1. Introduction
The scientific literature on text readability is closely related to research on textual complexity and comprehension, as it investigates how linguistic features affect processing effort and, consequently, readers’ ability to understand written texts (Dale & Chall, Reference Dale and Chall1949; DuBay, Reference DuBay2004; Klare, Reference Klare1963; Nippold and Tomblin, Reference Nippold and Tomblin2014; Serki & Bolkan, Reference Serki and Bolkan2024; Zainurrahman et al., Reference Zainurrahman, Yusuf and Sukyadi2024 for a recent literature review). Readability is commonly defined as the ease/complexity with which readers can understand the content of a sentence or a written text (Weiss & Meurers, Reference Weiss and Meurers2019). From this perspective, a well-written text is one that achieves high readability by minimizing unnecessary processing effort for the reader, while preserving the accuracy and relevance of the conveyed information, thereby facilitating efficient and effective communication. Readability research translates these cognitive principles into concrete guidelines for effective communication. These include maintaining clarity, providing appropriate examples, and avoiding vague or underspecified language (Williams & Nadel, Reference Williams and Nadel1989). Although such principles may appear intuitive, they are grounded in empirical research and play a crucial role in ensuring that texts are not only readable but also communicatively effective.
Despite their apparent simplicity, these principles are often neglected. As George Orwell (Reference Orwell1946) noted in his widely cited essay Politics and the English Language, poorly written texts typically exhibit two recurrent flaws: staleness of imagery and lack of precision. The first reflects a loss of clarity and vividness that can be mitigated, we argue, using concrete, image-rich language, while the second results from vague or overly general wording, which can be reduced, we suggest, by employing more specific and accurate lexical choices.
Understanding the factors that shape the comprehensibility and informational richness of texts is therefore fundamental to achieving effective communication. However, the task is non-trivial, as comprehensibility is deeply subjective and relative: what one reader perceives as simple and transparent may appear demanding or ambiguous to another, depending on individual differences in linguistic and cultural experience, prior knowledge, cognitive development, and motivational factors (Dahl et al., Reference Dahl, Carlson, Renken, McCarthy and Reynolds2021) (Perfetti, Reference Perfetti1985). This inherent variability makes the assessment of textual clarity and informativeness particularly challenging and underscores the importance of considering individual differences in linguistic experience, reading ability, educational background, and cognitive development alongside objective linguistic indicators. This issue is especially relevant in educational contexts, as children in late primary and early secondary school transition from learning to read to reading to learn (Chall, Reference Chall1983). Moreover, children’s subjective perceptions of whether a text feels accessible or confusing may influence their willingness to engage with reading activities and cognitively demanding materials over time. Within the context of our investigation into how linguistic factors influence children’s perceptions of text quality and comprehensibility, we focus on particular lexical characteristics of the words used in text: concreteness and specificity.
Concreteness refers to the extent to which a word or concept can be directly experienced through the senses (e.g., “truth” is a highly abstract word, “tooth” is a highly concrete word), while specificity pertains to the level of detail and precision with which a word refers to a particular category or instance (e.g., “creature” and “religion” are general words, “macaque” and “Taoism” are specific words) (as defined in Lamarra et al., Reference Lamarra, Villani, Mazzuca, Borghi and Bolognesiunder review; Villani & Bolognesi, Reference Villani and Bolognesi2025).
Previous studies have examined how these lexical properties affect language processing, in tasks such as word recognition and semantic categorization (Lamarra et al., Reference Lamarra, Villani and Bolognesi2026; Mosenthal, Reference Mosenthal1996; Paivio, Reference Paivio1990). However, while it is well established that concrete expressions evoke vivid mental imagery and engage both visual as well as language-based representations (Paivio, Reference Paivio1971; Reference Paivio1991), and that specific terms reduces ambiguity and enhance informational richness (Kintsch & van Dijk, Reference Kintsch and Van Dijk1978), their impact on perceived clarity and informativeness remains insufficiently explored, especially in developing readers.
While previous studies have, for instance, examined the impact of clarity on learning outcomes (e.g., Serki & Bolkan, Reference Serki and Bolkan2024, for a recent empirical study and literature review), it is not quite clear which linguistic variables determine how developing readers perceive lexical clarity and informativeness. To address this gap, the present research seeks to understand whether and how lexical concreteness and specificity influence children’s perception of textual clarity and informativeness between the ages of 9 and 12. In this context, clarity is defined as the ease of understanding and informativeness as the richness and depth of information conveyed.
By combining a reader-centered approach with controlled lexical manipulations, this study aims to clarify the distinct roles of concreteness and specificity in shaping young readers’ evaluations of text comprehensibility.
Whereas concreteness is expected to mainly enhance perceived clarity by fostering vivid mental imagery and facilitating comprehension, specificity is hypothesized to mainly increase perceived informativeness by providing more detailed and precise semantic content. Based on these theoretical premises, the present study formulates the following research questions and related hypotheses:
-
- RQ1: To what extent does lexical concreteness influence children’s perceived clarity and informativeness of written texts between the ages of 9 and 12?
-
○ Hp. 1: Texts containing concrete words will be perceived as clearer by children aged 9–12 compared to texts containing more abstract words. The potential effect of lexical concreteness on perceived informativeness will be investigated in an exploratory manner.
-
-
- RQ2: To what extent does lexical specificity influence children’s perceived clarity and informativeness of written texts between the ages of 9 and 12?
-
○ Hp. 2: Texts containing highly specific words will be perceived as more informative than those containing more general words. The potential effect of lexical specificity on perceived clarity will be investigated in an exploratory manner.
