Narrative Skills in Autistic and Non-Autistic Preschool Children: A Scoping Review

Abstract

Background/Objectives: Narrative skills play an important role in children’s overall development from a very young age, and they are linked to social behavior, as well as several emotional and cognitive outcomes. Young autistic children often experience difficulties in their narrative skills and these difficulties may impact their social interactions. The present study reviews recent findings to detect factors influencing narrative development in autistic and non-autistic preschool children, and to identify trends or gaps in the existing literature. Following screening and eligibility assessment, 39 studies met the inclusion criteria and were included in the review. Methods: The Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines were followed. Results: Non-autistic children show a clear, age-related progression in narrative skill development, moving from simple to complex structures at the level of microstructure and advanced inferential abilities at the level of macrostructure, which are strongly linked to core language and cognitive development. Conversely, autistic children primarily face challenges in narrative macrostructure and coherence, demonstrating deficits in integrating information and making inferences, which is consistent with weak central coherence in autism. Conclusions: The evidence suggests that narrative development in autism reflects qualitative differences rather than mere delay, particularly in the organization and integration of macrostructural story elements. These findings underscore the importance of interventions that move beyond surface-level linguistic skills to explicitly target global coherence, causal structuring, and inferential reasoning. Future research should further clarify developmental trajectories and the mechanisms linking narrative competence with broader social and cognitive outcomes.

1. Introduction

Narrative skills play an important role in the lives of children from a very young age. Children engage with narratives by listening to others’ stories, sharing their personal experiences, and by both telling and retelling real or fictional stories to others. While young children’s early attempts at narration may be limited by their linguistic abilities, language skills evolve over time, allowing them to succeed in increasingly complex narratives (Veneziano & Nicolopoulou, 2019). Several factors have been found to influence the development of young children’s narrative skills, such as their age, the socio-cultural and socio-economic status of the family, i.e., the broader cultural context encompassing shared norms, values, and communicative practices that shape storytelling (Gardner-Neblett & Iruka, 2015), the parental socialization practices, including sharing past experiences and events, book-reading routines, and parental scaffolding during storytelling, and the nature of mother-child interactions (Bohanek et al., 2009; Smith et al., 2019). For example, when parents share family stories or support their children in describing events using prompting questions or tend to read books together on a frequent basis, they actively model narrative construction (McCabe & Peterson, 1991), and thus, integrate storytelling into daily life (S. G. Paris et al., 2005).

However, to effectively narrate a story, children must acquire both content knowledge of basic episodic structure, i.e., the goal-attempt-outcome sequence in a single episodic unit, and an understanding of overall narrative structure, which extends beyond single episodes and reflects the overall coherence of the story as a whole and the organization of the story’s characters, events, and their interrelationships within a causally connected network (Shapiro, 1991). These tasks demand a range of cognitive, linguistic, pragmatic, and socio-cultural skills that enable children to construct a coherent mental model of the narrative, understand causal relationships that highlight characters’ goals and mental states as drivers of action in a narrative, grasp temporal relations between events, and sequence events appropriately (Colletta et al., 2010; Hamilton et al., 2021). Thus, children’s narrative skills extend beyond mere word retrieval or the linear presentation of events and characters (Stadler & Ward, 2005), incorporating pertinent details of the plotline of the stories (Hamilton et al., 2021; Dore et al., 2018; C. Norbury et al., 2014; Pavias et al., 2016). Specifically, producing a coherent and meaningful story requires children to possess knowledge of the story events and recall personal and social experiences for inference-making, which enables the child to derive information that is not explicitly stated by integrating textual cues with prior knowledge (Kang et al., 2012). These processes draw upon event scripts, defined as structured mental representations of typical sequences of actions in familiar situations that support the prediction and organization of narrative information. Also, to efficiently tell a story, the child needs to organize events sequentially (Kelter et al., 2004), and maintain an understanding of multiple characters and their perspectives; for example, an effective child narrator needs to be able to distinguish between story actors, who refer to characters performing observable actions, agents, who imply intentionality and goal-directed behavior, and persons, who denote more psychologically elaborated characters with internal states, motivations, and perspectives (Mason & Just, 2009). The child also needs to produce linguistically sophisticated narratives across various levels of language processing, including phonology, vocabulary, morphology, and syntax (C. Norbury et al., 2014; Oakhill & Cain, 2011; Westerveld & Gillon, 2010; Westerveld et al., 2012).

The structural properties of a narrative are often examined through its microstructure and macrostructure features. Macrostructural analysis focuses on the coherence of the narrative production, which refers to the overall meaningful organization of a story, including the logical sequencing of events and the integration of characters, goals, and outcomes into a unified whole (Mäkinen et al., 2014). In addition to coherence, macrostructure also includes the content adequacy criterion, which refers to the extent to which a narrative includes all essential and relevant story elements necessary for the listener to fully understand the plot and its causal–temporal relations. Within this framework, event content reflects the number and relevance of key story events represented in the narrative, such as the story’s initial events, the story protagonists’ actions to reach the goal, and outcomes, which all contribute to how comprehensively the plot is conveyed. This plot system, also known as story grammar, is characterized as the backbone of a narrative (Geelhand et al., 2020; Dosi & Boni, 2023). As such, narratives can be categorized as complete, simplified, or distorted, reflecting qualitative differences in coherence (Mäkinen et al., 2014). Specifically, complete narratives include all essential story elements and logical connections; simplified narratives present reduced structural complexity but maintain overall coherence; and distorted narratives contain disorganized or inconsistent elements that disrupt meaning. Additional narrative-level constructs include programming, which refers to the planning and organization of the linguistic output (Mäkinen et al., 2014), and the appropriate use of referential expressions that underlie the story’s overall clarity and comprehensibility (He et al., 2025). In contrast, microstructure focuses on cohesion, i.e., the linguistic devices that connect elements within the text, such as connectives, referential markers, and lexical ties, and the linguistic form at the level of individual utterances. This includes grammatical sentence structure (i.e., the organization and accuracy of sentence-level grammar), lexical diversity, and syntactic complexity (i.e., the extent of hierarchical structuring, such as embedding and subordination), which influence how ideas are expressed, which reflect the child narrator’s narrow syntactic skills (Heilmann et al., 2010). Lexical measures also include usual verbal designation (UVD), which refers to the use of conventional or expected lexical labels for story elements, such as naming characters or objects using typical terms. This measure captures lexical appropriateness and familiarity in narrative production. Narrative length and structural complexity are reflected in communication units (C-units), which are independent clauses and their modifying clauses, and T-units, defined as one independent clause plus any dependent modifiers of that clause. C- and T-units have been frequently used as a measure of utterance segmentation in narrative analysis, and reflect how children organize and express their ideas at the sentence level (Mäkinen et al., 2014; He et al., 2025; Charest et al., 2020). Overall productivity, defined as the total number of words or C-/T-units produced, provides complementary information about the quantity of language generated in the narrative. Both macrostructure and microstructure are essential for creating informative and engaging narratives, as they work together to create a cohesive and immersive storytelling experience (Charest et al., 2020). Understanding both aspects helps in assessing a child’s narrative development and overall language skills.

Exposure to stories from a very young age through diverse media (e.g., from picture books to television and apps) enhances children’s language, literacy, and socio-cognitive abilities (Veneziano & Nicolopoulou, 2019; Adornetti et al., 2022). Original story production (also known as telling) and story retelling, while related, seem to engage different underlying skills (Vretudaki, 2022; Yuan et al., 2018). Original storytelling requires children to create a narrative from prompt-based cues and organize story elements in a logical, causal, and temporal framework (Kirby et al., 2021), while also providing sufficient background information (Heilmann et al., 2010). Conversely, story retelling demands that children process and remember all the information presented to them either orally or visually, construct mental representations in a meaningful sequence, and express the narrative in a coherent manner (Tompkins et al., 2013). Research suggests that story retelling, but not telling, positively correlates with children’s memory skills (Mäkinen et al., 2018), while both modes of narrative production are grounded in children’s comprehension abilities (Mäkinen et al., 2018; Potocki et al., 2013). Furthermore, story retelling serves as a vital precursor to children’s ability to generate their own personal and fictional narratives and develop writing skills (Spencer & Petersen, 2020). Consequently, narrative skills not only enhance oral literacy, but also lay the groundwork for future academic learning, including reading comprehension, writing (Hayward et al., 2009; Griffin et al., 2004), and even mathematical reasoning skills (O’Neill et al., 2004). Particularly, fictional narratives have been identified as a strong predictor of academic success (Petersen et al., 2010) and later literacy achievement (Abbott & McCarthey, 2001).

Narrative comprehension or listening text comprehension, which is also an intricate process, encompasses not only basic word and phrase understanding, but also higher-order cognitive skills and an expanding awareness of social understanding, including mental states (i.e., intentions, beliefs) and behaviors. To assess these skills, narrative questioning can vary from surface-level recall to deeper integration of cognitive and social information. Specifically, identification questions prompt basic recall of narrative elements, such as naming characters or events, whereas elaboration questions encourage children to expand on motivations, causal links, or internal states, supporting inferential and descriptive responses. Kintsch’s construction-integration model of text comprehension (Kintsch, 1988, 1994) postulates that both foundational language and cognitive skills, such as vocabulary, syntactic knowledge, working memory, and attention, alongside Theory of Mind (ToM), which refers to the ability to recognize that another person’s knowledge is different from our own, thus, allowing us to predict and interpret others’ behavior based on attributed mental states (Dore et al., 2018), contribute to formulating a coherent mental representation of a narrative (Florit et al., 2014; Kendeou et al., 2008; Kim & Phillips, 2014; Kim, 2015; Strasser & del Río, 2014). The processing and parallel construction of complex linguistic (e.g., syntactic and lexical structures) and cognitive representations (e.g., event models and causal relations) during narrative performance involve the ability to simultaneously manage multiple levels of information during narrative production, including linguistic encoding, event sequencing, and integration of social–cognitive knowledge. Thus, foundational language and cognitive skills seem to be vital for discourse comprehension and broader learning tasks (Lepola et al., 2012).

1.1. Narratives of Non-Autistic Children

Narrative skills, encompassing both comprehension and production, undergo significant quantitative and qualitative development during the preschool years, typically emerging around ages 4 to 5 (Kendeou et al., 2008; Leinonen et al., 2003). In non-autistic children, these skills begin to form early in childhood and continue to evolve throughout adolescence, reflecting variability in individual developmental trajectories (Harvey et al., 2023).

At approximately 40 months of age, children begin to create personal narratives, recounting lived experiences while incorporating contextual and evaluative information (Chang, 2004). From ages three to four, children are capable of telling simple stories that exhibit a basic structure, consisting mainly of temporally sequenced events rather than thematically coherent narratives (Kendeou et al., 2008; Slobin, 2004). Thus, they often lack the more sophisticated storytelling strategies seen in older children (Chang, 2004). By the age of four, children show progress in character and action descriptions (Muñoz et al., 2003). However, substantial developmental advancements between ages four and five, particularly in terms of vocabulary, character viewpoint representation, and overall story structure, have been well documented (Westerveld et al., 2012). Despite these improvements, challenges persist in structuring narratives around clear goals, initiating events, and accurately conveying causal relationships (Khan et al., 2016). Concurrently, the ability to build a coherent mental representation of a story through inference becomes more solidified around the age of four years (Kendeou et al., 2008; Lepola et al., 2012; Filiatrault-Veilleux et al., 2016). By age five, improvements are evident in children’s ability to connect events causally and temporally, add setting details, and utilize rich semantic features when describing the events and the story setting (Castilla-Earls et al., 2015; Vretudaki & Tafa, 2022). Nonetheless, the formation of an overarching plot or goal often remains challenging. The critical developmental period between ages five and six is characterized by a significant enhancement in narrative complexity, as children increasingly employ advanced temporal and referential markers, background information, evaluations, and intricate sentence constructions (He et al., 2025). Narrative production is more detailed and the hierarchical organization of events is based on cause–effect relationships (Pistav Akmese & Kanmaz, 2021). However, research suggests that narrative skills remain qualitatively underdeveloped until approximately age six, with children frequently demonstrating difficulty connecting events cohesively and producing structured stories without external support (Boudreau, 2008). By around ten years of age, children’s narrative abilities closely resemble those of adults, featuring comprehensive episodes, causal and temporal connections, and references to characters’ mental states such as emotions (King et al., 2014).

