’s Hebrew. Linguistics, 28(6), 1135–1166. DOI: 10.1515/ling.1990.28.6.1135

Berman, R. A. (1995). Narrative competence and storytelling performance: How children tell stories in different contexts. Journal of Narrative and Life History, 5(4), 285–313.

Berman, R. A. (1996). Form and function in developing narrative abilities. In D. I. Slobin, J. Gerhardt, A. Kyratzis & J. Gus (Eds.), Social interaction, social context and language (pp. 343). Mahwah, NJ: Lawrence Erlbaum.

http://dx.doi.org/10.1075/sibil.23.07aki

http://dx.doi.org/10.1515/9783110854190

http://dx.doi.org/10.1017/S0305000900011314

http://dx.doi.org/10.1080/01638538809544714

http://dx.doi.org/10.1515/ling.1990.28.6.1135



Berman, R. A. (2001). Setting the narrative scene: How children begin to tell a story. In A. Aksu-Koç, C. Johnson & K. Nelson (Eds.), Children’s language (vol. 10, pp. 1–31). Mahwah, NJ: Lawrence Erlbaum.

Berman, R. A. (2009). Language development in narrative contexts. In E. L. Bavin (Ed.), The Cambridge handbook of child language (pp. 355–376). New York: Cambridge University Press. DOI: 10.1017/CBO9780511576164.020

Berman, R. A., & Neeman, Y. (1994). Development of linguistic forms: Hebrew. In R. A. Berman & D. I. Slobin (Eds.), Relating events in narrative: A crosslinguistic developmental study (pp. 285–328). Hillsdale, NJ: Lawrence Erlbaum.

Berman, R. A., & Nir-Sagiv, B. (2007). Comparing narrative and expository text construction across adolescence: A developmental paradox. Discourse Processes, 43(2), 79–120. DOI: 10.1080/01638530709336894

Berman, R. A., & Nir-Sagiv, B. (2009). Clause-packaging in narratives: A crosslinguistic de-velopmental study. In J. Guo, E. Lieven, S. Ervin-Tripp, N. Budwig, S. Özçalişkan & K. Nakamura (Eds.), Crosslinguistic approaches to the psychology of language: Research in the tradition of Dan I. Slobin (pp. 149–162). Mahwah, NJ: Lawrence Erlbaum.

Berman, R. A., & Slobin, D. I. (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum.

Chen, L., & Pan, N. (2009). Development of English referring expressions in the narratives of Chinese–English bilinguals. Bilingualism: Language and Cognition, 12(4), 429–445. DOI: 10.1017/S1366728909990216

De Beaugrande, R. (1980). Text, discourse and process. Norwood, NJ: Ablex.Eisenberg, A. R. (1985). Learning to describe past experiences in conversation. Discourse

Processes, 8(2), 177–204. DOI: 10.1080/01638538509544613Givón, T. (1995). Functionalism and grammar. Amsterdam: John Benjamins. DOI: 10.1075/z.74Goralnik, E. (1995). Language screening test for Hebrew-speaking preschool children. Netanya,

Israel: Gai Agencies.Guo, J., & Chen, L. (2009). Learning to express motion in narratives by Mandarin speaking chil-

dren. In J. Guo, E. Lieven, S. Ervin-Tripp, N. Budwig, S. Özçalişkan & K. Nakamura (Eds.), Crosslinguistic approaches to the psychology of language: Research in the tradition of Dan I. Slobin (pp. 193–208). Mahwah, NJ: Lawrence Erlbaum.

Gutiérrez-Clellen, V. F. (2002). Narratives in two languages: Assessing performance of bilingual children. Linguistics and Education, 13(2), 175–197. DOI: 10.1016/S0898-5898(01)00061-4

Halliday, M. A. K., & Hasan, R. (1976). Cohesion in spoken and written English. London: Longmans.

Hickmann, M. (1995). Discourse organization and the development of reference to person, space and time. In P. Fletcher & B. MacWhinney (Eds.), The handbook of child language (pp. 194–218). Oxford: Blackwell.

Hickmann, M. (2003). Children’s discourse: Person, space and time across languages. Cambridge: Cambridge University Press.

Hickmann, M. (2004). Coherence, cohesion, and context: Some comparative perspectives in narrative development. In S. Strömqvist & L. Verhoeven (Eds.), Relating events in narrative: Typological and contextual perspectives (pp. 281–306). Mahwah, NJ: Lawrence Erlbaum.