-
-
- RQ3: Does the combination of lexical concreteness and specificity impact children’s perceived clarity and informativeness of written texts between the ages of 9 and 12?
-
○ Hp. 3: Based on previous behavioral evidence, we argue that texts containing highly concrete and specific words will be perceived as clearer and more informative than texts containing highly general and abstract words. We do not formulate strong hypotheses about the potential effects of texts that are general but concrete, or abstract but specific, on perceived clarity and informativeness.
-
Across all hypotheses, we also examine the effect of readers’ age on the response variables. We do not advance a strong a priori prediction regarding age effects, since prior evidence suggests that individual differences in children’s language-mediated abstraction skills (i.e., the ability to generate taxonomic associations ranging from very general to very specific) do not systematically interact with age (Villani et al., Reference Villani, Loia and Bolognesi2025).
Investigating children’s subjective perceptions of these dimensions offers a more comprehensive account of how young readers process and evaluate written material. Ultimately, this approach contributes to identifying lexical mechanisms that support comprehension and engagement during reading, with potential implications for educational practices.
2. Theoretical background
2.1. Clarity and informativeness as cognitive and communicative dimensions
The tension between being clear and being precise can be situated within a broader cognitive and communicative framework.
Within communication theory, clarity and informativeness are commonly associated with effective communication and successful knowledge transfer (De Beaugrande & Dressler, Reference De Beaugrande and Dressler1981; Grice, Reference Grice, Cole and Morgan1975; Halliday & Hasan, Reference Halliday and Hasan1976).
Grice’s conversational maxims of quality and quantity establish that a communicative contribution should be truthful and sufficiently informative, but not excessively so. In textual terms, these principles highlight the need to balance clarity, which ensures transparency and lack of ambiguity, with informativeness, that guarantees semantic density and relevance. In practice, these two variables are strongly interconnected, given that comprehension depends on the balance between linguistic transparency and informational load. Texts that are clear but insufficiently informative risk being trivial or unhelpful, whereas highly informative texts that lack clarity may overwhelm the reader and hinder understanding.
However, providing a precise and operationalized definition of clarity and informativeness is far from straightforward. These notions rarely receive explicit, operationalized definitions in the scientific literature, largely because they emerge from the interaction of multiple linguistic, cognitive, and pragmatic factors rather than from a single identifiable property of the text. As a result, many studies treat them as intuitively understood constructs – implicit outcomes of broader parameters such as linguistic transparency, cohesion, lexical choice, and semantic richness. Moreover, different disciplines treat and conceptualize these dimensions from distinct theoretical perspectives: whereas communication theory emphasizes pragmatic appropriateness, psycholinguistics focuses on processing ease, and applied linguistics highlights their pedagogical implications. This diversity further contributes to the inherently subjective nature of clarity and informativeness and complicates the task of delineating them with precision. Nevertheless, it is possible to identify several core characteristics that emerge consistently across studies in different fields.
2.2. Clarity
From a cognitive perspective, clarity can be defined as a text’s ability to minimize processing effort (Bulté & Housen, Reference Bulté and Housen2012; Miestamo et al., Reference Miestamo, Sinnemäki and Karlsson2008). This property tends to increase with linguistic complexity (Dahl, Reference Dahl2004; Dahl et al., Reference Dahl, Carlson, Renken, McCarthy and Reynolds2021), i.e., to the difficulty with which a message is encoded or decoded, which in turn depends on the transparency and lack of ambiguity of the communicated message. In this view, clarity corresponds to low cognitive complexity, as clear language reduces mental load and facilitates the construction of coherent mental models.
This cognitive interpretation aligns closely with research on knowledge communication, which emphasizes that clarity is essential to avoid misunderstandings and ensure successful knowledge transfer. Bischof and Eppler (Reference Bischof and Eppler2010) describe clarity as the absence of ambiguity, difficulty, or doubt, referring to the Oxford Dictionary definition. Their work highlights how unclear texts hinder comprehension, make information harder to apply, and interfere with learning or problem-solving, which in cognitive terms corresponds to an unnecessary increase in processing load. This point is further reinforced by the student’s perspective, according to which clarity is one of the primary criteria for evaluating the quality of teaching (Smidt et al., Reference Smidt, Li, Bunk, Kochem and McAndrew2017). Clear texts enable readers to readily identify communicative intentions and relevant information without the need for complex or effortful inferential processes. This effect is achieved through concision, transparency, and the avoidance of confusion. In this regard, Nordquist (Reference Nordquist2019) argues that concision is virtually synonymous with clarity, as both involve the use of brief, precise, and contextually appropriate language. Concision, or clarity in this sense, does not imply oversimplification, but rather the elimination of unnecessary complexity that does not contribute to meaning. It is therefore essential for conveying messages accurately and efficiently and constitutes a core dimension of effective writing and communication (Fitria, Reference Fitria2021; Harris, Reference Harris2017).
2.3. Informativeness
Whereas clarity primarily concerns processing ease and the lack of confusion, informativeness relates to the richness, precision, and relevance of the content conveyed by a text (Wu & Giles, Reference Wu, Giles, Vanderwende, Daum and Kirchhoff2013). Informativeness emerges from the interaction of multiple linguistic factors that jointly determine a text’s informational value.
From a cognitive perspective, a text can be considered informative to the extent that it supports the construction of a rich and well-specified mental representation. This involves not only the presence of information, but also the degree to which information is structured and relevant to communicative goals. Consequently, informativeness must be carefully balanced: while greater semantic detail can enhance comprehension, excessive informational density or poor organization may overload cognitive resources and hinder understanding.
