Background and aim: it has been suggested that story retell will lead to more informative or syntactically more complex stories, since an exact verbal model is given. However, story generation might better reflect genuine storytelling skills, such as narrative organization (e.g. Kunnari et al., 2016). Previous results regarding differences between girls and boys are inconclusive. The aim of the current study is to compare story retell and story generation in a large sample of Swedish preschoolers.

 Materials and methods: data were collected within the framework of an intervention study (Gerholm et al., 2018) and included 431 children (mean age 5;3, range 3;10 – 6;4), of which 24 % were bi- or multilingual and 6.7 % had a language disorder. The Renfrew Bus Story (Renfrew, 1995; Svensson & Tuominen-Eriksson, 2002) was used to elicit story retells and Frog, Where Are You? (Mayer, 2003), was employed to generate stories. Receptive vocabulary was assessed with PPVT-4 (Dunn & Dunn, 2007). Testing was video and audio recorded and the stories were transcribed in ELAN software. Measures regarding story content, story length, syntactic complexity and morphological accuracy were extracted from the narratives. Scoring of story content was based on the Bus Story manual and on the event structure of Frog, Where Are You?. Story length was defined as number of unified predicates (Berman, 1988). Syntactic complexity was defined as number of subordinate clauses and morphological accuracy as the proportion of morphologically well-formed utterances. The scores regarding story content/information were z-transformed to enable comparisons between the two story conditions.

 Results: in story generation, children told significantly longer stories (M_generation = 28, SD_generation = 15; M_retell = 17, SD_retell = 7, p < .001) and had a larger proportion of morphosyntactically correct utterances (M_generation = 73 % well-formed clauses, SD_generation = 19) compared to story retell (M_retell = 65 % well-formed clauses, SD_retell = 23, p < .001). Syntactic complexity was also significantly higher in the story generation condition, but the difference was possibly due to outliers. In story retell, children included more story content (M_retell = 0.33, SD_retell = 0.18) than in story generation (M_generation = 0.29, SD_generation = 0.15, p < .001). Boys told significantly longer stories than girls in both conditions, no other differences between girls and boys were found. Children with additional other language/s than Swedish spoken in the home did not differ from monolingual Swedish children on any of the language measures. Children with a known language disorder (LD) did not differ from typically-developing (TD) children on any of the language measures. Range of performance was substantial within the TD group. 

Conclusions: in line with previous work, story retell led to more informative stories. However, there was no support for the claim that story retell would lead to more syntactically complex stories compared to story generation. Lack of differences between TD and LD children may indicate the presence of undetected language disorders in the TD group.