Additionally, studies further highlight variability in narrative performance among non-autistic preschool children, particularly in relation to bilingualism and language exposure. Hipfner-Boucher et al. (2015) reported differences in microstructural narrative measures across subgroups of English language learners, while macrostructural performance remained comparable. Similarly, Rodina (2017) found that bilingual preschoolers demonstrated comparable macrostructural abilities across languages, with microstructure being more sensitive to language exposure. Longitudinal evidence from Lindgren (2019) further showed that narrative macrostructure develops rapidly during the preschool years, reaching a plateau around age six.

It is noteworthy that young children’s storytelling abilities can vary depending on the elicitation methods employed. For instance, narratives are typically longer and better organized, with more references to mental states, when children retell stories based on pictures compared to simply creating stories from visual stimuli (Lever & Sénéchal, 2011). The choice of materials used to elicit narratives, such as picture books, digital texts, audio recordings, or videos, substantially influences the quality of children’s storytelling. Because young children often struggle to comprehend a narrative without visual cues, their stories tend to exhibit greater grammatical complexity when supported by visual aids rather than when relying solely on audio prompts. Frequently, narrative ability is embedded within broader linguistic assessments of typical development. For instance, Thordardottir et al. (2010) demonstrated that in 5-year-old French-speaking children, narrative structure scores show a systematic increase alongside vocabulary and morphosyntax, suggesting that while narrative is a distinct communicative skill, it remains deeply integrated with core language knowledge.

Narrative comprehension refers to conceptual knowledge that facilitates the understanding of words and text, and plays a crucial role in cognitive development, particularly during the preschool years. Narrative comprehension is a multifaceted process that relies upon the simultaneous advancement of various skills, including story structure comprehension, ToM, and perspective-taking abilities. Specifically, children initially construct basic story frameworks known as narrative scripts, which outline familiar sequences of events (Lervåg et al., 2018). They then use the semantic, conceptual, and narrative relations to grasp the internal experiences of the story characters, such as their thoughts and emotions (Astington, 1993), and to enhance their representational abilities, empathy, and working memory, all of which contribute to their understanding of both the external and internal nuances of the stories (Cain et al., 2001). As they grow older, children transition to more complex narrative schemas, which encompass key elements of stories (including characters, settings, and conflicts) along with an understanding of how events are sequenced both temporally and causally (A. H. Paris & Paris, 2003). Evidence suggests that inferencing skills, evaluated through storytelling tasks, correlate with story comprehension among preschoolers (Tompkins et al., 2013). Finally, mixed findings have been reported regarding the link between narrative comprehension and reading comprehension, with some studies illustrating the connections between narrative skills and reading outcomes (Justice et al., 2011), and others failing to find this predictive link (Adlof et al., 2021). In summary, while non-autistic children exhibit progressive advancements in narrative skills throughout early childhood, the pace and quality of this development vary significantly among individuals. Understanding these developmental trajectories provides valuable insights for educators and other practitioners working to support children’s narrative competencies. Indeed, relevant research has shown that highly contextualized activities, including shared book reading, literacy exposure and dynamic adult-child interactions during storybook reading, may promote children’s linguistic capacities and narrative skills (Smith et al., 2019; S. G. Paris et al., 2005).

1.2. Narratives of Autistic Children

Autism spectrum disorder (ASD) or autism is a neurodevelopmental condition present from early childhood with symptoms varying in severity. It is characterized by persistent differences in social communication and interaction, alongside restricted or repetitive patterns of behavior, interests, or activities (Geelhand et al., 2020). Given that narrative production and comprehension rely heavily on social–cognitive, linguistic, and inferential processes, examining narrative skills in autistic preschool children is particularly important. Autistic children face great challenges in their daily social interactions and communication (King et al., 2014). Research on the narrative skills of autistic individuals has mainly focused on the narrative production of school-aged children and adolescents, or even adults (Lee et al., 2018; McCabe et al., 2013; Rollins, 2014), often presenting divergent findings (Baixauli et al., 2016). Some studies postulate that compared to their non-autistic peers, narratives of autistic individuals are less coherent and structured, shorter, simpler, with weaker macrostructural structure (Dosi & Boni, 2023; Siller et al., 2014). Their narrations are often characterized by fewer causal statements (Diehl et al., 2006), less use of mental state language, fewer or more ambiguous referential expressions, or expressions of emotional states (Colle et al., 2008), and more pragmatic errors (Capps et al., 2000). Autistic children often demonstrate imaginative and organizational difficulties in narratives. Their narrative profiles typically show challenges relating the background information, concluding the narrative, and establishing cohesive connections between story episodes. They also tend to prioritize attention to story details over the global, or macrostructural, features (Barnes & Baron-Cohen, 2012). Other studies suggest that when language and cognitive skills are matched, autistic and non-autistic individuals show few differences in the structural language of their narratives (Dillon & Underwood, 2012). However, as these findings are based on narratives elicited from wordless picture books (e.g., the Frog Story), little is known about how they construct fictional stories of their own that better reflect everyday educational and social contexts. Findings regarding the microstructure of the narratives of autistic children are also ambiguous. Some researchers failed to find any differences in narrative cohesion and story length between autistic and non-autistic children (Diehl et al., 2006; Novogrodsky & Edelson, 2016), while others reported weaker cohesion, shorter mean length of utterances, poorer vocabulary diversity, and lower syntactic complexity in autistic children, especially in fictional narratives (Dosi & Boni, 2023; Kuijper et al., 2017). Research indicates that autistic children use significantly fewer propositions and restricted vocabulary in fictional narrative tasks, as compared to their non-autistic peers (C. Norbury et al., 2014). Studies also highlight that primary-school autistic children tend to incorporate internal states less frequently in their narratives. Consequently, their stories often display reduced emotional expression or limited attention to the thoughts and feelings of the characters (Hilvert et al., 2016; Kauschke et al., 2016). Finally, while research on narrative comprehension in school-aged autistic children is limited, existing reports consistently show that their narrative abilities are significantly poorer than those of their non-autistic peers (C. F. Norbury & Bishop, 2002). These documented difficulties include challenges with drawing appropriate inferences during everyday storytelling tasks and with grasping the overarching context when retelling fictional scenarios (Diehl et al., 2006).

1.3. Aim and Objectives

Despite the critical role of preschool years in all facets of development, i.e., cognitive, linguistic, and social, relatively few studies have addressed narrative production and comprehension, mainly in non-autistic children of this age group. Published work on the narrative skills of preschool autistic children is scarce.

In light of this identified gap, the present PRISMA Extension for Scoping Reviews (PRISMA-ScR) aims to synthesize empirical evidence regarding the narrative skills of monolingual preschool children aged 4 to 6 years, both autistic and non-autistic. The scoping review will pursue the following objectives:

(a)

To identify the factors influencing narrative abilities across high-functioning autistic and non-autistic preschoolers.

(b)

To evaluate how autism influences the narrative production and comprehension skills of preschool-aged children.

(c)

To highlight gaps in the existing literature and recommend directions for future research.

2. Materials and Methods

The scoping review was conducted following the procedures recommended by the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines (Page et al., 2021). To minimize potential biases, multiple researchers participated in the selection and analysis of the included publications.

2.1. Search Strategy

A search strategy was devised and carefully performed in the following electronic databases: ProQuest, APA PsycInfo, ERIC and PubMed. The following keywords and Boolean operators were used: (“autism” OR “autism spectrum disorder” OR “ASD” OR “Asperger syndrome”) AND (“neurotypical” OR “non-autistic” OR “typically developing children”) AND (“narrative” OR “narrative comprehension” OR “story retelling” OR “story generation”) AND (“pre-school” OR “preschoolers” OR “early childhood”). Search filters were applied to include peer-reviewed articles published in English between 2000 and 2025. Alerts for new articles were created for each database. Searches started on the 1st of May 2025, and the final search was performed on 18 April 2026. The chosen articles were retrieved in full-text and were read and assessed by the first three authors separately for the final list of studies to be included. The reference and citation lists of included studies were also screened to locate additional potentially eligible studies. The year 2000 was selected as a cut-off point to ensure alignment with contemporary diagnostic frameworks for autism and to reflect advances in narrative assessment methodologies, thereby enhancing comparability across studies.

The selection of keywords and Boolean operators was guided by the aim to capture studies addressing narrative production and comprehension in both autistic and non-autistic preschool populations. However, given the variability in terminology used across studies, particularly with respect to age groups, which are not always explicitly labeled as “preschool” in titles or abstracts, the search strategy was complemented by manual screening of reference lists and citation tracking of included studies. This approach was adopted to enhance the comprehensiveness of the review and to identify potentially relevant studies that may not have been retrieved through database searches alone. Despite these efforts, it is acknowledged that some studies may not have been included due to inconsistencies in indexing and reporting.

2.2. Eligibility Criteria

A set of eligibility criteria was defined to guide the inclusion and exclusion process of the collected studies. In particular, the inclusion criteria involved research articles that: (1) were published in peer-reviewed journals between 2000–2025, with 2000 being selected as the cut-off year to ensure alignment with contemporary diagnostic frameworks for autism and to reflect substantial advances in narrative assessment methodologies and early identification practices; (2) were written in the English language; (3) were original research studies except case studies and reports; (4) focused on narrative production and/or comprehension; (5) the sample consisted of monolingual and bilingual preschool non-autistic and/or autistic children according to their clinical diagnoses; (6) the sample included children aged 4–6 years, or provided clearly extractable and separately reported data for this age group when broader age ranges were included; (7) examined linguistic and cognitive as well as demographic factors that could influence the development of narrative production and comprehension; (8) provided enough description to guarantee that methodologically sound research practices were used, and (9) the participants had a full IQ mean above 70, based on the score on a standardized intelligence test. Studies were excluded if they (1) involved participants with other medical conditions or disorders (e.g., attention deficit hyperactivity disorder, psychiatric disorder), (2) included comorbidities (e.g., autism with learning disorder), (3) focused on interventions using pre- and post-assessments, (4) included participants outside the specified age range, or (5) had extremely small sample sizes (fewer than five participants), as such designs may reflect case-study findings and limit comparability and generalizability. Studies including broader age ranges were considered eligible only when data for children within the target age range (4–6 years) were clearly reported or could be extracted separately. This criterion was applied to ensure consistency while allowing the inclusion of relevant studies that encompass preschool-aged participants within wider developmental samples. In addition, studies comparing typically developing children with clinical populations other than autism were excluded. This decision was made to maintain a clear and focused comparison between autistic and non-autistic children, in line with the primary aim of the review. Including studies involving other clinical populations could introduce additional heterogeneity and confound the interpretation of findings, as different developmental conditions are associated with distinct narrative profiles. Priority was given to studies in which narrative production and/or comprehension constituted a primary or clearly defined outcome. However, studies in which narrative measures provided sufficiently detailed and relevant data were also considered for inclusion, even when narrative was not the sole focus of the study. To ensure a comprehensive overview, studies were not excluded if narrative was a secondary rather than a primary measure, provided they offered sufficiently detailed data on narrative macrostructure or microstructure. This allowed for the inclusion of significant normative studies (e.g., Thordardottir et al., 2010). that examine narrative skills as part of a broader linguistic profile in the 4–6 age range.

2.3. Data Collection

Two independent authors conducted the study selection process using the Rayyan platform. During the screening phase, they independently assessed all titles and abstracts, achieving a percentage agreement of 95% at the title screening stage and 92% at the abstract screening stage. Any disagreements were resolved through joint review and discussion until consensus was reached prior to full-text review. The same authors proceeded to data extraction through a synoptic table to identify the relevant articles of the eligible studies, including author(s), year of publication, country, sample size, characteristics of the sample (age, gender, and ethnicity), study design, and key findings. In addition, particular attention was given to identifying studies in which relevant age groups or narrative measures were not explicitly indicated in titles or abstracts, in order to minimize the risk of excluding pertinent literature.