Several interrelated linguistic dimensions contribute to informativeness. A central role is played by semantic precision, which narrows the referential scope of lexical items and enables clearer conceptual distinctions. By constraining the range of possible interpretations, specific words allow readers to construct more detailed and determinate mental representations, thereby increasing the perceived informational value of a text (Cruse, Reference Cruse1986; Murphy, Reference Murphy2002, Reference Murphy2010). However, greater specificity may also increase processing demands, particularly for younger or less experienced readers, highlighting the importance of considering reader characteristics when evaluating informativeness, as is evident in technical or highly specialized texts.
Closely related to specificity is lexical sophistication (Kyle & Crossley, Reference Kyle and Crossley2015; Meara, Reference Meara, Butler, Gómez González and Doval-Suárez2005), understood as the use of precise and semantically rich vocabulary. Sophisticated lexical choices enable fine-grained encoding of meaning and support the transmission of nuanced distinctions that would otherwise remain underspecified. In this sense, lexical sophistication contributes to informativeness by expanding a text’s conceptual space. Lexical variation, in turn, reduces redundancy and prevents the repeated activation of already processed representations, thereby increasing informational density without necessarily increasing text length.
A further key dimension concerns informational relevance, namely the extent to which relevant content is foregrounded and made salient to the reader. From this perspective, informativeness is not solely a matter of how much information is present, but also of how effectively key information is highlighted and integrated into the discourse structure (see also Sperber & Wilson, Reference Sperber and Wilson1986). This view aligns with research on knowledge communication, which emphasizes that informative texts are those that enable readers to reconstruct, apply, and transfer knowledge efficiently (Kamalski et al., Reference Kamalski, Sanders and Lentz2008; McNamara et al., Reference McNamara, Kintsch, Songer and Kintsch1996).
Crucially, informativeness must be balanced with clarity. As noted by Bischof and Eppler (Reference Bischof and Eppler2010), excessive detail, overly dense formulations, or an accumulation of poorly integrated information may lead to information overload, thereby reducing comprehensibility. Informativeness, therefore, does not increase monotonically with semantic complexity; rather, it emerges from an optimal balance between informational richness and cognitive accessibility. Moreover, perceptions of informativeness may vary developmentally, as readers of different ages may rely on partially different criteria when evaluating what counts as “informative” content.
2.4. Lexical factors influencing clarity and informativeness
Empirical research in cognitive, computational, and text linguistics has extensively investigated the linguistic features that shape the perceived complexity of texts and, consequently, influence readers’ judgments of clarity and informativeness (Burani et al., Reference Burani, Barca and Arduino2001; Collins-Thompson, Reference Collins-Thompson2014; North et al., Reference North, Zampieri and Shardlow2023). These studies converge in showing that perceived complexity does not arise from a single property of the text, but from the interaction of multiple lexical and semantic parameters that modulate processing effort and semantic accessibility in the working memory (Khairova, Reference Khairova2019; Štajner et al., Reference Štajner, Evans, Orasan and Mitkov2012; Cain et al., Reference Cain, Oakhill and Bryant2004; Daneman & Merikle, Reference Daneman and Merikle1996; Just & Carpenter, Reference Just and Carpenter1992; Linderholm & van den Broek, Reference Linderholm and van den Broek2002).
Among these parameters, lexical concreteness has emerged as a key factor in processing ease and readability. Tanaka et al. (Reference Tanaka, Jatowt, Kato and Tanaka2013) demonstrated that word concreteness plays a crucial role in determining text readability, showing that concrete words enhance comprehension, interest, and positive attitudes toward texts by activating sensory-based mental representations. Concrete words evoke vivid imagery and engage dual-coding mechanisms (Paivio, Reference Paivio1971), thereby facilitating semantic access and memory while reducing processing effort, despite activating richer sensory and multimodal mental representations. Conversely, lexical abstractness has been consistently associated with increased processing difficulty. Abstract words possess fewer perceptual features and are harder to visualize, making them less accessible, particularly for younger or less experienced readers (Beinborn et al., Reference Beinborn, Zesch and Gurevych2014).
Recent work shows that the use of abstract words in conversational contexts is often associated with states of uncertainty, epistemic exploration, and the need for clarification, frequently co-occurring with questions, requests for explanation, and expressions of curiosity. In contrast, the use of concrete words tends to correlate with communicative contexts characterized by greater certainty, confidence, and reduced need for additional clarification (Mazzuca et al., Reference Mazzuca, Villani, Lamarra, Bolognesi and Borghi2025). These findings suggest that concreteness may contribute to perceived clarity not only by facilitating cognitive processing, but also by signaling epistemic stability and communicative certainty in discourse. At the same time, communicative contexts characterized by greater abstraction and interpretative openness may also promote curiosity, reflection, and exploratory forms of engagement, even when they are perceived as less immediately clear.
Lexical specificity, on the other hand, has been shown to influence textual comprehensibility, albeit in a more nuanced way. Overall, more specific lexical items reduce interpretive ambiguity by constraining the range of possible referents, thereby supporting less effortful representations (Cruse, Reference Cruse1986; Murphy, Reference Murphy2002, Reference Murphy2010). From a cognitive perspective, research on categorization suggests that different levels of specificity differentially affect processing: while basic-level terms optimize accessibility and recognizability, highly specific terms increase informational precision (Rosch, Reference Rosch, Rosch and Lloyd1978; Rosch et al., Reference Rosch, Mervis, Gray, Johnson and Boyes-Braem1976). For this reason, specific words, which narrow the referent to a particular category or instance, may enhance perceived informativeness by providing detailed and precise information. Recent research shows that highly specific words are indeed processed faster than highly general words, independently of their concreteness (Lamarra et al., Reference Lamarra, Villani and Bolognesi2026). These experimental findings contribute to supporting the theoretical differentiation between concreteness and specificity, and the different effects that each of these two variables has on language processing and comprehension.
