Running head: PERSONAL NARRATIVE ASSESSMENT FOR SCHOOL CHILDREN 1 Personal narrative assessment for Cantonese-speaking school-age children SO Ka Sin Kasty

This study investigated the oral personal narrative production in 60 Cantonesespeaking children between 6 and 9 years of age. Three language measures, namely Total Personal Narrative Score, Macrostructure Score and Microstructure, were derived from three personal narratives produced by each participant after listening to the sample stories. The participants also completed the Cantonese Grammar subtest of the Hong Kong Cantonese Oral Language Assessment Scale (HKCOLAS). Analysis of the data demonstrated that there were significant group differences in narrative production skills with respect to all the tested language measures. Further investigation identified that the high point element ‘Resolution’ at the macrostructure level was the most sensitive to age, and ‘Conjunction’ was a better age indicator than ‘Referencing’ at the microstructure level. The correlation between the Macrostructure and Microstructure scores was positive but weak, while the correlation between Total Personal Narrative and Cantonese Grammar scores was positive and strong. No significant group difference was found with respect to the length of narratives. These findings supported McCabe and Bliss (2003) and Cheng’s (2004) reports that personal narrative production can be further developed as an efficient tool for screening and assessment of language capabilities of Cantonese-speaking school-age children. PERSONAL NARRATIVE ASSESSMENT FOR SCHOOL CHILDREN 3 Personal narrative is one of the earliest developing genres of narration, which grows throughout the preschool and early elementary school years. It generally develops around events that children personally experienced in the past, such as experience of getting hurt and holidays (Peterson & McCabe, 1983; Preece, 1987). Children produce personal narratives for communication with peers and adults, and this ability is found to be related to their academic, social and linguistic skills (e.g. Crais & Lorch, 1994; Scarborough, 2001). Despite the importance of personal narrative skills, there is a lack of research focusing on its development in Cantonesespeaking children. To address this inadequacy, this study aims to provide descriptive information about the personal narrative skills of Cantonese-speaking school-age children in Hong Kong, and develop an assessment framework that can be easily used in school settings. Levels of Personal Narrative Assessment Personal narrative assessment involves macro-structure and micro-structure levels of analysis. For macrostructure, the literature reported the use of high point and story grammar analysis. “High point analysis” involves the identification of the high point elements that describe a child’s discourse-level abilities with respect to the overall narrative structure (McCabe & Rollins, 1994). High-point elements (HP), including introducer, orientation, complicating action, evaluation, resolution and coda, have been reported to be sensitive to age changes (McCabe & Bliss, 2003). “Story grammar analysis” examines how a narrative is structured around the explicit goals of a protagonist (Stein & Glenn, 1979). Story grammar analysis examines the extent to which stories are structure around the explicit goals of a protagonist. Between the two, high point analysis has been predominantly applied in the macro-analysis of personal narratives (e.g. Labov, 1972; Liles, 1985). Literature reviews suggested an PERSONAL NARRATIVE ASSESSMENT FOR SCHOOL CHILDREN 4 underestimation of children’s personal narrative abilities with the use of story grammar analysis. For instance, if a child tells a story about a time she was hurt but untreated, it would be considered as a good narrative using of high point analysis but defined as a primitive story using story grammar analysis. Another concern is that story grammar analysis often failed to discriminate impaired narration, as research indicated that children with language impairment can produce all key story grammars during story-retelling (e.g. Ripich & Griffith, 1988). For microstructure, McCabe and Bliss (2003) proposed a model for assessing syntactic structures of personal narratives, namely the Narrative Assessment Profile. Six microstructure components were investigated, including topic maintenance, event sequencing, explicitness, referencing, conjunction and fluency (McCabe & Bliss, 1998). Table 1 illustrated the definition of each component. To (2010) studied the use of two components, namely referencing and conjunction, for assessing narrative production of school-age children, and both revealed age-related changes. Table 1. Definition of Microstructure components Component Definition Topic Maintenance