2.4. Quality Assessment

Quality assessment of the studies that have been selected was conducted using the standardized JBI Critical Appraisal Checklist for Analytical Cross-Sectional Studies (Moola et al., 2020). The specific checklist recommends that inspecting authors evaluate the following processes in cross-sectional studies: clear recruitment and sampling criteria for the groups, detailed descriptions of the groups’ demographic, language and cognitive profiles, appropriateness of screening tests as well as narrative production and comprehension measures, and assessment for possible bias during analyzing and reporting of study results. The first three authors independently assessed the quality of the individual studies by following the signaling questions from the checklist, and an agreement was reached by further discussion in case of disagreement on the quality of interest.

2.5. Study Selection Criteria

The initial database search identified a total of 140 records. Initially, eleven records were identified as duplicates and were removed to avoid errors of overrepresentation. The remaining 129 articles were then screened based on the following criteria: (i) accessibility, i.e., full text availability, (ii) type of publication; specifically, only peer-reviewed journal publications written in English were included. Other types of publications—such as journal editorials, book chapters and conference proceedings, and journal commentaries were excluded, and (iii) research design; the studies should be original research articles involving empirical quantitative methods and findings on fictional stories from children aged 4–6 years. A total of 19 records were excluded, as 8 were not accessible, 7 were published either in non-peer-reviewed journals or were book chapters, and 4 were not empirical studies. The remaining 110 articles were screened based on their title and abstract. A total of 64 records were excluded because the participants did not fall within the age range specified by the inclusion criteria (i.e., 4–6 years), and 2 records because the sample size was very small (<5 children), while 5 records were excluded for the wrong study design, specifically, they involved qualitative case study designs. After a thorough evaluation of the remaining manuscripts, 39 articles satisfied all the inclusion and exclusion criteria and were included in the scoping review. Figure 1 shows the trial selection flowchart for the studies including autistic and non-autistic preschool-aged children.

3. Results

3.1. Identification and Selection of the Studies

The included studies were characterized by cross-sectional (N = 32) and longitudinal (N = 7) research designs. They were published between 2004 and 2025, with most having been conducted in the United States (N = 10), followed by Canada (N = 6), Australia (N = 3), Finland (N = 2), Germany (N = 2), Greece (N = 2), Italy (N = 2), Norway (N = 2), Brazil (N = 1), Chile (N = 1), China (N = 1), Egypt (N = 1), Israel (N = 1), New Zealand (N = 1), Sweden (N = 2), Russia (N = 1), and Turkey (N = 1). The final selection comprised 33 studies focusing exclusively on (monolingual or bilingual) non-autistic participants, 5 studies that compared (monolingual or bilingual) autistic and non-autistic preschoolers, and 1 study involving only autistic monolingual participants. The sample sizes varied substantially, ranging from 9 to 386 participants. The total number of participants was 3239 for the non-autistic preschool children and 130 for the autistic preschool-aged children. The age of participants was consistently within the 4–6 range, in accordance with the inclusion criteria. Although our primary focus was on the 4–6 age cohort, studies including broader age ranges were included only when data for this specific age group were clearly reported or could be extracted separately, ensuring consistency with the predefined inclusion criteria.

The included studies utilized a broad range of assessments, focusing on multiple facets of narrative skills and associated language and cognitive abilities. Specifically, the literature addressed three primary narrative domains: narrative production (N = 19 studies), narrative comprehension (N = 9 studies), and narrative/story retelling (N = 15 studies). Beyond direct narrative analysis, the reviewed research also incorporated measures of foundational language skills. These included assessments of phonology, expressive and/or receptive vocabulary, morphology, and syntax, in addition to pragmatic abilities. Furthermore, studies often extended their focus to underlying cognitive mechanisms, examining short-term and/or working memory capacities, and both microstructure (e.g., sentence-level complexity and cohesion) and macrostructure (e.g., overall story organization and thematic coherence) within the narratives.

Table 1. Studies of the narrative skills of preschool autistic and non-autistic children.