Finally, recent research has shown that the combination of lexical concreteness and specificity gives rise to different types of words (e.g., Villani et al., Reference Villani, Loia and Bolognesi2024). For instance, the word “substance” is fairly general, although concrete. Conversely, “yeast” is fairly specific and concrete. The same difference applies to abstract concepts: “art” is abstract and general, while “Impressionism” is abstract but specific. Such words are processed in different ways, with concrete-specific words showing an advantage in semantic processing over abstract-general words (Lamarra et al., Reference Lamarra, Villani and Bolognesi2026). Moreover, on a qualitative level, it appears that concrete-specific words define prototypical concrete words, while abstract-general words define prototypical abstract words. Extending these observations from individual lexical items to the textual level, the interaction between concreteness and specificity can be expected to shape different dimensions of textual complexity. In particular, concreteness may facilitate semantic access and mental imagery, promoting clarity, whereas specificity may contribute to perceived informativeness by providing more detailed and precise referential content. Consequently, texts combining highly concrete and specific lexical items may be expected to optimize both clarity and informativeness.
3. Methods
The study was not pre-registered. In accordance with the Transparency and Openness Promotion (TOP) Guidelines, the dataset generated in the current study, the full list of stimuli, and analysis scripts are available in the online repository on Open Science Framework (OSF): https://osf.io/3hvjn/.
3.1. Participants
A total of 127 children (9–12 years old) were recruited from one primary and one secondary school in Imola (Italy), as part of an official, school-supported research activity conducted in collaboration with the University of Bologna. Twenty-four students with special educational needs or diagnosed specific learning disorders were included in the data collection to ensure inclusivity; however, they were excluded from the statistical analyses to maintain sample homogeneity and comparability of the data. The final sample thus comprised 103 participants. This includes two fifth-grade primary school classes (n = 33, 17 female, age 9–10), two first-year secondary school classes (corresponding to sixth-grade classes in the American school system), (n = 32, 16 female, age 10–11), and two second-year secondary school (corresponding to seventh-grade classes) (n = 38, 20 female, age 11–12). So, the final sample was relatively balanced across age groups and gender, with comparable numbers of participants in each age range and a similar distribution of female and male students across grades.
The study was approved by the Ethics Committee of the University of Bologna with informed consent and a privacy policy signed by the school principal with the approval of the children’s parents or caregivers. The school interventions were officially included in the official educational plan provided by the school for the school year 2023–2024 (POF 2023–2024, protocol number 0006311, dated 12/06/2023 and classified as “IV.5 – Progetti e materiali didattici”), as a formal collaboration between the public institution IC7 public school in Imola (Italy) and the research group Abstraction.
3.2. Materials
The dataset of original sentences is taken from Lamarra et al. (Reference Lamarra, Villani, Mazzuca, Borghi and Bolognesiunder review). The dataset contains balanced and normed sentences obtained by manipulating the specificity level of a target noun. For each basic-level concept, a general – specific pair was constructed by replacing the target noun with either a more specific hyponym or a more general hypernym. Both abstract and concrete domains were represented.
All sentences were matched for morphological structure and consisted of a subject, verb, and noun only (e.g., “Ho fatto una torta”/“I made a cake”; “Ho espresso un giudizio”/“I made a judgment”). According to the original norming data, the selected items did not differ significantly in naturalness, familiarity, or degree of abstractness (see Lamarra et al., Reference Lamarra, Villani, Mazzuca, Borghi and Bolognesiunder review). From this set, we selected 32 sentences distributed across four conditions (abstract – general, abstract – specific, concrete – general, concrete – specific; 8 items per condition), likely to contain words within the vocabulary knowledge of the target age group. Particular attention was paid to avoiding highly specialized or developmentally inappropriate lexical items, although some variability in familiarity across participants cannot be entirely excluded. In addition, 8 basic-level sentences (4 abstract, 4 concrete) were included as a baseline. The dataset resulted in 40 sentences.
Finally, to provide discourse context and construct the texts, we embedded each target sentence in a short narrative generated using OpenAI’s conversational agent (ChatGPT) based on GPT4. This was achieved by adding one sentence preceding and one sentence following the target sentence. The exact prompt is available in the OSF repository.
This procedure resulted in 40 short texts, each consisting of three sentences, with the target word appearing in the middle sentence (see Table 1 for examples, and the complete list on the OSF online repository).
Examples of experimental texts with English translations
Table 1. Long description
The table is organized into three columns: Category, General text, and Specific text.
* The first row under the Concrete category features a General text about making a dessert (dolce) and a Specific text about making a tart (crostata).
* The second row under the Abstract category features a General text about solving a problem (problema) and a Specific text about solving a riddle (indovinello).
Each cell contains the Italian text followed by its English translation. Target sentences are highlighted in bold, and the specific target concepts (dessert, tart, problem, riddle) are underlined. For example, in the Concrete Specific cell, the bolded English text reads: I made a delicious tart. The word tart is underlined.
Note: Target sentences are in bold, and target concepts are underlined.
The short texts were distributed in a semi-random order across 10 experimental lists (20 per list) using a design such that each participant was exposed to only one level of specificity for each lexical item (e.g., “crostata”, corresponding to “tart”, or “torta”, corresponding to “cake”, or “dessert”), while maintaining a balanced number of concrete and abstract items across lists. Note that the experimental manipulation concerns the target sentences in which the level of specificity/concreteness of the target word varies, whereas the context sentences generated using ChatGPT, as illustrated in the example in Table 1, were repeated across conditions and therefore should not constitute a potential confound.