Jisa, H. (1984). French preschoolers’ use of et pis (“and then”). First Language, 5, 169–184. DOI: 10.1177/014272378400501501

Jisa, H. (1987). Sentence connectors in French children’s monologue performance. Journal of Pragmatics, 11(5), 607–621. DOI: 10.1016/0378-2166(87)90182-2

http://dx.doi.org/10.1017/CBO9780511576164.020

http://dx.doi.org/10.1080/01638530709336894

http://dx.doi.org/10.1017/S1366728909990216

http://dx.doi.org/10.1080/01638538509544613

http://dx.doi.org/10.1075/z.74

http://dx.doi.org/10.1016/S0898-5898(01)00061-4

http://dx.doi.org/10.1177/014272378400501501

http://dx.doi.org/10.1016/0378-2166(87)90182-2



Karmiloff-Smith, A. (1979). A functional approach to child language: A study of determiners and reference. Cambridge: Cambridge University Press.

Karmiloff-Smith, A. (1985). Language and cognitive processes from a developmental perspec-tive. Language and Cognitive Processes, 1(1), 61–85. DOI: 10.1080/01690968508402071

Kohnert, K. (2010). Bilingual children with primary language impairment: Issues, evidence and implications for clinical actions. Journal of Communication Disorders, 43, 456–473. DOI: 10.1016/j.jcomdis.2010.02.002

Kupersmitt, J. (2004). Form-function relations in Spanish narratives: A comparative study of bilingual and monolingual speakers. In S. Strömqvist & L. Verhoeven (Eds.), Relating events in narrative: Typological and contextual perspectives (pp. 395–434). Mahwah, NJ: Lawrence Erlbaum.

Kupersmitt, J., & Berman, R. A. (2001). Linguistic features of Spanish-Hebrew children’s nar-ratives. In L. Verhoeven & S. Strömqvist (Eds.), Narrative development in a multilingual context (pp. 277–318). Amsterdam: John Benjamins. DOI: 10.1075/sibil.23.11kup

Labov, W., & Waletzky, J. (1967). Narrative analysis: Oral versions of personal experience. In J. Helm (Ed.), Essays on the verbal and visual arts (pp. 12–44). Seattle: University of Washington Press.

MacWhinney, B. (2000). The childes project: Tools for analyzing talk. Mahwah, NJ: Lawrence Erlbaum.

Makdissi, H., & Boisclair, A. (2006). Interactive reading: A context for expanding the expression of causal relations in preschoolers. Written Language & Literacy, 9(2), 177–211. DOI: 10.1075/wll.9.2.02mak

Mandler, J. M. (1978). A code in the node: The use of a story schema in retrieval. Discourse Processes, 1(1), 14–35. DOI: 10.1080/01638537809544426

McCabe, A., & Peterson, C. (1991). Getting the story: A longitudinal study of parental styles in eliciting oral personal narratives and developing narrative skill. In A. McCabe & C. Peterson (Eds.), Developing narrative structure (pp. 217–253). Hillsdale, NJ: Lawrence Erlbaum.

McCutchen, D., & Perfetti, C. A. (1982). Coherence and connectedness in the development of discourse production. Text, 2(1-3), 113–139.

Minami, M. (2008). Telling good stories in different languages: Bilingual children’s styles of sto-ry construction and their linguistic and educational implications. Narrative Inquiry, 18(1), 83–110. DOI: 10.1075/ni.18.1.05min

Montanari, S. (2004). The development of narrative competence in the L1 and L2 of Spanish-English bilingual children. International Journal of Bilingualism, 8(4), 449–497. DOI: 10.1177/13670069040080040301

Nakamura, K. (2009). Language and affect: Japanese children’s use of evaluative expressions in narratives. In J. Guo, E. Lieven, S. Ervin-Tripp, N. Budwig, S. Özçalişkan & K. Nakamura (Eds.), Crosslinguistic approaches to the psychology of language: Research in the tradition of Dan I. Slobin (pp. 225–240). Mahwah, NJ: Lawrence Erlbaum.

Nelson, K. (1986). Event knowledge: Structure and function in development. Hillsdale, NJ: Lawrence Erlbaum.