Study	Country	Method	Results
Studies with non-autistic children
Bitetti et al. (2020)	USA	Participants: 200 Spanish-English non-autistic children, aged between 3;0 to 5;3 years old Measures: Narrative production—story telling through wordless picture books, separately in Spanish and English Coding: narrative macrostructure, narrative comprehension; Language ability tests in Spanish and English to assess language dominance	- Children’s microstructural abilities in each language were significantly related to their macrostructural abilities within that language. - Vocabulary significantly predicted macrostructure, but not microstructure. - No dominance effects in either microstructure or macrostructure in either language.
Bohnacker (2016)	Sweden	Participants: 52 Swedish-English non-autistic children, aged between 5 and 7 years old Measures: Narrative production—story telling in Swedish and English Coding: narrative macrostructure and microstructure, narrative comprehension	- Bilingual preschoolers scored lower than older children in macrostructural narrative skills, as well as in narrative comprehension in both languages, Swedish and English.
Bonifacci et al. (2018)	Italy	Participants: 64 bilingual non-autistic children, aged between 4 and 5 years old (various first languages: Tagalog, Urdu, Arabic, Romanian, Bengali, Spanish, Chinese, Russian, Hindi, Albanian, French and Polish; second language: Italian); 64 monolingual Italian-speaking non-autistic children, aged between 4 and 5 years old Measures: General language ability battery testing, story sequencing, articulation and vocabulary, phonological awareness, morphosyntactic comprehension, phonological memory, and letter knowledge	- Bilingual preschoolers did not differ from their monolingual peers on narrative macrostructural complexity (i.e., inclusion of Goals, Attempts, Outcomes, Mental states, Setting). - Macrostructural parameters were predicted by Mean Length of Utterances in monolinguals, but not in bilinguals.
Castilla-Earls et al. (2015)	USA	Participants: 104 Spanish-speaking non-autistic children, aged between 3 and 5 years old Measures: Narrative production—story retelling Coding: Index of Narrative Complexity	- 4-year-olds: use of characters, actions, & internal responses. - 5-year-olds: characters, settings, initiating events, actions, internal responses, temporal markers, & knowledge of dialogue.
Crawshaw et al. (2020)	Germany	Participants: 46 non-autistic children, aged between 4 and 5 years old (27 boys—19 girls), Sample sizes unbalanced in number and gender Measures: Narrative production—story retelling, focusing on story content, Same story (The mole and the green star)—2 narrative input conditions: (a) illustrated storybook format & (b) non-verbal animated cartoon-video format	The video condition gave significantly more elaborated retellings (in Who, What, and Where components, as compared to When, How, and Why components).
Filiatrault-Veilleux et al. (2016)	Canada	Participants: 121 non-autistic children, 3 age groups (3–4-, 4–5- & 5–6-year-olds), 62 boys Measures: Narrative comprehension assessing causal inference type targeting elements of the story grammar, iPad application	- 3–4-year-olds: emergence of inferential comprehension -interindividual variability. - 4–5-year-olds: ability to infer the problem of the story, the internal response of a character and prediction of the next steps in the story. - 5–6-year-olds: ability to infer the goal of the main character, the attempt to solve the problem and the resolution of the story. - Improvement of the quality of responses with age.
Florit et al. (2014)	Italy	Participants: 156 non-autistic children, aged 4–5.11 years old, 71 girls (Time 1) & 152 of the above children, 68 girls (Time 2, 8 months later), middle to lower-middle socio-economic status Measures: Longitudinal design, The Test for Listening Comprehension-TOR 3–8, Lower-level semantic components, Expressive & Receptive word knowledge, Higher-level semantic components: Inferential tasks, Short-term & working memory tasks	- Lower-level semantic components strongly affected listening comprehension. - Receptive vocabulary was not significantly linked to expressive word knowledge, but both uniquely accounted for listening comprehension. - Higher-level semantic skills: inferential skills did not directly account for later listening comprehension. - Short-term & working memory showed weak associations with listening comprehension.
Friend and Bates (2014)	USA	Participants: 42 non-autistic children aged between 4 and 5 years old & their caregivers Measures: Narrative production (“Frog, where are you?”), a complexity scale with 4 subscales used to code children’s narratives: (a) elements, (b) sequence, (c) syntax & (d) lexicon, executive function (EF) task: attention network test, language sample: a 10 min interaction between child and caregiver during play at home	- No reciprocal relation between EF task and narrative ability: advanced narratives at 4.5 years of age indicative of better performance on EF task at 5, but EF task at 4.5 years did not predict narrative ability at age 5. - The ability to focus attention and resist distraction at 4.5 years conferred benefits in the ability to construct a complex and coherent narrative at 5. - No strong evidence for a convergence of narrative and EF skills at either 4.5 or 5 years of age.
Gagarina (2016)	Germany	Participants: 21 Russian-German non-autistic preschoolers aged 4 years old, 15 non-autistic Russian-German first graders aged 7 years old, 22 Russian-German non-autistic third graders aged 9 years old Measures: Narrative telling in each language via a wordless picture book; assessment of macrostructure and use of mental state terms	- Significant improvement for macrostructure and mental state terms between the preschool and first-grade period. - Larger improvement mainly observed for Russian, which is explained by the Russian curriculum explicitly teaching narrative skills during early literacy training. - Weak cross-language association of narrative macrostructural scores across the two languages.
Hipfner-Boucher et al. (2015)	Canada	Participants: 2 groups of English Language Learners (ELLs: ELL minority language group & ELL English language group), and a group of English monolingual (EL1) preschool children. 25 participants in each group, averaged 56 months of age. Measures: Narrative retell tasks, examining macrostructure and microstructure	- No group differences in the number of utterances or story grammar. - ELL minority language group significantly different from the EL1 and the ELL English language group on all microstructure measures (number of different words, sentence length, and grammaticality). - ELL English language users indistinguishable from EL1 group. - Need to gather detailed linguistic information about the home language environments of ELL children when involving them in language- or literacy-related tasks.
Isbell et al. (2004)	USA	Participants: 38 non-autistic children aged between 3 and 4 years old. Measures: Narrative retelling from storytelling (Group 1) & reading (Group 2), 12-week study, 24 stories from picture books	- The storytelling group performed better on the retelling than the story reading group in (a) formal ending, (b) describing the setting, (c) naming the moral and (d) remembering the characters of the story. - The reading group relied on illustrations to retell the story. - Storytelling helped children’s imaginative recollection of the story. - No significant differences in the verbal factors between the two groups.
Kanellou et al. (2016)	Greece	Participants: 237 non-autistic children aged between 4 to 7 years old, Greek-speaking, from various regions of Greece Measures: Narrative production & retelling skills, 4 wordless illustrated stories (2 stories for retelling and 2 stories for telling/narrative production), Criteria assessed: microstructure/cohesion (conjunction and lexical cohesion) & macrostructure/coherence (story grammar and temporal sequencing of actions and events)	- Regardless of age, better performance in retelling (as compared to telling/narrative production)—a less demanding task. - Gender differences (girls performed better than boys). - Narratives showed better story structure with age: children aged 5–6 years performed better than those aged 4–5 years. - For the 4–5 and 5–6 age groups, the difference in all criteria, in both telling/story production and retelling tasks, was statistically significant except for the criterion of temporal succession. - For the 5–6 and over 6 age groups, the difference in the total retelling scores and in the criterion of content adequacy was statistically significant. - The microstructure could not explain (a) the developmental differences in retelling between the 5–6 and the above 6 age groups and (b) the temporal succession in retelling between the ages of 4–5 and 5–6 years. - Poor scores on the content adequacy criterion (also including the character development component) in all age groups. - Qualitative analysis of narratives showed different narrative levels (labelling, listing, connecting, sequencing and narrating)—for labelling, listing, and connecting the rate seemed to decrease from the 4-5 to 5–6 age groups, while for narrating the rate seemed to gradually increase.
Karlsen et al. (2021)	Norway	Participants: 215 non-autistic children (Mage = 4.3 years), 109 boys & 106 girls Measures: Longitudinal study—initial assessment at age 4 and the second assessment at age 7; At age 4: Story retelling (Norwegian version of the Renfrew Bus Story Test), receptive vocabulary (Norwegian version of the British Picture Vocabulary Scale), sentence memory (Sentence Memory subtest of the Norwegian language screening test ‘Language 6–16’), and grammatical knowledge (Norwegian version of the Grammatic Closure subtest from the Illinois Test of Psycholinguistic Abilities). At age 7: a listening comprehension test (Norwegian version of the Neale Analysis of Reading Ability).	- Narrative skills & core language skills were related, but best conceptualized as different skills. - At age 4: Associations found between narrative retelling & core language construct (vocabulary, grammar and verbal memory). - Moderate associations were observed between age 4 and age 7 in listening comprehension and core language skills (vocabulary, grammar and verbal memory). - With age, syntactic complexity increased in retellings as compared to self-generated stories, and the use of prompts (verbal, picture) appeared to improve coherence.
Kendeou et al. (2008)	Canada & USA	Participants: 113 non-autistic children aged between 4 and 6 years old (Time 1), and 108 of them between 6 and 8 years old (Time 2) Measures: Longitudinal study, Time 2: 2 years later, two narratives for narrative comprehension assessment (“The Cat’s Purr” on audiotape & “The Autumn Leaves”, audio-visual episode from television series), Open-ended comprehension questions followed each narrative, Evaluation of phonological awareness, letter and word identification (Woodcock Reading Mastery Test) & vocabulary (Peabody Picture Vocabulary Test—III) [Time 1], two different narratives for narrative comprehension (“The Rabbit and the Moon” on audiotape & “Granny’s New Glasses” audio-visual episode from television series), Open-ended comprehension questions followed each narrative [Time 2], Same tests of basic language skills used at Time 1 were used at Time 2.	- Children as young as 4 years old engaged in inferential processes that were very similar to those that older children and adults used when reading. - Dissociation between children’s inferential comprehension and their basic language skills. - Clear developmental pattern with goal inferences contributing to comprehension as early as 4 years of age and continuing strongly at age 6. - Causal inferences (antecedents and consequences) significantly contributed to narrative comprehension at age of 6. - Character actions and emotions did not contribute significantly to comprehension at ages 4 and 6. - Results indicate how children perform in narrative comprehension when prompted but not how they would perform in spontaneous production. - Television viewing or aural listening activities boosted the development of inferential skills that could later transfer to reading.
Khan et al. (2016)	USA	Participants: 386 non-autistic children aged between 3 and 6 years old Measures: Narrative retelling: two randomly selected stories read to each child, Wordless picture books developed for this study (Wolf Cleans His Bedroom, Tiger Gets Ready for Bed, Raccoon Makes Lemonade, and Rabbit Goes for a Bike Ride)—all stories had an identical format	- Story-structure ability increased significantly across age groups. - 4-year-olds: (a) difficulty encoding goal-based or purposeful actions, but (b) transition from producing incomplete episodes in their narratives to “minimally complete” episodes, containing initiating events, attempts, and consequences. - 5-year-olds: (a) related more purposeful attempts in their narratives, demonstrating an emerging knowledge of goals and plans, but (b) exhibited difficulty in identifying superordinate goals (i.e., larger goals that motivate the formation of subsequent goals in the story) in their narratives (only 25% of 5-year-olds succeeded in this task) - Significant improvement of story-structure ability between 3–4 and 4–5 years. - Mean story-structure abilities across ages 5 and 6 years were not significantly different (most children included all story components in their narratives). - Episode-related narrative abilities (subgoals, subproblems, subsolutions, subresolutions) improved with age (39% of 4-year-olds & 70% of 5-year-olds produced all four episode-related components in their narratives).
Lepola et al. (2012)	Finland	Participants: 130 Finnish-speaking non-autistic children, 62 boys and 68 girls, Time 1 (age 4), Time 2 (age 5), Time 3 (age 6) Measures: 2-year longitudinal study, Early narrative listening comprehension, Language skills (receptive vocabulary, sentence memory, & phonological awareness)	- Inference-making skills contributed both directly and indirectly, through vocabulary knowledge, to later narrative listening comprehension. - Narrative listening comprehension at age 4 contributed to listening comprehension at age 6. - Vocabulary did not predict later narrative listening comprehension. - Sentence memory at age 4 did not predict narrative listening comprehension at age 6. - Phonological awareness at age 4 was not related to narrative listening comprehension, nor to inference skills. - Inference skills are already in place at the age of 4 years and make a unique contribution to the development of listening comprehension and vocabulary. - Making inferences after picture book viewing at age 5, better at narrative retelling and comprehension at age 6. - Narrative listening comprehension at age 4 was a significant predictor of inference skills at age 5, which, in turn, directly predicted narrative listening comprehension at age 6. This suggests a reciprocal relationship between listening comprehension and inference skills through development.
Lindgren (2019)	Sweden	Participants: 17 non-autistic children aged between 4 and 7 years old. Longitudinal design: assessment at age 4;4, 5;10 and 7;4 Measures: Comprehension and production of narrative macrostructure in Swedish monolingual children	- Steep development from age 4;4 to 5;10 in both comprehension and production of macrostructure. - Children reach a plateau around age 6—only some further development in comprehension to age 7;4. - Consistent differences between comprehension and production (higher scores in comprehension) and between stories across time points.
Mäkinen et al. (2014)	Finland	Participants: 172 non-autistic children aged between 4 and 8 years old Measures: Picture-elicited narrations which were analysed for productivity, syntactic complexity, referential cohesion & event content.	- Significant differences between 4- and 5-year-olds in productivity and event content. - Significant differences between 5- and 6-year-olds in referential cohesion. - The number of different word tokens was useful in explaining the event content, whereas the number of communication units, or episodes, did not.
Méndez et al. (2018)	USA	Participants: 14 Spanish-English bilingual non-autistic children aged between 4 and 6 years Measures: Picture-elicited narrative retellings in both languages, which were analysed for microstructure (mean length of utterance in words, lexical diversity, use of subordinate clauses), and macrostructure (introduction, conflict resolution, conclusion), as well as use of mental state terms	- There were significant relations between the two languages, English and Spanish, for macrostructure but not microstructure. - Development of microstructure was independent across the two languages.
Nicolopoulou and Richner (2007)	USA	Participants: 30 non-autistic children, 3 age groups: early 3-year-olds, early 4-year-olds & late 4-year-olds Measures: Storytelling & story-acting, children voluntarily dictated self-initiated stories to teachers and peers every day, as a self-chosen free-play activity, and acted them out later with their friends. Children chose their own characters, subjects, and plots. The stories were neither solicited by adults nor channelled by props, story stems, or suggested topics.	- Continuous shift from lower to more advanced levels of character representation with age. - Actors (represented purely by actions and not being further described) were surpassed by agents (whose internal mental states need to be inferred) in the stories of 4-year-olds, and almost matched by persons (having ToM or else mental representational capacities) in the stories of 5-year-olds. - Girls’ stories included higher levels of character representation than boys in all age groups. - Between ages 4 and 5, a major developmental advance, both quantitative and qualitative, was noticed in character representation.
Pistav Akmese and Kanmaz (2021)	Turkey	Participants: 100 non-autistic children, 50 in the age group of 4–5 years, 50 children in the age group of 5–6 years Measures: Descriptive survey, Receptive and expressive verbal language skills, Narrative production, Mean Length of Utterance (MLU), Number of Different Words (NDW), Total Number of Words (TNW)	- Significant differences between the two groups, regarding the receptive and expressive language, and between NDW and TNW scores. - Positive associations between children’s age, receptive and expressive language, the MLU, NDW, and TNW. - Statistically significant differences in favour of the girls in MLU and NDW measures in both age groups. - Mothers’ education did not have a statistically significant effect in either group regarding receptive & expressive language, MLU, TNW, and TNW. - Longer stories produced by children between 5–6 years than by 4–5-year-old children. - Developmental differences in NDW and TNW between the two age groups.