3.3. Procedure
Within each class, following a brief welcome and mutual introduction, the experimenter verbally explained the task to the students. They were informed that, after reading a series of texts collectively, they would be asked to evaluate how easy the texts were to understand and how much information they conveyed. To increase engagement, the activity was also framed as a sort of game in which the children would take on the role of individual “judges” of each of the texts shown on the interactive whiteboard. To clarify the task to the children, the analysts showed the following texts to the participants and explained the type of ratings they had to provide: “Ho preso una cosa in un negozio. – I bought a thing in a shop”, and “Ho preso un cacciavite a stella per montare la macchinina. – I took a star screwdriver to assemble the toy car”. Children were also verbally encouraged to think about how much useful or relevant content the texts conveyed when responding to the informativeness questions.
These examples were introduced only to illustrate the kinds of features the children would later evaluate and were not intended to suggest how they should respond. The experimenter emphasized that the examples were hypothetical and served only to illustrate the task.
Instructions emphasized that there were no correct or incorrect answers and that the children should express their judgments based solely on their own opinions and impressions.
As previously anticipated, each text was projected onto an interactive board in the classroom, in large black fonts on a white background, and read aloud by an experimenter.
The decision to conduct a collective reading session was made in consultation with the schoolteachers, who advised against the initial plan of having children read and rate the texts autonomously. According to the teachers’ pedagogical experience, an individual reading design would likely have resulted in unreliable data, given the length of the task and the sustained, repetitive attentional demands it imposed, which were not considered optimal.
The collective reading of a text projected on the screen, followed by individual responses, was therefore adopted as a compromising solution aimed at maximizing task engagement and data quality. This procedure ensured that all participants were exposed to the same input under comparable conditions, while still allowing judgments to be expressed and recorded independently. At the same time, we acknowledge that collective reading may reduce natural variability in reading pace and comprehension strategies and may introduce a degree of shared contextualization. However, given the classroom setting and the developmental characteristics of the participants, this design was considered the most ecologically valid and methodologically robust option. Accordingly, the present findings should be interpreted as reflecting children’s evaluations of texts under guided listening and shared classroom reading conditions rather than fully autonomous individual reading situations.
After each text, children responded individually to four questions on a paper questionnaire, using verbal categorical responses on 5-point Likert scales. This format has proven particularly suitable for pupils in the age range targeted here (Hall et al., Reference Hall, Hume and Tazzyman2016; Mellor & Moore, Reference Mellor and Moore2014; Royeen, Reference Royeen1985; van Laerhoven et al., Reference Van Laerhoven, van der Zaag-Loonen and Derkx2004). Participants evaluated each text along four dimensions: clarity (ease and confusion) and informativeness (quantity and relevance). Ease of understanding was assessed with the question “How easy was it for you to understand the text?”, and confusion with “How confusing did you find the text?”. Informativeness was measured by asking “How much information did you find in the text?” (quantity) and “Were the information in the text important for understanding what it was about?” (relevance). Please see the OSF repository for experimental protocols.
Data collection was conducted over two consecutive days across the 6 classes, with each experimental session lasting approximately 90 minutes per class.
4. Analyses
All analyses and data visualizations were performed in R (R Core Team, 2019) using RStudio (Version 4.4.3). Analyses were conducted in two main steps. First, we examined the relationships among the four rating dimensions – Clarity_1 (easiness), Clarity_2 (confusion), Informativeness_1 (quantity), and Informativeness_2 (relevance). Second, we investigated how text characteristics (concreteness, specificity, and their combination, operationalized as independent variables), as well as age, influenced children’s evaluations, operationalized as dependent variables.
To assess associations among the four rating dimensions and to identify potential collinearity, Spearman’s rank-order correlation coefficients were computed. This nonparametric method was chosen because responses were collected on 5-point ordinal Likert scales and to reduce sensitivity to non-normal distributions and outliers.
Multicollinearity among predictors was further evaluated by computing variance inflation factors (VIFs) using the vif() function from the car package (Fox, Reference Fox2005). VIFs were calculated for each model predicting one rating dimension from the remaining three.
To examine the effects of text characteristics and age on children’s evaluations, linear mixed-effects models were fitted using the lmer() function from the lme4 package (Bates et al., Reference Bates, Machler, Bolker and Walker2015). Separate models were estimated for each of the four rating dimensions (Clarity_1, Clarity_2, Informativeness_1, Informativeness_2) as dependent variables. Three fixed-effects structures were tested in separate models: (a) text concreteness (two levels: abstract, concrete), (b) text specificity (three levels: basic, general, specific), and (c) combined text type (six levels: abstract – basic, concrete – basic, abstract – general, abstract – specific, concrete – general, concrete – specific). In all models, text characteristics were included in interaction with age (three groups: 9–10, 10–11, and 11–12 years). Random intercepts were specified for participants and texts.
The significance of fixed effects and their interactions was assessed using Type III analyses of variance with Satterthwaite’s approximation for degrees of freedom, applied to fitted linear mixed-effects models and implemented with the anova() function. When significant effects were observed, post-hoc pairwise comparisons were conducted using the emmeans package (Lenth et al., Reference Lenth, Piaskowski, Banfai, Bolker, Burkner, Gine-Vazquez, Herve, Jung, Love, Miguez, Riebl, Singmann, Derumigny and Schmidt2024), with Tukey adjustments applied to control for multiple comparisons.
Results from the correlation analyses and the mixed-effects models are reported and discussed separately in the Results section.