Nir, B., & Berman, R. A. (2010). Complex syntax as a window on contrastive rhetoric. Journal of Pragmatics, 42(3), 744–765. DOI: 10.1016/j.pragma.2009.07.006

Özçalişkan, Ş. (2009). Learning to talk about spatial motion in language-specific ways. In J. Guo, E. Lieven, S. Ervin-Tripp, N. Budwig, S. Özçalişkan & K. Nakamura (Eds.), Crosslinguistic approaches to the psychology of language: Research in the tradition of Dan I. Slobin (pp. 263–276). Mahwah, NJ: Lawrence Erlbaum.

http://dx.doi.org/10.1080/01690968508402071

http://dx.doi.org/10.1016/j.jcomdis.2010.02.002

http://dx.doi.org/10.1075/sibil.23.11kup

http://dx.doi.org/10.1075/wll.9.2.02mak

http://dx.doi.org/10.1080/01638537809544426

http://dx.doi.org/10.1075/ni.18.1.05min

http://dx.doi.org/10.1177/13670069040080040301

http://dx.doi.org/10.1016/j.pragma.2009.07.006



Paradis, J. (2005). Grammatical morphology in children learning English as a second language: Implications of similarities with specific language impairment. Language, Speech, and Hearing Services in the Schools, 36, 172–187. DOI: 10.1044/0161-1461(2005/019)

Paradis, J., Rice, M., Crago, M., & Marquis, J. (2008). The acquisition of tense in English: Distinguishing child L2 from L1 and SLI. Applied Psycholinguistics, 29, 1–34. DOI: 10.1017/S0142716408080296

Paris, A. H., & Paris, S. G. (2003). Assessing narrative comprehension in young children. Reading Research Quarterly, 38(1), 36–76. DOI: 10.1598/RRQ.38.1.3

Pearson, B. Z. (2001). Logic and mind in Spanish-English children’s narratives. In L. Verhoeven & S. Strömqvist (Eds.), Narrative development in a multilingual context (pp. 373–398). Amsterdam: John Benjamins. DOI: 10.1075/sibil.23.14pea

Peterson, C., & McCabe, A. (1983). Developmental psycholinguistics: Three ways of looking at a child’s narrative. New York: Plenum.

Peterson, C., & McCabe, A. (1991). Linking children’s connective use and narrative macro-structure. In A. McCabe & C. Peterson (Eds.), Developing narrative structure (pp. 29–54). Hillsdale, NJ: Lawrence Erlbaum.

Reilly, J., Losh, M., Bellugi, U., & Wulfeck, B. (2004). “Frog, where are you?” narratives in chil-dren with specific language impairment, early focal brain injury, and Williams Syndrome. Brain and Language, 88(2), 229–247. DOI: 10.1016/S0093-934X(03)00101-9

Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Hillsdale, NJ: Lawrence Erlbaum.

Schiffrin, D. (1986). Functions of ‘and’ in discourse. Journal of Pragmatics, 10, 41–66. DOI: 10.1016/0378-2166(86)90099-8

Silliman, E. R., Bahr, R. H., Brea, M. R., Hnath-Chisolm, T., & Mahecha, N. R. (2002). Spanish and English proficiency in the linguistic encoding of mental states in narrative retellings. Linguistics and Education, 13(2), 199–234. DOI: 10.1016/S0898-5898(01)00062-6

Slobin, D. I. (2004). Relating events in translation. In D. Ravid & H. B. Shyldkrot (Eds.), Perspectives on language and language development: Essays in honor of Ruth A. Berman (pp. 115–130). Dordrecht: Kluwer.

Snow, C. E., & Dickinson, D. K. (1991). Skills that aren’t basic in a new conception of literacy. In E. M. Jennings & A. C. Purves (Eds.), Literate systems and individual lives: Perspectives on literacy and schooling (pp. 179–191). New York: State University of New York Press.

Stein, N. L., & Glenn, C. G. (1979). An analysis of story comprehension in elementary school children. In R. O. Freedle (Ed.), New directions in discourse processing (vol. 2, pp. 53–120). Norwood, NJ: Ablex.

Talmy, L. (1991). Path to realization: A typology of event conflation. Proceedings of the 17th Annual Meeting of the Berkeley Linguistics Society (pp. 480–519) Berkeley, CA: Berkeley Linguistics Society.

Talmy, L. (2000). Toward a cognitive semantics, vol. II: Typology and process in concept structur-ing. Cambridge, MA: MIT Press.

