Analysing narratives of Maltese bilingual children: A pilot study on linguistic profiles in maltreatment research

Background
Narratives are considered an effective tool to generate quasi-naturalistic oral representations of language ability, particularly in the context of assessment (Gatt & Borg Cutajar, 2023). Microstructural aspects of narratives, which focus on linguistic structures embedded within the narrative, are typically more language-specific than macrostructural aspects (Lindgren & Bohnacker, 2022). Malta presents a relatively unique linguistic landscape, with both English and Maltese functioning as majority languages, each with stable and unified input (Gatt & Dodd, 2022). However, these typologically dissimilar languages produce disctinct linguistic profiles that require analyses that are sensitive to this specific bilingual context.

Method
This pilot study assessed the feasibility of data collection tools and analytical methods for investigating the linguistic profiles of Maltese bilingual 5-to-8-year-old children with a history of maltreatment. The study employed a small-scale, descriptive cross-sectional design with three 5-to-8-year-old Maltese bilingual children with a history of maltreatment and five children in a comparison group. Child participants completed the Multilingual Assessment Instrument for Narratives (MAIN, Gagarina et al., 2020). Narrative macro- and microstructure were analysed using MAIN scores, along with additional measures for lexical diversity, mean length of utterance (MLU) and mismatch analyses. Given the sample size and the non-normative distribution of the sample, non-parametric tests were run in attempt to very tentatively investigate potential relationships that might indicate increased sensitivity of one type of analysis over another.

Results
Preliminary findings suggest across-group differences in macro- and microstructure were, at times, more pronounced in L2 performance, however, interpretation is constrained by a very small sample size. Internal State Terms from the MAIN were insufficient for capturing microstructure variation, necessitating additional analysis. A moving-average-type-token-ratio proved more appropriate than a standard type-token ratio. Although no statistically significant differences across groups were generated for MLU, it remains a relevant measure main study. Mismatch analysis indicated across-group differences in uncorrected morphological mismatches in the L2, which approached significance (p = 0.053).

Conclusion
The small sample size calls any interpretations, including those suggested by the non-parametric statistical tests, into question. Further analysis with a larger sample size may indicate across-group differences in L2 performance.