5. Results
5.1. Correlations
Before fitting the mixed-effects models, we examined the relationships among the four rating dimensions to assess their degree of conceptual overlap and to evaluate whether they could be treated as related yet distinct constructs (Figure 1).
Correlation matrix of the study’s rating dimensions. Numbers inside the circles display Spearman’s rank correlation coefficients (ρ). Circle size reflects the magnitude of the correlation. Blue indicates positive correlations, and red indicates negative correlations.
Figure 1. Long description
The matrix consists of a four by four grid where only the lower triangle is populated with data circles. The variables on both axes are clarity underscore 1, clarity underscore 2, informativeness underscore 1, and informativeness underscore 2.
* Row 1, Column 1: A large dark blue circle with the value 1.00, representing the self-correlation of clarity underscore 1.
* Row 2, Column 1: A medium blue circle with the value 0.67.
* Row 2, Column 2: A large dark blue circle with the value 1.00, representing the self-correlation of clarity underscore 2.
* Row 3, Column 1: A small light blue circle with the value 0.34.
* Row 3, Column 2: A small light blue circle with the value 0.40.
* Row 3, Column 3: A large dark blue circle with the value 1.00, representing the self-correlation of informativeness underscore 1.
* Row 4, Column 1: A small light blue circle with the value 0.34.
* Row 4, Column 2: A small light blue circle with the value 0.37.
* Row 4, Column 3: A medium blue circle with the value 0.57.
* Row 4, Column 4: A large dark blue circle with the value 1.00, representing the self-correlation of informativeness underscore 2.
At the bottom, a horizontal color scale bar ranges from negative 1 in dark red, through 0 in white, to positive 1 in dark blue. All circles in the matrix are shades of blue, indicating positive correlations.
Spearman’s rank-order correlations revealed a moderate association between the two clarity measures (ρ = .67) and between the two informativeness measures (ρ = .57), suggesting satisfactory internal consistency within each construct. By contrast, correlations across constructs were noticeably weaker. Both clarity measures showed only low to moderate correlations with the informativeness measures (ρs ranging from .34 to .40), indicating that, although related, clarity and informativeness capture partially independent aspects of children’s text evaluations rather than reflecting a single underlying dimension. This interpretation was further supported by multicollinearity diagnostics. Variance Inflation Factors (VIFs) for all predictors remained well below commonly accepted thresholds (all VIFs < 2), indicating no collinearity and justifying the inclusion of the four rating dimensions as separate dependent variables in subsequent analyses.
5.2. To what extent does lexical concreteness influence children’s perceived clarity and informativeness of written texts between the ages of 9 and 12?
For Clarity 1 (easiness), we found a main effect of concreteness (F (1, 37.52) = 7.16, p = .011), showing that concrete texts were rated as easier to understand (EMM = 4.24, SE = 0.09) than abstract texts (EMM = 3.95, SE = 0.09). Lexical concreteness, therefore, reliably and significantly improves perceived ease of processing. This effect emerged independently of age, as neither a significant main effect of age (p = .192) nor a significant interaction between age and concreteness (p = .339) was observed. The absence of interaction suggests that the facilitating role of concreteness on perceived ease is stable across the age range examined.
A comparable pattern was observed for Clarity 2 (confusion). We found a significant main effect of concreteness, F (1, 37.94) = 5.70, p = .022, with concrete texts eliciting lower confusion ratings (EMM = 3.90, SE = 0.09) than abstract texts (EMM = 3.62, SE = 0.09). This result further supports the idea that lexical concreteness reduces interpretive uncertainty and cognitive effort. Importantly, no main effect of age was observed (p = .108); nor did age interact with concreteness (p = .499), indicating that the beneficial effect of concreteness on confusion judgments was stable across the examined age range. Together, these findings indicate that concreteness has a general, age-independent benefit on clarity-related judgments.
In contrast, a different pattern emerged for Informativeness 1 (quantity). We found no significant main effect of concreteness (p = .686), suggesting that concrete lexical content did not increase the perceived amount of information conveyed by a text. Instead, ratings were primarily shaped by age, as evidenced by a significant main effect of age, F (2, 99.39) = 8.84, p < .001. Post-hoc comparisons showed that children aged 11–12 assigned significantly lower informativeness ratings than both the 9–10 (EMM = 0.39, SE = 0.12, p = .003) and 10–11 (EMM = 0.44, SE = 0.12, p < .001) age groups, whereas no difference emerged between the two younger groups (EMM = –0.05, SE = 0.12, p = .918). Importantly, this age effect was not accompanied by an interaction with concreteness (p = .174), indicating that age-related differences in perceived informational quantity applied similarly to both concrete and abstract texts.
A partially complementary pattern emerged for Informativeness 2 (relevance). As with Informativeness 1, concreteness did not exert a significant effect on children’s judgments (p = .746), nor did it interact with age (p = .183), indicating that the perceived relevance of the information was not modulated by the concreteness of lexical content.
5.3. To what extent does lexical specificity influence children’s perceived clarity and informativeness of written texts between the ages of 9 and 12?
To ensure methodological consistency and comparability with the concreteness analyses, we applied the same mixed-effects modeling approach to examine the role of specificity on children’s judgments of clarity and informativeness.
The mixed-effects model predicting Clarity 1(easiness) ratings revealed no significant main effects of specificity level (p = .225) or age (p = .190), nor a significant interaction between the two factors (p = .235). Thus, perceived easiness did not vary systematically as a function of lexical specificity across age groups.
A different pattern emerged for Clarity 2 (confusion). While the main effects of specificity and age did not reach significance (specificity, p = .082; age, p = .114), the interaction between specificity level and age was significant, F (4, 1911.66) = 2.98, p = .018.