Tapiero, I., van den Broek, P., & Quintana, M. P. (2002). The mental representation of narrative texts as networks: The role of necessity and sufficiency in the detection of different types of causal relations. Discourse Processes, 34(3), 237–258. DOI: 10.1207/S15326950DP3403_1

Tomasello, M. (2000). First steps toward a usage-based theory of language acquisition. Cognitive Linguistics, 11(1-2), 61–82.

Trabasso, T. (2005). Goal plans of action and inferences during comprehension of narratives. Discourse Processes, 39(2-3), 129–164. DOI: 10.1080/0163853X.2005.9651677

http://dx.doi.org/10.1044/0161-1461(2005/019)

http://dx.doi.org/10.1017/S0142716408080296

http://dx.doi.org/10.1598/RRQ.38.1.3

http://dx.doi.org/10.1075/sibil.23.14pea

http://dx.doi.org/10.1016/S0093-934X(03)00101-9

http://dx.doi.org/10.1016/0378-2166(86)90099-8

http://dx.doi.org/10.1016/S0898-5898(01)00062-6

http://dx.doi.org/10.1207/S15326950DP3403_1

http://dx.doi.org/10.1080/0163853X.2005.9651677



Trabasso, T., & Nickels, M. (1992). The development of goal plans of action in the narration of a picture story. Discourse Processes, 15(3), 249–275. DOI: 10.1080/01638539209544812

Trabasso, T., & Rodkin, P. (1994). Knowledge of goals/plans: A conceptual basis for narrating “Frog, where are you?” In R. A. Berman & D. I. Slobin (Eds.), Relating events in narrative: A crosslinguistic developmental study (pp. 85–107). Hillsdale, NJ: Lawrence Erlbaum.

Trabasso, T., & Stein, N. L. (1997). Narrating, representing, and remembering event sequences. In. P. W. van den Broek, P. J. Bauer & T. Bourg (Eds.), Developmental spans in event compre-hension and representation: Bridging fictional and actual events (pp. 237–270). Hillsdale, NJ, England: Lawrence Erlbaum Associates.

Trabasso, T., Suh, S., & Payton, P. (1995). Explanatory coherence in understanding and talking about events. In M. A. Gernsbacher & T. Givón (Eds.), Coherence in spontaneous text (pp. 189–214). Amsterdam: John Benjamins.

Trabasso, T., van den Broek, P., & Suh, S. Y. (1989). Logical necessity and transitivity of causal relations in stories. Discourse Processes, 12(1), 1–25. DOI: 10.1080/01638538909544717

Tur-Caspa, H., & Dromi, E. (2001). Grammatical deviations in the spoken and written language of Hebrew-speaking children with hearing impairments. Language, Speech, and Hearing Services in Schools, 32, 79–89. DOI: 10.1044/0161-1461(2001/007)

van Dijk, T. A. (1977). Text and context. London: Longman.Viberg, Å. (1994). Bilingual development of school-age students in Sweden. In G. Extra &

L. Verhoeven (Eds.), The crosslinguistic study of bilingual development (pp. 181–198). Amsterdam: North Holland.

Viberg, Å. (2001). Age-related and L2-related features in bilingual narrative development in Sweden. In L. Verhoeven & S. Strömqvist (Eds.), Narrative development in a multilingual context (pp. 87–128). Amsterdam: John Benjamins. DOI: 10.1075/sibil.23.04vib

Vion, M., & Colas, A. (2005). Using connectives in oral French narratives: Cognitive constraints and development of narrative skills. First Language, 25(1), 39–66. DOI: 10.1177/0142723705046899

Appendix I. Sentence level errors in the Hebrew L2 stories

Table A. Number of children per age group who produced one and at least two of the dif-ferent sentence level errors in their storiesAge group 6-year-olds (N=20) 8-year-olds (N=17)

One Two (+) One Two (+)agreement 5 7 7 2morphological markers 6 4 4 1verb pattern 4 2 2 0word order 3 1 3 0syntactic deviations 3 4 3 0collocations 6 5 7 3lexicon 7 2 3 2

http://dx.doi.org/10.1080/01638539209544812

http://dx.doi.org/10.1080/01638538909544717

http://dx.doi.org/10.1044/0161-1461(2001/007)

http://dx.doi.org/10.1075/sibil.23.04vib

http://dx.doi.org/10.1177/0142723705046899



Types of error: Definitions and examplesThe error analysis was meant to measure grammatical deviations at the level of morphology, syntax, and pragmatics, with specific categories adapted from Tur-Kaspa & Dromi, (2001) and Kupersmitt & Berman, (2001). Errors are marked with an asterisk and correct forms given be-tween square brackets. Missing categories are represented with the symbol Ø.