Describes how the utterances relate to a central topic. Event sequencing Involves the presentation of events in chronological order. Informativeness Evaluates if a narrative contains full elaboration of the plot, and contains also the number of high point elements Referencing Describes the adequacy in the identification of individuals, features and events Conjunction Consists of words (e. g. and, then) or phrases that link utterances and events, which contribute to coherence. Fluency Includes lexical or phrasal interruptions in utterances. PERSONAL NARRATIVE ASSESSMENT FOR SCHOOL CHILDREN 5 Current Practice of Narrative Assessment and its Clinical Uses Compared to naturalistic conversation, narrative serves as a more productive context for the elicitation of complex syntactic structures that normally develop in school-age children (Gummersall & Strong, 1999). There are various types of elicitation methods currently used in narrative assessment, with fictional and personal narratives commonly adopted. Fictional narrative assessment is predominantly employed in the field of speech-language pathology (e.g. Gillam et al, 2004; Justice et al, 2006). Past studies of oral narrative skills of school-age children primarily focused on fictional story-retell and generation rather than personal narratives. Numerous studies on fictional narratives of children have been implemented in English-speaking communities. These studies identified age-sensitive narrative development until age 9 (e.g. Owens, 1999). Researches on fictional story-retelling of Chinese-speaking children were divided into two levels. Macroanalysis concerns development of overall narrative organization in terms of causal and temporal relationships (Tam, 2006; Wong, 1995). Microanalysis evaluates the number and diversity of cohesive ties, which are the grammatical and lexical links within sentences that hold a text together in both linguistic and nonlinguistic content (Snow & Ninio, 1986). Cohesive ties were classified into five parameters, namely conjunction, ellipsis, lexical cohesion, referencing and substitution (Law, 2001; Leung, 2008; Tsui, 2004). Conjunction is defined as a part of speech that connects words, phrases or clauses together, while referencing points to recurring entitles that mentioned earlier in a text (To, 2006). Man (2010) and To (2010) examined these two cohesive ties, and reported that the measure of conjunction was a better predictor of age than referencing. As the studies involved fictional narratives for measurement, the findings may not generalize to personal narratives due to genre difference. PERSONAL NARRATIVE ASSESSMENT FOR SCHOOL CHILDREN 6 For personal narrative assessment, research evidence has in fact suggested that children are more competent at producing advanced story grammar elements, such as evaluations, explanations and ending, in personal narratives than those in other genres (Losh & Capps, 2003). These findings have been contributed to its high frequency of use in naturalistic contexts. Preece (1987) and McCabe et al (2008) studied a range of narrative forms produced in conversation among preschoolers and school-age children, in which 80% of the samples obtained were personal narratives. Studies of Personal Narrative in the Western Literature Research in personal narratives proceeded in two directions. One focuses on personal narrative development across age. Age-related variations of personal narrative skills were noted, in which older children produced personal narratives with greater length and more story grammar elements (e.g. Merritt et al, 1989). The other direction contrasts personal narrative performance between children with typical language development and language impairment. Van Bysterveldt (2012) compared micro-components of personal narrative of school-age children with Down Syndrome and normal language development, with significant correlation found between age and number of different words. As English-speaking children were included as participants in the above studies, the findings may not be valid for Cantonesespeaking group. Studies of Personal Narrative in Cantonese The macrostructure level of personal narrative in Cantonese was previously studied using high point analysis. Cheng (2004) investigated the macrostructure level of personal narratives, and identified an increase in the use of high point element ‘Evaluation’ in telling personal narratives in a group of school-age children. The PERSONAL NARRATIVE ASSESSMENT FOR SCHOOL CHILDREN 7 same variable was investigated in Ho’s study (2007), in which no significant changes were found across preschool, school-age children and adults. The inconsistency was likely to be the result of small sample size in the latter study, with only few specific components of personal narratives studied.