Post-hoc comparisons revealed that this interaction was driven by the youngest age group. Children aged 9–10 years found texts with highly general words significantly more confusing (EMM = 3.54, SE = 0.13) compared to texts with highly specific words (EMM = 4.02, SE = 0.13), t (54.6) = -3.28, p = .005. No other pairwise contrasts were significant (all ps > .05), and we did not find any significant differences between age groups across the three levels of text specificity (all ps > .05). This pattern suggests that sensitivity to lexical specificity in reducing confusion is present primarily in younger children and diminishes with age.
For Informativeness 1 (quantity), the model revealed significant main effects of specificity level, F (2, 36.32) = 4.03, p = .026, and age, F (2, 103.04) = 8.06, p < .001, as well as a significant interaction between the two factors, F (4, 1912.10) = 3.89, p = .004.
Post-hoc contrasts showed that overall specific texts (EMM = 3.35, SE = 0.09) were rated as significantly more informative than general texts (EMM = 3.05, SE = 0.09), t (36.4) = -2.84, p = .020. No differences were found between basic texts (EMM = 3.18, SE = 0.12) and either specific (t (36.4) = -1.28, p = .41) or general texts (t(36.4) = 1.03, p = .56).
Within age groups, children aged 9–10 found general texts (EMM = 3.05, SE = 0.118) as less informative than specific texts (EMM = 3.52, SE = 0.118), t(65.5) = -3.802, p < .001. No other contrasts reached significance (ps >. 06). Children’s ratings of informativeness differed significantly between age groups across all text levels. Basic texts were rated as more informative by children aged 9–10 (EMM = 3.40, SE = 0.147) than those aged 11–12 (EMM = 2.96, SE = .142), t(266) = 2.975, p = .009. General texts were judged as more informative by children aged 10–11 (EMM = 3.29, SE = 0.12) than 11–12 (EMM = 2.96, SE = 0.14), t(153) = 3.76, p < .001. Finally, specific texts were rated as less informative by children aged 11–12 (EMM = 3.02, SE = 0.11) compared to both 9–10-year-olds (EMM = 3.52, SE = 0.12), t(154) = 3.93, p < .001, and 10–11-year-olds (EMM = 3.51, SE = 0.12), t(153) = 3.80, p < .001. For Informativeness 2 (relevance), the model showed a significant main effect of specificity level, F(2, 36.71) = 3.26, p = .050, and a main effect of age, F(2, 103.98) = 6.41, p = .002. However, the interaction between specificity level and age was not significant (p = .645). Post-hoc contrasts revealed that specific texts (EMM = 3.65, SE = 0.07) were rated as conveying more important information than general texts (EMM = 3.43, SE = 0.07), with the difference approaching significance, t(36.4) = 2.39, p = .056. No other contrast reached significance (ps > .18).
5.4. Does the combination of lexical concreteness and specificity impact children’s perceived clarity and informativeness of written texts between the ages of 9 and 12?
While the previous analysis examined the independent contributions of concreteness and specificity to clarity and informativeness, lexical proprieties co-occur and interact in natural texts. To capture these combined effects, we decided to apply the fitted mixed-effects models including both concreteness and specificity as a unified text-type factor. These analyses allow us to examine how different lexical configurations jointly shape children’s perceptions of clarity and informativeness across age groups.
With respect to Clarity 1 (easiness), the analyses did not reveal robust overall effects of either text type (p = .063) or age (p = .188). However, we found a significant interaction between text type and age, F(10, 1905.76) =1.88, p = .044, indicating that children of different ages were differentially sensitive to combinations of concreteness and specificity. From post-hoc contrasts within age group, a consistent pattern emerged for the youngest and oldest children. In both groups, texts combining abstract and general lexical choices were experienced as particularly difficult, whereas texts characterized by concrete and specific vocabulary were perceived as easier to understand. Specifically, children aged 9–10 judged abstract – general texts as significantly more difficult than concrete – specific ones (EMM = 3.84, SE = 0.15; EMM = 4.49, SE = 0.15, respectively), t(49.4) = -3.44, p = .002, and the same contrast was observed among children aged 11–12, (EMM = 3.65, SE = 0.15; EMM = 4.26, SE = 0.15, respectively), t(46.3) = -3.26, p = .024 (see Figure 2). No other differences between text types reached significance within these age groups, all ps > .10.
Mean ratings for Clarity 1 (perceived easiness) as a function of text type (combining lexical concreteness and specificity) and age group. Points represent estimated marginal means (EMMs) derived from the mixed-effects model, and error bars indicate standard errors.
Figure 2. Long description
The Y axis is labeled Clarity 1 easiness, ranging from 1 not easy at all to 5 very easy. The X axis is labeled Age, with three categorical groups: 9-10, 10-11, and 11-12.
Each age group contains six vertical colored bars representing different text types. From left to right within each cluster, the categories are:
- abstract_basic (light blue with a solid line and circle marker)
- abstract_general (orange with a dashed line and triangle marker)
- abstract_specific (yellow with a dashed line and square marker)
- concrete_basic (green with a dash-dot line and plus marker)
- concrete_general (pink with a dotted line and crossed-square marker)
- concrete_specific (purple with a dash-dot-dot line and asterisk marker)
Data trends show that for the 9-10 age group, ratings are highest for concrete_specific (approximately 4.5) and lowest for abstract_general (approximately 3.8). As age increases to 11-12, most ratings converge or slightly decrease, with abstract_basic and abstract_specific ending near 4.0. Error bars extend vertically from each data point, and individual data points are scattered at the bottom of the graph near the 1 and 2 ratings. A legend at the bottom identifies the six text types by color and marker style.