Ralli et al. (2021)	Greece	Participants: 237 non-autistic children, 98 boys—139 girls, Greek-speaking, Middle class from various parts of Greece, 3 age groups: (a) 79 children (33.3%) 4–5 year old (Mage = 4.5), (b) 58 children (24.5%), 5–6 year old (Mage = 5.5), (c) 100 children (42%), 6–7 year old (Mage = 6.5) Measures: Story Retelling, Oral Language Measures: Vocabulary Knowledge Tasks, Phonological Awareness & Processing Tasks, Morphological Awareness Tasks, Pragmatics Task	- Statistically significant differences in performance in all oral language tasks across age groups: older children performed better. - Story retelling microstructure and macrostructure improved with age. - Vocabulary skills (listening comprehension, receptive vocabulary, naming, and word definition) predicted story retelling through morphological skills for the 4–5-year-old children. - Vocabulary skills associated with children’s story retelling performance. - 4–5-year-olds: both vocabulary and morphological skills contributed to story retelling. - 5–6-year-olds: vocabulary, morphological, and phonological skills contributed to story retelling. - 6–7-year-olds: vocabulary, morphological, phonological, and pragmatic skills contributed to story retelling.
Rezzonico et al. (2016)	Canada	Participants: 23 English-Cantonese bilingual non-autistic children aged 4 years old; 24 English-Cantonese bilingual non-autistic children aged 5 years old Measures: Narrative story telling (a wordless picture story) in both languages, English and Cantonese; assessing macrostructure and microstructure (sentence length, lexical diversity, morphosyntactic quality); Language ability assessment: expressive and receptive vocabulary	- Higher macrostructural scores in English than in Cantonese across both groups. - The 5-year-old group showed higher macrostructural performance scores and better microstructure (mean length of utterance in words, morphosyntactic quality) than the 4-year-old group in both languages. - Cantonese macrostructure and microstructure predicted English macrostructure and microstructure, respectively.
Rodina (2017)	Norway	Participants: 16 Norwegian (dominant language)-Russian (minority language) bilingual non-autistic children aged 4 years old, 16 Norwegian monolingual non-autistic children aged 4 years old, 16 Russian monolingual non-autistic children aged 4 years old Measures: Narrative production separately in each language, i.e., Norwegian and Russian (assessment of both microstructure and macrostructure), narrative comprehension	- No significant differences found between bilinguals and monolinguals in Norwegian, which was the dominant language. - Monolinguals differed from bilinguals in Russian (minority language) in both microstructure and macrostructure, suggesting that narrative production is sensitive to the amount of exposure to each language. - Narrative microstructure and macrostructure equally developed in each language in the bilingual group.
Safwat et al. (2013)	Egypt	Participants: 60 non-autistic children, 29 boys & 31 girls—4 groups: Group A: 2–2.11-year-olds, Group B: 3–3.11-year-olds, Group C: 4–4.11-year-olds, Group D: 5–5.11-year-olds Measures: Narrative story retelling (a wordless picture story, specially designed for this study)—assessing macrostructure & microstructure, Language ability assessment: Standardized Arabic language test, Articulation, General & neurological examination, & audiological evaluation	Results for Group C & D only: - Children were able to produce a proper story frame, providing a topic, an introduction, the story events using complex verbs, use of references, and finding out the main problem and its solution. - At 4 years; significant increase in the use of independent connectives, but at 5 years an unexpected significant decrease in the use of independent connectives. - Group D exhibited greater grammatical (e.g., past and future tenses) & syntactic complexity. - Younger children used more irregular plurals than older children. - Narrative productivity and MLU showed increase with age. - Decrease in diversity in the vocabulary used across all age groups. - No significant differences between boys and girls in narrative skills.
Silva et al. (2014)	Chile	Participants: 60 non-autistic children, aged between 5 and 6 years old, from 3 public and private Chilean schools Measures: Narrative skills: (a) picture-walk (look the story from a wordless picture-book: ‘A boy, a dog, and a frog’), (b) narrative comprehension questions (10 questions about the characters, setting, mental states, such as feelings & thoughts, the initiating event, the problem and its resolution, and predictions targeting memory and story comprehension, (c) book storytelling	- Cultural context did not affect performance in constructing fictional stories. - School type affected performance, with public schools scoring higher in all tasks than private schools. - Questions about the key story elements improved children’s coherence and macrostructure of narratives. - The type of questioning task promoted thinking about relations between the events and the construction of coherent stories. - Storytelling after answering questions was of higher structural quality.
Souza and Cáceres-Assenço (2021)	Brazil	Participants: 39 non-autistic children aged between 4 and 6 years old, 48.7% boys, 3 age groups: (a) 4-year-olds (25.6%), (b) 5-year-olds (33.3%), (c) 6-year-olds (41%), family variables: mothers with higher education (38.5%) or a master’s degree (38.5%) Measures: Language assessment: vocabulary, phonology, fluency, speech, phonological short-term memory, Narrative production (book “Frog, where are you?”)	- 92.3% of children had age-appropriate expressive vocabulary (usual verbal designation—(UVD). - The semantic fields with the highest UVD were “animals”, “shapes and colours”, “toys and musical instruments”, “transportation”, and “professions”. - The predominant type of narrative was causal (when causality between scenes was established to demonstrate that one event occurred as a result of another), followed by intentional (when mental states or feelings of the characters were expressed), as children grew older. - No correlations found between performance in vocabulary and narrative tasks. - The socio-cultural background, the socioeconomic status of the family, and a facilitating environment significantly affected children’s performance in language and narrative tasks.
Thordardottir et al. (2010)	Canada	Participants: 78 monolingual speakers of Quebec French—three age groups (4;6, 5;0 and 5;6 years) Measures: Tests of language knowledge and linguistic processing, addressing vocabulary, morphosyntax, syntax, narrative structure, nonword repetition, sentence imitation, rapid automatized naming, following directions, and short-term memory	- Systematic increase with age for most of the measures. - Performance in language assessment of French-speaking 5-year-olds comparable to the normal range of non-autistic monolingual French-speaking children.
Tompkins et al. (2013)	USA	Participants: 42 non-autistic children aged between 3.8 and 5.8 years old, Mother-child participation, Medium to high socio-economic status Measures: Vocabulary: Receptive (PPVT-4: Peabody Picture Vocabulary Test-4th edition) & Expressive (EVT-2: The Expressive Vocabulary Test-2nd edition, Narrative comprehension: 5 questions assessed literal understanding of the story (initiating event, overarching goal, characters, setting, and outcome resolution) & 5 questions assessed children’s ability to make inferences (dialogue, causal inference, feelings, predicting what happens next, and the theme), Narrative production from a wordless picture book (Frog Goes to Dinner—Mayer, 1992)	- Children exhibited competence with inferences about concrete activities, character states, dialogue, and emotions—Relatively fewer inferences regarding goals, actions, objects, causal antecedents and consequences, or places. - Only certain inference categories, e.g., goals, actions, and character states, were significantly related to story comprehension. - Character states, goals and actions were particularly important in predicting children’s narrative comprehension, after controlling for children’s age and vocabulary. - Expressive vocabulary was a better predictor of narrative comprehension than receptive vocabulary.
Uccelli and Páez (2007)	USA	Participants: 24 bilingual non-autistic children were assessed in English and Spanish at the end of kindergarten (Time 1) and at the end of first grade (Time 2) Measures: Narrative production in each language, English and Spanish; narrative microstructure (lexical diversity) and macrostructure (story structure)	- Significant improvement in both microstructure and macrostructure in English from preschool age to first grade. - Significant gains in Spanish observed only for macrostructure. - Vocabulary has significantly predicted narrative macrostructure scores in both languages.
Veraksa et al. (2020)	Russia	Participants: 269 non-autistic children, aged between 5 and 6 years old (Mage = 5.6 years; Sd = 0.48), 133 boys & 136 girls Measures: NEPSY II: (a) Verbal working memory (sentence repetition), (b) Visual working memory (memory of designs), Story retelling (“The Jackdaw and the Pigeons” by L.N. Tolstoy)—in the absence of a story, up to 3 reading rounds, Narrative production (create a story on a series of pictures: (i) “The Tower”, (ii) “The Cat and the Dog”, (iii) “The Broken Cup”)—children assembled the pictures in alogical sequence of events, and built a story, Text’s macrostructure: semantic completeness & adequacy, story programming, speech pace, number of words and paragrammatisms (i.e., number of lexical and grammatical mistakes), narrative type (complete, i.e., the child includes full amount of relevant information; simplified, i.e., the child provides partial, yet, relevant information; or distorted, i.e., the child provides inaccurate or irrelevant information), and narrative structure, Text’s microstructure: lexical and grammatical accuracy	- Visual and verbal working memory related to semantic completeness of narration, programming, number of words and sentences, and narrative structure and type. - Only verbal working memory correlated with lexical and grammatical story design and semantic adequacy. - Visual working memory was associated primarily with the semantic breadth of the utterances. - Verbal working memory was related to pictured story comprehension. - Macrostructure associated more strongly with verbal than visual working memory capacity.
Westerveld et al. (2012)	New Zealand	Participants: 73 non-autistic children, aged between 4 and 4.11 years old, 43 girls (59%) & 29 boys (41%), low-mid to high socio-economic backgrounds, NZ English Measures: Longitudinal design (Time 1: age 4, Time 2: age 5), Narrative ability at age 4 and age 5, (a) Listened an audio-recording of unfamiliar story, while looking at the pictures of the story book on a computer screen, (b) Oral narrative comprehension (ONC): 8 factual and inferential questions (targeting the main idea, sequence, story details, and predictive information) followed the first exposure to the story, (c) Story retelling, without the use of pictures, following the second exposure to the story: examination of oral narrative quality (ONQ) with (a) macrostructure (6 text structure elements: introduction, main character/s, supporting character/s, conflict, resolution, conclusion, holistic coherence ‘theme’) & (b) microstructure (semantic diversity: Number of Different Words (NDW), verbal productivity: total number of utterances (UTT), grammatical ability: mean length of utterance in morphemes (MLU-M), and grammatical competency: grammatical accuracy (GA) in percentiles of grammatically correct utterances, Receptive vocabulary skills at age 4	- Meaningful story retelling samples in most 4-year-old children. - Normal distributions at 4 years on ONC, ONQ, and MLU-M. - Significant improvement on all measures from 4 to 5 years of age. - At age 4: significant moderate correlations between ONC, ONQ, and receptive vocabulary. - At age 5: significant moderate correlations between ONQ, NDW, MLU-M, and ONC. - The ONC measure was more suitable for the 4-year-olds; a ceiling effect observed at 5 years of age.
Yang et al. (2022)	Australia	Participants: 25 Mandarin–English speaking children between 3 and 6 years old Measures: remote online story retelling in Mandarin and English; microstructural elements total number of words, number of different words, total number of C- and T-units, and mean length of utterance in words; macrostructural elements (Introduction, Theme, Main Character, Supporting Character(s), Conflict, Coherence, Resolution and Conclusion); weekly diary detailing daily Mandarin, English, and mixed language input and output per hour	- Similar narrative micro- and macrostructural across the two languages, Mandarin and English. - Cross-linguistic differences found in microstructure; more nominal phrases in English (vs. Mandarin); more modifier phrases in Mandarin (vs. English). - Age was significantly positively correlated with macrostructure across both languages. - Language experience had no significant effect on either narrative macrostructure or microstructure.
Study with autistic children
Westerveld and Roberts (2017)	Australia	Participants: 29 autistic children aged between 4 and 5.9 years old, Ability to speak in short sentences, Age equivalent score of 36 months on the expressive communication subscale of the VABS-II Measures: Level of nonverbal ability: Visual Reception and Fine Motor skills (Mullen Scales of Early Learning—MSEL), Communication skills (Vineland Adaptive Behavior Scales-II—VABS-II), Receptive vocabulary (Peabody Picture Vocabulary Test 4—PPVT-4), Narrative comprehension: children listened twice to an audio-recording (“Ana gets Lost”), while looking at the pictures from the book on a computer screen—8 factual and inferential questions, Narrative retelling without the use of pictures—only neutral prompts to start and/or continue retelling the story, Micro-and macrostructure	- Significant difficulties in narrative comprehension, intelligibility, and grammatical accuracy. - Children were better able to answer factual questions than inferential questions—very few children answered inferential questions. - Significant positive correlations among vocabulary, semantic measures, and narrative comprehension. - Microstructure: (a) relative strengths in productivity, semantic diversity, and grammatical complexity; (b) relative weaknesses in grammatical accuracy. - Macrostructure: great challenges in producing a coherent narrative—Evidence supports the weak central coherence theory, which holds that autistic children prioritize details over the whole. - Receptive vocabulary significantly correlated with oral narrative production (number of different words) and narrative comprehension.
Studies with non-autistic and autistic children
Gabig (2008)	USA	Participants: 10 non-autistic children and 15 autistic children (13 boys) aged between 5 and 7.11 years old Measures: Nonverbal intelligence, Language skills, Story retelling task (The Renfrew Bus Story) & Verbal working memory tasks	- Non-autistic children: (a) significantly better scores on verbal working memory, (b) significantly more story propositional units recalled, (c) significant between-group differences in story retelling, which requires simultaneous processing and construction of complex linguistic and cognitive representations. - Autistic children: (a) verbal working memory deficits, (b) recalled significantly fewer story propositional units, (c) receptive vocabulary linked with sentence imitation and story recall, but not with nonword repetition or digits span, (d) significant difficulties in tasks requiring processing and parallel construction of complex linguistic and cognitive representations.
Govindarajan and Paradis (2022)	Canada	Participants: 9 bilingual autistic children aged 5;4 to 9;1 (diverse first languages-English as second language), 10 bilingual non-autistic children aged 5;4 to 9;1 (diverse first languages-English as second language), 10 bilingual children with DLD aged 5;4 to 9;1 (diverse first languages-English as second language) Measures: Narrative telling through a pictured story; assessment of narrative microstructure (utterance length, syntactic complexity, lexical diversity, story length) and narrative macrostructure (character introductions, settings, initiating events, attempts, outcomes, use of mental state terms)	- The autistic bilingual group scored lower than the non-autistic bilingual group in both microstructure and macrostructure; no differences observed between bilingual autistic children and bilingual children with DLD. - The bilingual autistic group also scored significantly lower than the rest of the groups in macrostructural elements requiring perspective-taking, such as pronoun use. - The non-autistic bilingual group exhibited higher macrostructural scores and used longer utterances and more complex syntax than the autistic bilingual group.
He et al. (2025)	China	Participants: 19 non-autistic children (Mage = 5.14 years) and 19 autistic children (Mage = 5.57 years) (age range: 3.5–7.2 years old), Mandarin speaking Measures: Narrative Assessment Protocol–2 (NAP-2): assessment of microstructural (linguistic & syntactic complexity) & macrostructural (story organization and coherence) features, Story retelling (wordless picture book ‘Raccoon Makes Lemonade’), Story length, Linguistic performance (mean length of utterance, number of different words, vocabulary diversity)	- No significant between-group differences in story length or linguistic complexity (microstructure). - Autistic group: Significantly lower scores than non-autistic peers in macrostructural measures: total NAP-2 score, story grammar, and storytelling conventions. - Nonverbal and receptive vocabulary influenced groups’ total score but not story grammar or conventions.
Kenan et al. (2019)	Israel	Participants: 24 non-autistic children and 24 autistic children aged between 4.10 and 7 years old, only boys, IQ > 80 Measures: Autism Diagnostic Observation Scales (ADOS), Wechsler Preschool and Primary Scale of Intelligence (3rd ed), Picture book, Language measures: (a) total length of story and syntactic complexity, (b) semantic-pragmatic measures: details of history, key ideas, evaluative comments and irrelevant text	- Narratives of non-autistic children included more central ideas, environments, characters, actions, and objects. - No between-group difference in the number of comments irrelevant to the story. - Narratives of autistic children included fewer central ideas, environments, characters and actions, but no objects, compared to the narratives of non-autistic children. - Semantic difficulties & heterogeneity within the autistic group. - Difficulties in autistic children explained by cognitive challenges in the Weak Central Coherence Account. - Difficulties in ToM and abstract concepts in autistic children.
Nuske and Bavin (2011)	Australia	Participants: 14 non-autistic children aged between 4.2 and 5.4 years old (7 boys & 7 girls) and 14 autistic children aged between 4.6 and 7.11 years old (13 boys & 1 girl) Measures: two narrative comprehension tasks: short stories were read followed by comprehension questions, Understanding Spoken Paragraphs & Inferential Processing, Receptive Vocabulary and Picture Completion sub-tests from the WPPSI-3	- Non-autistic children: Extracting the main idea of the narratives significantly correlated with performance on all other comprehension tasks. - Autistic children: (a) Extracted the main idea of stories with clues from the title, (b) did not spontaneously make inferences regarding event scripts, (c) unable to automatically integrate information and draw conclusions from scripts.