AGREEMENT [AGR] — Hebrew is highly inflected and requires that verbs, nouns, adjectives, and pronouns agree for number, gender and person.

Noun/Pronoun-Verb agreement: *haya [hayu] lahem arba kisa’ot ‘(There) *was-sing-masc to-them [=they had] four chairs-pl-masc’; ve hi *nika [nikta] oto im smartut ‘she *cleaned-masc him with a cloth’

Noun-Pronoun agreement: hu lo ohev et ha-ambatia *ha-ze [ha-zo] ‘he doesn’t like *this-MASC bath-fem’

MORPHOLOGICAL MARKERS [MM] — refers to omission or wrong selection of determin-ers, connectors, or prepositions. Under this category we also included the omission of the oblig-atory object marker et.

ra’u et *Ø- xatul [ha-xatul] ‘(they) saw *Ø-cat [the-cat]’; ha-xatul yashan *Ø-salsela [ba-salsela] ‘the cat slept *Ø-basket [in-the basket]’; ha-xatul shavar *Ø-OBJ ha-calaxot [et-obj ha-calaxot] ‘the cat broke *Ø-OBJ the dishes’

VERB PATTERN [VPT] — Hebrew verbs are classified according to a system of conjugation patterns (biniyainim) which mark features such as valence, transitivity, causativity, or reflexivity.

Ha-yalda *hityiabsha [yibsha] et ha-xatul ‘the girl dried up the cat’ (in this case a reflexive pat-tern was selected instead of a causative pattern).

WORD ORDER [WO]

calaxot ha-xatul shavar [ha-xatul shavar calaxot] ‘dishes the cat broke’ [the cat broke (the) dishes)

SYNTACTIC DEVIATIONS [SYN] — including omission of objects in obligatory contexts, wrong selection of verb tense in subordinated contexts, or wrong formation of complex sen-tences.

Ima notenet *Ø et [lahem, la-yeladim] et ha-oxel ‘(the) mother gives *Ø [them, the children] the food’

COLLOCATIONS [COL] — including inappropriate combinations of lexical (verb-noun, noun-adjective) and non-lexical (verb-prep, verb-adverb) categories.

hem axlu be-sheqet bli *dibur-n [ledaber] ‘they ate silently without talk-n [talking]’

LEXICON [LEX] — include a range of compensatory strategies to cope with lexical difficulties such as (a) use of a word referring to some of the semantic attributes of the unknown word (e.g., yetush ‘mosquito’ for zvuv ‘fly’); (b) use of a word similar in phonological form, including phonemic metathesis (e.g., sargu oto for serku oto ‘combed him’); (c) replacing a word by a very general term, including super-ordinate semantic categories (e.g., xaya ‘animal’ for zvuv ‘fly’); (d) replacing a word by a first language word; (e) paraphrasing, such as a description of the term (e.g., masheu kmo she samim al ha-rosh ‘something like you put on your head’); (f) other strate-gies used to overcome lexical difficulties such as asking the investigator.



Appendix II.

a. Story content

Picture number Episodic component Content1 Setting Mother washing the dishes, boy and girl eating their

meal, cat sleeping on a bench.2 Initiating event Bee flies into kitchen. Boy and girl stop eating. Boy

tries to get rid of the bee. Cat wakes up. Mother still washing the dishes.

3 Attempt 1 Cat jumps onto bee. Bee flies down toward the meal on the table. Boy and girl stand up, shout and raise their hands in the direction of the cat. Mother turns around and looks surprised.

4 Outcome 1 Bee Bee flies away.4 Outcome 2 Cat Cat lands on the table and spills the soup. Children

and mother look at the cat.5 Attempt 2 Mother washes the cat. Boy looks and girl holds towel.6 Outcome 3 Cat is clean and happy. Mother and children smile.

b. The motion chain (Goralnik, 1995)

The development of coherence and cohesion in monolingual and sequential bilingual children’s narratives: Same or different?

Documents