details the included studies involving the narrative skills of preschool autistic or non-autistic children, listed alphabetically by author.

3.2. Narrative Skills in Non-Autistic Children

Research focusing on narrative skills in non-autistic development demonstrates a clear and progressive developmental trajectory across preschool years across both monolingual and bilingual children, marked by increasing complexity in both microstructure (linguistic features) and macrostructure (story organization). The results provide interesting findings in relation to the developmental progression of narrative components. More specifically, there is a strong indication that children tend to exhibit a chronological advancement in the elements they include in their narratives. For instance, 4-year-olds begin using characters, actions, and internal responses, while 5-year-olds incorporate more sophisticated components like settings, initiating events, temporal markers, and dialogue knowledge (Castilla-Earls et al., 2015; Bohnacker, 2016; Gagarina, 2016; Uccelli & Páez, 2007; Yang et al., 2022). A distinct improvement in story-structure ability is noted between 3 to 4, and 4 to 5 years of age, with 4-year-olds moving from incomplete to “minimally complete” episodes, and 5-year-olds relating more purposeful attempts, though they may still struggle with identifying superordinate goals (Khan et al., 2016). Older children (5 to 6 years and up) show better story structure and an increase in the rate of higher-level narrative performance (Kanellou et al., 2016; Bohnacker, 2016; Gagarina, 2016; Rezzonico et al., 2016; Uccelli & Páez, 2007; Yang et al., 2022). The ability to infer story elements also progresses with age, starting with inferential comprehension in 3- to 4-year-olds, advancing to inferring the problem and predicting steps in 4- to 5-year-olds, and finally inferring the main character’s goals, attempts, and resolution in 5- to 6-year-olds (Filiatrault-Veilleux et al., 2016). Notably, the tendency for bilingual non-autistic children is to exhibit greater improvement in narrative macrostructure as compared to microstructure, which may be due to the fact that macrostructure relies less on narrow language abilities, including vocabulary and morphosyntax, which have been both shown to be vulnerable in bilingual child populations (Bohnacker, 2016; Gagarina, 2016; Uccelli & Páez, 2007). However, contradictory evidence also exists showing that bilingual non-autistic children’s narrative micro- and macrostructural abilities show similar growth rates across the two languages (Yang et al., 2022).

Additionally, interesting findings are revealed in relation to children’s linguistic and cohesive narrative development. Linguistic measures of narratives, such as productivity and event content, show significant differences between 4- and 5-year-olds, while referential cohesion significantly improves between 5- and 6-year-olds (Mäkinen et al., 2018). Measures like Mean Length of Utterance (MLU), Total Number of Words (TNW), and Number of Different Words (NDW) also show positive associations with age, indicating longer and more diverse stories in older children (Safwat et al., 2013; Rezzonico et al., 2016). Syntactic complexity increases with age, particularly in retellings (Karlsen et al., 2021; Safwat et al., 2013; Rezzonico et al., 2016). In regard to children’s comprehension and inferential skills, the reviewed findings suggest that inferential processes similar to those used by older individuals are evident in children as young as 4 years old, with goal inferences being present early, and causal inferences (antecedents and consequences) developing strongly by the age of 6 (Kendeou et al., 2008). Inference-making skills contribute significantly to later listening narrative comprehension, and a reciprocal relationship emerges between listening comprehension and inference skills over time (Lepola et al., 2012). Expressive vocabulary is noted as a better predictor of narrative comprehension than receptive vocabulary (Mäkinen et al., 2018; Uccelli & Páez, 2007).

Another component of narrative ability identified by the reviewed studies relates to the influence of the task’s characteristics and the context of narration. Narrative retelling is generally found to be a less demanding task than telling/narrative production, with children performing better in retelling (Kanellou et al., 2016). The method of story presentation impacts the retelling quality, with video conditions leading to more elaborated retellings than illustrated storybook formats (Crawshaw et al., 2020), and storytelling yielding better recollection of setting, moral, and characters than story reading (Isbell et al., 2004). Questioning tasks about key story elements or conducting storytelling after question-answering can improve the structural quality, coherence, and macrostructure of the narrative (Silva et al., 2014).

Interestingly, the reviewed findings suggest that narrative skills are related to, yet distinct from, core language skills like vocabulary, grammar, and verbal memory (Karlsen et al., 2021). The partial dissociation between narrative skills and structural language abilities receives support from studies with bilingual non-autistic children and their macrostructural abilities which show higher improvement than microstructure, possibly because macrostructure refers to the links between abstract representations of events that are not completely tied to bilinguals’ otherwise weak language abilities (Bohnacker, 2016; Gagarina, 2016; Uccelli & Páez, 2007). Vocabulary and morphological skills consistently predict story retelling performance across different age groups, with phonological and pragmatic skills becoming additional contributors mainly for older children (Ralli et al., 2021). Verbal and visual working memory are related to the semantic completeness and structure of narration, with verbal working memory being more strongly associated with macrostructure and grammatical complexity (Veraksa et al., 2020). There is mixed evidence for a reciprocal relation between executive functions and narrative ability, although attention skills may enhance complex narrative construction in older children (Friend & Bates, 2014). Socio-cultural background and a facilitating environment (e.g., consistent exposure to rich language input, frequent storybook reading, responsive conversational partners, curricula explicitly teaching narrative skills during early literacy training, and opportunities for shared narrative creation) are also noted to significantly affect language and narrative performance (Souza & Cáceres-Assenço, 2021; Gagarina, 2016). Finally, several studies focused on gender differences and noted that girls performed better than boys in narrative skills, including character representation, retelling and telling, and language measures, such as MLU and NDW (Pistav Akmese & Kanmaz, 2021; Kanellou et al., 2016; Nicolopoulou & Richner, 2007). Other research, however, found no significant gender differences (Safwat et al., 2013).

3.3. Narrative Skills in Autism

Studies focusing solely on autistic children or comparing them to their non-autistic peers highlight specific challenges in narrative abilities, particularly in areas related to global meaning and coherence. More specifically, they highlight several macrostructure challenges and point out that autistic children consistently show significantly lower scores in macrostructural measures, such as total story grammar and storytelling conventions compared to non-autistic children, despite potentially showing no significant difference in microstructural measures, like story length or linguistic complexity (He et al., 2025). This finding is consistent with the local processing bias approach (Happé & Booth, 2008), suggesting that the observed pattern is a direct result of a cognitive prioritization of specific, local details over the synthesis of the wider picture. This reflects an enhanced local coherence rather than a difficulty in global integration (Westerveld & Roberts, 2017; Kenan et al., 2019). The narratives of autistic children include fewer central ideas, environments, characters, and actions than those of their non-autistic peers. Furthermore, the prominence of autistic characteristics negatively correlates with the number of central ideas and characters mentioned (Kenan et al., 2019). Evidence from bilingual autistic children is slightly different, pointing towards generalized weakness in both microstructure and macrostructure as compared to their bilingual non-autistic peers, further suggesting that bilingualism incurs an extra burden on autistic children’s use of vocabulary and morphosyntax while narrating a story, besides pragmatic difficulties (Yang et al., 2022). Interestingly, the only study with bilingual autistic preschoolers (Yang et al., 2022) also shows that the macrostructural element most vulnerable to autism in macrostructure was the use of pronouns, since a great portion of referentially ambiguous pronominal elements was a characteristic trait in the bilingual autistic preschoolers’ narrative production performance only (Yang et al., 2022). Since pronoun use lies at the interfaces between vocabulary, morphosyntax and pragmatics, and it represents a structure that is sensitive to one’s perspective-taking skills, the particular finding in Yang et al. (2022) implies that the increased use of ambiguous pronouns in the bilingual autistic children as compared to their non-autistic peers may stem from the former group’s ToM weaknesses.

Additionally, the reviewed findings identify comprehension and inferential difficulties in young autistic children. More specifically, they support that autistic children demonstrate significant difficulties in narrative comprehension, particularly in answering inferential questions compared to factual questions (Westerveld & Roberts, 2017). They may be unable to spontaneously make inferences regarding event scripts or automatically integrate information and draw conclusions, even if they can extract the main idea through clues provided by the adult examiner (Nuske & Bavin, 2011). Finally, in terms of microstructure in narration, the reviewed studies indicate that while autistic children may show relative strengths in productivity, semantic diversity, and grammatical complexity, they exhibit relative weaknesses in grammatical accuracy (Westerveld & Roberts, 2017). Furthermore, they often demonstrate lower performance in verbal working memory and recall significantly fewer story propositional units than their non-autistic peers, suggesting challenges in processing and constructing complex linguistic and cognitive representations (Gabig, 2008). Nonverbal ability and receptive vocabulary influence the total narrative score but not specifically story grammar or conventions (He et al., 2025). Positive correlations were also found among vocabulary, semantic measures, and narrative comprehension in autistic children (Westerveld & Roberts, 2017).

3.4. Summary of the Reviewed Findings

The reviewed studies on non-autistic development reveal a clear, age-related progression in narrative skills, moving from basic components in 4-year-olds to complex story structures, inferential comprehension, and advanced linguistic coherence by the age of 6. Narrative ability is intertwined with core language skills (vocabulary, morphology), though less strongly in bilingual non-autistic children as compared to their monolingual peers, and cognitive factors (working memory, attention), while there is also evidence that the presentation format (e.g., video over book) and task type (retelling over production) influence performance. Girls often show better performance than boys. In contrast, autistic children demonstrate significant difficulties in narrative macrostructure and coherence (story grammar, central ideas), consistent with the local processing bias account. They experience particular challenges with inferential comprehension and demonstrate lower performance in verbal working memory, even when their microstructural (linguistic complexity) or basic vocabulary skills are relatively preserved.

4. Discussion

The present scoping review of the literature has focused on research regarding the narrative skills of monolingual preschool non-autistic and autistic children between 4 and 6 years of age. Specifically, the review aimed to identify the individual and contextual factors influencing narrative abilities in both groups, and compare the narrative skills in production and comprehension of non-autistic preschool children to those of autistic children. Another purpose of this study was to discover the gaps in the existing literature and recommend directions for future research.

4.1. Narrative Production in Non-Autistic and Autistic Children

The findings regarding the spontaneous production of narratives in non-autistic preschoolers closely align with established developmental milestones. Research highlights that by the age of 4 years, children include core narrative elements such as characters, actions, and internal responses in their storytelling (Friend & Bates, 2014; Bohnacker, 2016; Gagarina, 2016; Uccelli & Páez, 2007). This accomplishment expands significantly by the age of 5 years, incorporating more temporal markers, knowledge of dialogue, and a broader range of semantic components (setting, initiating events, actions) (Kanellou et al., 2016). Thus, around the age of 5 years, children start to produce stories with more details, presenting an understanding of how story events relate to one another and generating multi-episode sequences in their narratives (Muñoz et al., 2003).

Qualitative analysis of spontaneously produced narratives (Mäkinen et al., 2018) revealed a clear developmental trajectory regarding distinct narrative levels, moving from simple labeling and listing (decreasing in frequency from ages 4–5) to more sophisticated sequencing and narrating. Mäkinen et al. (2018) also focused on the development of narrative structure and its relationship with narrative productivity based on picture-elicited narrations. Significant differences were found between 4- and 5-year-olds in productivity and event content, and between 5- and 6-year-olds in referential cohesion. The period between 5 and 6 years has been described as the ranking period, as children produced stories in a logical, hierarchical order and in accordance with cause–effect relationships (Pistav Akmese & Kanmaz, 2021; Rodina, 2017). However, despite improvements in structure, consistent challenges across the preschool years were observed in content adequacy, a measure of how well children’s narratives include all relevant story elements and causal links, as well as in the development of character depth (Mäkinen et al., 2018; Pistav Akmese & Kanmaz, 2021). Therefore, there are researchers who argue that although preschool children have developed certain knowledge and capacities for story production before the age of 6 years, their narrative skills are weak, as they still have difficulties in linking events and characters, and are in need of support to produce a structured, cohesive, and consistent narrative story (Boudreau, 2008), with cross- and within-episodes connections (Hamilton et al., 2021). Thus, although preschool children gradually become better at organizing their story production, the content and depth of their narratives remain underdeveloped until later childhood.

Spontaneous narrative production in preschool autistic children is marked by significant qualitative limitations in macrostructure, despite potentially adequate linguistic complexity at the sentence level (Kenan et al., 2019). While microstructure measures (e.g., grammatical complexity and semantic diversity) showed relative strengths compared to other language domains, on macrostructural measures, including total story grammar and storytelling conventions, autistic preschoolers’ performance was consistently poor (Gabig, 2008). This discrepancy—possessing the linguistic tools (microstructure) but displaying reduced global organization (macrostructure)—is effectively explained by the local processing bias (Happé & Booth, 2008) which suggests that the autistic processing style involves a superior ability and preference for focusing on and encoding local details, which inadvertently leads to a cognitive prioritization of these microstructural elements over the synthesis of a global plot structure (Gabig, 2008). This trade-off is evident in the narrative output, where autistic children’s stories characteristically included fewer central ideas, settings, and characters compared to their non-autistic peers, reflecting an organizational choice rather than a failure to connect events. Reduced content density in narratives has been found to correlate negatively with the prominence of autistic characteristics, indicating that more pronounced autistic features are associated with sparser narrative content (Gabig, 2008). Finally, while productivity measures (MLU, NDW) increased with age, vocabulary diversity decreased across all age groups (Gabig, 2008), implying that lexical limitations may confound richness in narrative content.

The findings of the present review regarding macrostructural development can also be interpreted in light of longitudinal evidence from non-autistic populations. For instance, Lindgren (2019) documented rapid growth in narrative macrostructure between the ages of four and six, followed by a plateau in later years. This pattern suggests that the preschool period represents a critical window for the acquisition of core narrative organization skills, thereby providing an important developmental benchmark against which differences observed in autistic populations can be interpreted.

4.2. Story Retelling in Non-Autistic and Autistic Children

Narrative retelling emerged as a less demanding task compared to spontaneous production in both preschool non-autistic and autistic children, evidenced by their better overall performance across all age groups. Performance differences between retelling and production tasks suggest that the cognitive load associated with creating novel content in spontaneous storytelling is substantially higher than that for reproducing a story. In non-autistic preschoolers, retelling performance was robust across age cohorts (Castilla-Earls et al., 2015; Crawshaw et al., 2020). Specifically, story structure improved significantly between the ages of 4 and 5 years (Isbell et al., 2004), and by the age of 5 and 6 years, children were able to incorporate the essential story components, while total retelling scores and content adequacy showed significant progress (Filiatrault-Veilleux et al., 2016; Kanellou et al., 2016). However, the microstructure (linguistic complexity) alone could not account for the significant developmental differences observed in retelling quality between the 5-year-olds and older groups, nor explained the temporal succession differences between 4–5 and 5–6-year-olds (Khan et al., 2016; Kanellou et al., 2016; Ralli et al., 2021). Thus, it could be claimed that this progress may be mainly driven by macrostructural components, reinforcing the hypothesis that narrative quality is more dependent on macro-level organization than sentence-level complexity in this age range. Likewise, research on retelling in bilingual autistic children shows significant relations between the two languages in macrostructure but not in microstructure (Govindarajan & Paradis, 2022), further implying that the two levels of narrative development follow dissociable trajectories in bilingual non-autistic child populations. Research also exists from non-autistic bilingual children showing no connections in macrostructure between the two languages (Bitetti et al., 2020). This discrepancy may be attributed to differences in the input quantity and quality on bilingual language development that is often difficult to measure in studies with bilingual children.

As story retelling is a less demanding task than spontaneous production, it can reveal the residual strengths in autistic children’s narrative profile. Preschool autistic children utilized the existing story knowledge in retelling narratives, although certain challenges regarding the processing, cognitive representations, and macrostructural measures (e.g., story grammar) remained (He et al., 2025; Gabig, 2008). Furthermore, while receptive vocabulary was linked to the number of different words used in retellings, the overall story structure remained weak (He et al., 2025). This outcome reinforces the view that structural difficulties in autism may not be purely lexical, but involve cognitive demands related to organizing and structuring the information provided (Yang et al., 2022).

4.3. Narrative Comprehension and Inference Skills in Non-Autistic and Autistic Children

The ability to comprehend narratives, particularly through inferential processing, appears to develop across the preschool years, although it is characterized by high inter-individual variability. Studies point out that inferential comprehension emerges around the age of 4 years, initially focusing on concrete elements, such as character actions, dialogue, and immediate emotional states (Filiatrault-Veilleux et al., 2016), and contributing to children’s listening comprehension and vocabulary development (Lepola et al., 2012). By the age of 4 years, non-autistic children infer the story’s problem and predict sequencing, while between the ages of 5 and 6 years, they have mastered inferring the main character’s overarching goal, the attempts made to solve the problem, and the final resolution (Filiatrault-Veilleux et al., 2016). Furthermore, research evidence highlighted the existence of a reciprocal relationship between inferencing and listening comprehension, as listening comprehension at age 4 predicted inference skills at age 5, which, in turn, predicted better listening comprehension at age 6, suggesting mutual development rather than a simple unidirectional influence between the two domains (Lepola et al., 2012). However, once children’s age and vocabulary were controlled, inferencing ability did not significantly predict overall comprehension compared to the production of character states, goals, and actions (Florit et al., 2014; Kendeou et al., 2008; Nuske & Bavin, 2011). Finally, it is important to note that performance significantly improved across all comprehension tasks when children were asked elaborative questions regarding story elements, indicating that active questioning promotes better structural coherence and extraction of the main idea (Silva et al., 2014). Findings on non-autistic preschoolers regarding their narrative comprehension skills are in accordance with the narrative skills of older children who also need to initially construct narrative scripts and make inferences to understand the external and internal nuances of stories (Tompkins et al., 2013; Lervåg et al., 2018; Cain et al., 2001; Bohnacker, 2016).

Narrative comprehension constitutes a major challenge for preschool autistic children, as they consistently show significantly poorer performance than their non-autistic peers. This difficulty mainly derives from problems in inferential processing (Nuske & Bavin, 2011). Autistic children were generally better at answering factual (what, who) than inferential questions, as they showed difficulties in spontaneously making inferences about event scripts, i.e., mental representations of typical sequences of events in familiar situations, or integrating information to draw conclusions (Westerveld et al., 2012). These challenges appear to be linked to both linguistic (e.g., vocabulary, semantics) and cognitive factors (e.g., working memory) (Westerveld & Roberts, 2017).

4.4. Language Skills and Narrative Performance in Non-Autistic and Autistic Children

The relationship between core language skills and narrative competence is significant but complex, suggesting that they are related but conceptually distinct processes. Even from the age of 4 years, strong correlations existed between story retelling and core language measures (vocabulary, grammar, verbal memory) (Karlsen et al., 2021; Bonifacci et al., 2018; Rezzonico et al., 2016; Uccelli & Páez, 2007). In addition, positive correlations were found between vocabulary, semantic measures, and comprehension, although vocabulary was found to be a necessary but not the sole component uniquely accounting for narrative comprehension performance (Florit et al., 2014). By the age of 5 years, tense marking, along with the use of verbs and conjunctions, improves, contributing to children’s ability to convey action and organize events in a coherent sequence. However, while narrative and core language skills are linked, they are best understood as partially overlapping but separable constructs. This is supported by the finding that syntactic complexity advancement in retellings (e.g., use of complex verbs and references) often outpaced concurrent changes in vocabulary, suggesting that organizational skill (e.g., syntax and grammar) matures differently from lexical breadth (Karlsen et al., 2021; Safwat et al., 2013).

Analysis of microstructure (linguistic complexity) indicated that differences in early narrative structure are not always significant across close age groups. For example, comparisons between the 4–5 and 5–6 age groups showed significant improvement across most criteria, except for the sequencing of temporal events (Mäkinen et al., 2018; Nicolopoulou & Richner, 2007). Additionally, microstructure alone could not account for the developmental differences observed in retelling between older groups, implying that complexity at the sentence level may not fully capture narrative growth (Ralli et al., 2021). When comparing linguistic complexity with narrative ability, the results confirm the importance of macrostructure. While microstructure (sentence-level features) was sufficient to distinguish between some age groups, it failed to explain the developmental leaps in retelling quality (e.g., between 5–6 and 6+ years) (Kendeou et al., 2008). Thus, the ability to organize events into a goal-driven and coherent plot (macrostructure) is the primary marker of narrative maturity, overshadowing sentence construction skills in later preschool years. This distinction is further illuminated by comparative findings. For example, non-autistic children generally produced narratives with richer central ideas, environments, characters, and objects than the autistic group (Kenan et al., 2019). Yet, no significant differences were found between groups in terms of irrelevant comments, suggesting that focus and relevance are relatively preserved, unlike the issues with content adequacy (Kenan et al., 2019). In addition, productivity metrics (MLU, NDW, TNW) consistently increased with age across groups, although the vocabulary diversity decreased across all age groups, reflecting possibly a consolidation phase where children use fewer and more precise terms in their narratives (He et al., 2025; Gabig, 2008). While their macrostructure was weak, the autistic preschoolers presented a more nuanced picture regarding microstructure. They showed relative strengths in productivity, semantic diversity (number of different words), and grammatical complexity, while they simultaneously exhibited weaknesses in grammatical accuracy (He et al., 2025; Westerveld et al., 2012). Receptive vocabulary emerged as a strong predictor of their oral narrative production and narrative comprehension skills (Gabig, 2008). However, this relation was not universal, as non-word repetition and digit span (measures of phonological/visual memory) were not linked to receptive vocabulary (He et al., 2025), indicating that the impact of vocabulary is largely semantic and related to linguistic output, rather than the children’s auditory processing speed. The partial dissociation between microstructure and macrostructure is also evidenced in studies with bilingual non-autistic children who manage to improve in narrative macrostructure despite their lower lexical and syntactic skills as compared to their monolingual autistic peers (Bohnacker, 2016; Gagarina, 2016; Uccelli & Páez, 2007; Yang et al., 2022). The distinction between narrative macrostructure and microstructure is further supported by evidence from studies on non-autistic populations. For example, Hipfner-Boucher et al. (2015) found that while macrostructural organization (e.g., story grammar) remained stable across groups, microstructural features, such as lexical diversity and grammaticality, varied depending on language exposure. Similarly, Rodina (2017) reported that bilingual children exhibited comparable macrostructural abilities across their two languages, whereas microstructural performance was more sensitive to differences in language input. These findings reinforce the view that macrostructure may reflect more general cognitive and narrative organization skills, while microstructure is more closely tied to language-specific proficiency.

The findings of the present review should also be interpreted in light of evidence highlighting the role of language exposure and bilingualism in shaping narrative abilities. Studies, such as Hipfner-Boucher et al.’s (2015) and Rodina’s (2017), demonstrate that variability in narrative performance among non-autistic children is strongly influenced by the quantity and quality of linguistic input. In particular, microstructural aspects of narratives appear to be especially sensitive to language dominance and exposure, whereas macrostructural organization remains relatively robust across different language contexts. This variability underscores the importance of considering linguistic background when interpreting differences between autistic and non-autistic groups.

Finally, the inclusion of studies that treat narrative as part of a wider language battery highlights the interdependence of linguistic domains. Findings from Thordardottir et al. (2010) reinforce the notion that for non-autistic preschoolers, narrative progression is not an isolated milestone but correlates with the maturation of vocabulary and syntactic processing skills. This contrast is particularly relevant when comparing these profiles to those of autistic children, whose narrative challenges, especially in macrostructure, may persist even when formal language measures are within the average range.

4.5. Cognitive, Social, and Environmental Factors, and Narratives in Non-Autistic and Autistic Children

Certain socio-cognitive factors have been found to play an important role in narrative competence. Particularly, skills such as ToM abilities (Mason & Just, 2009; Kim, 2015) and executive functions (e.g., working memory) (Friend & Bates, 2014) are required for constructing structurally sophisticated oral stories with more elaborate character perspectives. Deficits in verbal working memory were related to the recollection of significantly fewer propositional units, hindering children’s ability to hold complex linguistic and cognitive representations in mind long enough to integrate them in narrative production (Florit et al., 2014; Friend & Bates, 2014; Veraksa et al., 2020). Friend and Bates (2014) also noted that children’s ability to focus their attention on a task and inhibit or resist possible distractions at the age of 4.5 years confers benefits in the ability to construct a complex and coherent narrative at 5 years of age. These researchers observed that more advanced narratives at 4.5 years were indicative of better performance on executive function tasks at 5 years of age. The lack of a bidirectional relationship between executive functions, such as attention and inhibition, and narrative performance implies that narrative and executive functions comprise skill sets that develop asynchronously during the preschool years. The asynchronous developmental trajectories of these skills suggest that narrative development is not a by-product of executive functions, nor is executive function an outcome of narrative development.

In addition, differences in narrative production and comprehension may be associated with broader socio-cognitive processes, including ToM, abstract concept representation, and processing style, which support the ability to infer characters’ goals and motivations (Lepola et al., 2012). For example, challenges in generating inferences about event scripts or integrating script knowledge may relate to differences in these underlying processes. However, it is important to note that ToM and local processing bias were not consistently or directly measured across the reviewed studies. Therefore, their role should be interpreted cautiously as a potential explanatory framework rather than a confirmed or group-specific mechanism. Moreover, such socio-cognitive processes may be relevant to narrative performance in both autistic and non-autistic children, given their broader role in social understanding and language development; direct examination of ToM, processing style, and narrative skills within the same samples would allow for clearer delineation of the extent and specificity of these associations. Furthermore, answering inferential questions may be more demanding, potentially reflecting broader differences in ToM and abstract concept representation that support understanding character goals and motivations (Gabig, 2008). Interestingly, the extraction of the main idea correlated significantly with the performance on all comprehension tasks, suggesting that the inability to grasp the overarching theme may affect their narrative comprehension performance (He et al., 2025; Gabig, 2008).

Moreover, gender differences were apparent in retelling, with girls consistently outperforming boys (Nicolopoulou & Richner, 2007). Presentation modality was an additional important factor influencing children’s performance. Specifically, more elaborate retellings (covering Who, What, Where) were noted following video presentation compared to other modalities, such as book reading (Crawshaw et al., 2020). Exposure to stories through diverse media (e.g., from picture books to television and apps) has been found to enhance children’s language, literacy, and socio-cognitive abilities (Veneziano & Nicolopoulou, 2019; Adornetti et al., 2022). Finally, story retelling generated better performance in areas concerning narrative framing (formal ending, setting description, moral, and character recall), suggesting that it enhances imaginative recollection effectively (Castilla-Earls et al., 2015; Westerveld et al., 2012).

The narrative skills of preschool children are characterized by rapid development. While foundational language skills establish a necessary base, narrative structure depends on the effective deployment of goal-oriented inference and organization. Environ-mental factors, involving rich conversational interactions and targeted questioning, appear crucial for bridging the gap between simple event sequencing and the production of coherent stories. Socioeconomic status (SES), mothers’ education, and a family’s supportive environment, as well as school curricula explicitly teaching narrative skills during early literacy training, have been reported as significant predictors of linguistic and narrative performance (Silva et al., 2014; Souza & Cáceres-Assenço, 2021; Gagarina, 2016). The nature of parents’ scaffolding proved highly influential. Questions requiring elaboration (as opposed to simple identification) promoted thinking about relational events and fostered greater coherence (Souza & Cáceres-Assenço, 2021). Thus, children who engaged in storytelling after answering scaffolding questions tended to produce narratives of higher structural quality, suggesting that targeted questioning effectively aids in the construction of higher-level narrative frames. Finally, school type affected narrative outcomes, with preschoolers attending public schools scoring higher across all narrative tasks as compared to peers in private schools (Silva et al., 2014). This finding warrants further investigation into curriculum or classroom interaction styles. When assessing oral narrative skills, the choice of task administration and the potential challenges these tasks raise should also be taken into consideration. For example, telling or retelling a story without an illustrated book can be extremely difficult for young children and ‘non-compliance’ to the task may occur. Westerveld et al. (2012) reported that approximately 12% of their 4-year-old non-autistic participants refused to retell a story or provided a few utterances for analysis when they were asked to tell a story without pictures, making the researchers speculate that children’s non-compliance could be attributed to the complexity of the task.

4.6. Comparison Between Non-Autistic and Autistic Preschool Children’s Narrative Abilities

Comparisons between non-autistic and autistic preschoolers highlight that while both groups progress developmentally, they follow different underlying mechanisms and achieve different functional ceilings. The divergence lies in what drives narrative success. For non-autistic children, progress between the ages of 4 and 6 years mainly relied on mastering goal-directed structure leading to sophisticated narrative macrostructure (goal setting, episode components), even when microstructure competence ability was varying. For autistic children, low scores on content adequacy and story grammar across all ages imply a major difficulty in consistently internalizing these structural goals. This difficulty cannot be explained by linguistic complexity (microstructure), as both non-autistic and autistic groups showed comparable though independently developing grammatical complexity. The persistent inability of the autistic group in identifying central themes, main characters, and settings, alongside a lack of increased irrelevant commentary, indicates a fundamental difference in their information integration processes, thereby providing strong empirical support for the local processing bias account.

Moreover, cognitive predictors varied between groups. For non-autistic children, higher-level inferences predicted later comprehension, while vocabulary showed inconsistent effects. Conversely, autistic children were significantly impacted by verbal working memory deficits, leading to poorer retention of story details. This impacted their ability to construct complex narratives in real-time. The narrative challenges in autistic preschoolers were not attributed to a lack of vocabulary, raising questions about the strength of the role of cognitive mechanisms, including ToM and local processing bias, in children’s narrative performance. The relation between narratives, ToM and domain-general cognitive mechanisms has not been explicitly addressed in any study with either autistic or non-autistic preschoolers so far, so continued research into the specificity of these relationships is warranted to validate these assumptions.

The narrative challenges in autistic preschoolers were attributed less to a lack of vocabulary, and more to deficits in cognitive mechanisms necessary for meaningful language use, specifically ToM and local processing bias. Non-autistic children developed goal-based inference systematically, whereas autistic children failed to spontaneously generate script-based inferences, succeeding only on direct factual recall. While non-autistic children develop the ability to infer goals and plans between the ages of 4 and 5 years, the autistic group demonstrates difficulty utilizing even basic event scripts. This implies that scaffolding helps organize existing knowledge in non-autistic children, while difficulties in autism stem from a core deficit in representing abstract goals.

The role of memory also differed between the two groups. For instance, in non-autistic pre-school children, memory related to narrative structure, but its predictive significance for later comprehension was inconsistent. In contrast, autistic children demonstrated explicit verbal working memory deficits, which were linked to their inability to retain propositional units. This directly impaired their capacity for coherent narrative construction, mirroring their difficulty with tasks requiring the parallel processing of complex linguistic and cognitive representations. While both groups exhibited vocabulary differences (with autistic children being typically lower in lexical productivity measures), narrative impairment in autism appears rooted less in simple lexical access and more in the cognitive deployment of language (i.e., working memory, ToM, and local processing bias). In summary, the non-autistic profile reflects a gradual refinement of narrative structuring, moving from temporal sequencing to causal goal-planning, while the autistic profile is defined by a reduced tendency to adopt or effectively utilize this overarching goal structure, resulting in brittle comprehension and sparse and poorly integrated spontaneous narrative.

4.7. Gaps in the Existing Literature and Directions for Future Research

The interpretation of the findings within this review must be contextualized by several methodological limitations inherent in the primary studies. Sample heterogeneity and recruitment bias were evident, characterized by unbalanced sample sizes across age groups and gender, as well as unequal group distribution due to recruitment procedures. This imbalance challenges the robustness of fine-grained statistical comparisons, particularly those relying on interaction effects. Furthermore, the elicitation technique employed in several studies requires cautious interpretation; the pervasive use of illustrated storybooks in retelling tasks may have inadvertently aided comprehension, potentially offering a scaffolding effect similar to that observed with video stimuli, thereby masking true difficulties in memory-based recall, especially in autistic children. This is underscored by the observation of a ceiling effect in older children (6-year-olds) on story structure measures (where over 80% achieved maximal scores), suggesting that the elicitation task itself may have been less challenging than narrative recall from memory. In addition, certain factors (e.g., storybook exposure frequency, ToM) were investigated in only a small subset of studies. Consequently, conclusions regarding their role should be interpreted cautiously, and further research is needed to determine their contribution to narrative development. Relatively few studies directly compared autistic and non-autistic children. In some cases, conclusions regarding similarities or differences between groups are drawn across separate studies that may employ different constructs, measures, and methodologies. This variability may limit the comparability of findings and underscores the need for future research using shared narrative assessment frameworks and direct group comparisons. Another limitation of the present review relates to the variability in terminology used across studies, particularly in relation to age group classification. As not all relevant studies explicitly identify participants as “preschoolers” in titles or abstracts, some studies may not have been captured through database searches despite efforts to employ comprehensive search strategies and supplementary manual screening. Future reviews may benefit from broader or more flexible search approaches to further enhance coverage. Finally, the decision to exclude studies comparing non-autistic children with clinical populations other than autism was made to preserve conceptual clarity and focus. While this approach strengthens the interpretability of comparisons between autistic and non-autistic children, it may limit the broader generalizability of findings across different developmental conditions.

Based on these limitations, it is necessary to treat certain outcomes with caution. Specifically, results suggesting indirect predictive pathways (e.g., working memory not being an indirect predictor of comprehension) must be treated tentatively, given potential measurement overlap. Methodologically, the reliance on cross-sectional designs limits causal inference regarding age-related progression in story structure, making longitudinal data essential to confirm developmental sequencing. Moreover, several critical contributing factors were omitted from the analyzed data, including measures of comprehension monitoring, morphological understanding, quality of joint picture-book reading scaffolding, and broader social-pragmatic skills, all of which likely moderate narrative outcomes. A further constraint is the strictly structured research environment, which may not reflect the variability and complexity of narrative skill use in natural, daily interactions.

To advance the field of narrative performance in autistic preschoolers, future research should prioritize methodologies that address these gaps. Longitudinal designs are necessary to confirm the causal pathways between underlying cognitive skills (like memory and inference) and later narrative outcomes. Studies must strive for balanced sampling in both group sizes and gender distribution. Future work should also compare performance across multiple elicitation modalities (memory recall vs. picture-supported retelling) to discriminate structural mastery from scaffolding effects. Finally, future research must incorporate standardized, robust measures of grammatical understanding, social-pragmatic competence, and the quality of environmental exposure (e.g., storybook exposure frequency) to build more ecologically valid models of narrative development across the preschool years.

5. Conclusions

As children’s narrative skills are essential for effective communication, social interaction, and academic success, they are often the focus of clinical and research interest. Narratives can be used as ecologically valid assessment tools to detect subtle differences in children’s language production (Heilmann et al., 2010). Narrative analysis provides information regarding children’s language use at the level of microstructure (i.e., cohesion, narrative length, lexical diversity, linguistic complexity, grammatical and syntactical accuracy) (Heilmann et al., 2010; Westerveld et al., 2012) and macrostructure (i.e., coherence, story grammar, and structure of a story) (Spencer & Petersen, 2020). Thus, narration can be used for language evaluation purposes to create normative data in order to explore and understand the nature of several communication impairments and/or neurodevelopmental disorders (Lee et al., 2018; Rollins, 2014).

Thus, the insights gained from examining narrative production and comprehension in preschool years carry significant practical implications for both assessment and intervention. Given the strong, age-dependent relationship between macrostructural organization and inferential ability, interventions should move beyond simply targeting vocabulary or syntactic complexity. Instead, efforts should focus on explicitly scaffolding the hierarchical organization of information by teaching children how to identify initiating events, establish character goals, and establish causal links between actions and resolutions. Since the lack of narrative coherence keeps narrative skills weak, intervention should prioritize teaching the main idea and structure instead of focusing only on recalling isolated facts and events in the stories. Furthermore, as verbal working memory is crucial for maintaining the complex representations needed for coherent storytelling, enhancing working memory skills through specific training paradigms may offer an effective way to improve both narrative production and comprehension, especially for children exhibiting deficits in these areas. Concluding, it should be highlighted that fostering strong narrative skills during the preschool years is essential, as these abilities form the foundation for later literacy, abstract reasoning, and complex social communication skills.

Taken together, the inclusion of studies on bilingual and monolingual non-autistic children highlights that narrative development is shaped not only by developmental condition but also by linguistic experience and environmental factors, which should be carefully considered in future research and clinical assessment.