Mean ratings for Informativeness 1 (perceived quantity of information) as a function of text type (combining lexical concreteness and specificity) and age group. Points represent estimated marginal means (EMMs) from the mixed-effects model, and error bars indicate standard errors.
Figure 3. Long description
The Y-axis is labeled Informativeness 1 (quantity) with a scale from 1 (very little information) to 5 (a lot of information). The X-axis is labeled Age and is divided into three categories: 9-10, 10-11, and 11-12.
Data is presented as colored vertical bars with overlaid points representing E M M s and lines connecting these points across age groups. The six text types are:
- abstract_basic: light blue bar, solid circle point.
- abstract_general: orange bar, triangle point.
- abstract_specific: yellow bar, square point.
- concrete_basic: green bar, plus sign point.
- concrete_general: pink bar, square with an X point.
- concrete_specific: purple bar, asterisk point.
Trends across age groups:
- abstract_basic shows a slight downward trend from approximately 3.4 at age 9-10 to 2.9 at age 11-12.
- abstract_general remains relatively stable around 3.4 to 3.5 across the first two groups before dropping to 2.9.
- abstract_specific stays near 3.4 for the first two groups and drops slightly to 2.9.
- concrete_basic starts at 3.4, dips slightly at 10-11, and ends at 3.0.
- concrete_general is the lowest rated across all groups, starting at 2.7, rising to 3.1, and ending at 2.7.
- concrete_specific is the highest rated, starting at 3.7, maintaining that level at 10-11, and dropping to 3.1 at 11-12.
Error bars for all categories generally span about 1 unit on the Y-axis.
Post-hoc contrast between age groups showed that children aged 10–11 rated abstract-general texts as easier to understand than children aged 11–12, t(240) = 2.87, p = .012. No other contrasts reached significance, all ps > .16.
Taken together, these findings indicate that clarity judgments are explained (also) by the combination of concreteness and specificity and that this effect is modulated by age. In particular, the combination of abstractness and generality appears to place higher processing demands on children, while concreteness paired with specificity consistently facilitates perceived ease of comprehension.
A different picture emerged for Clarity 2 (confusion). Analyses revealed only a marginally significant effect of text type on clarity, F(5, 33.98) = 2.48, p = .051, whereas neither age (p = .011) nor the interaction between age and text type (p = .14) reached significance. This suggests that while children were sensitive to lexical differences when explicitly judging how easy a text was to understand, these differences did not translate as clearly into feelings of confusion. In other words, easiness and confusion appear to capture different but not opposite aspects of perceived clarity (this is also reflected in a correlation coefficient of .57 which is medium and not high).
Unlike clarity, the analyses on Informativeness 1 (quantity) showed significant main effects of text type (F(5, 33.27) = 4.35, p = .004) and age (F(2, 103.05) = 8.03, p <.001), as well as a robust interaction between the two F(10,1906.42) = 2.49, p = .006) (see Figure 3). These results indicate that children’s evaluations of how much information a text contained depended on both what kind of words were used and who was reading the text.
Post-hoc contrasts showed that overall concrete-general texts (EMM = 2.85, SE = 0.11) were perceived as significantly less informative than concrete-specific texts (EMM = 3.47, SE = 0.11), t(33.3) = -4.55, p < .001. No other comparisons were significant, ps > .06.
Post-hoc contrast within age groups showed that children aged 9–10 found concrete-general texts (EMM = 2.73, SE = 0.14) significantly less informative compared to all other type of text: concrete-specific (EMM = 3.66, SE = 0.14), t(67.2) = -5.64, p < .001; concrete-basic (EMM = 3.43, SE = 0.18), t(67.2) = -3.47, p = .011; abstract-general (EMM = 3.37, SE = 0.14), t(67.2) = 3.87, p = .003; abstract-specific (EMM = 3.39, SE = 0.14) t(67.4) = 4.04, p = .002; and abstract-basic texts (EMM = 3.37, SE = 0.18) t(67.2) = 3.20, p = .025. Children aged 10–11 also found concrete-general texts significantly less informative than concrete-specific texts (EMM = 3.10, SE = 0.14, EMM = 3.63, SE = 0.14, respectively), t(122.5) = -3.231, p = .022. None of the other contrasts reached significance, ps > .10.
Moreover, we found that abstract-basic texts were perceived as significantly more informative by children 9–10 (EMM = 3.37, SE = 0.18) than children aged 11–12 (EMM = 2.88, SE = 0.17), t(541) = 2.68, p = .020. Abstract-general texts were judged as more informative by both 9–10-year-olds (EMM = 3.37, SE = 0.14) and 10–11-year-olds (EMM = 3.48, SE = 0.14) compared to 11–12-year-olds (EMM = 2.89, SE = 0.13), with t(264) = 3.17, p = .005 and t(264) = 3.879, p < .001, respectively. Similarly, abstract-specific texts were rated as more informative by both children aged 9–10 (EMM = 3.39, SE = 0.14) and 10–11 (EMM = 3.39, SE = 0.14) than 11–12-year-olds (EMM = 2.91, SE = 0.13), t(265) = 3.24, p = .004 and t(264) = 3.20, p = .004, respectively. Concrete-general texts were perceived as more informative by children aged 10–11 (EMM = 3.10, SE = 0.14) than 11–12-year-olds (EMM = 2.70, SE = 0.13), t(265) = 2.64, p = .024. Conversely, children aged 9–10 (EMM = 2.73, SE = 0.14) rated these texts as significantly less informative than children aged 10–11, t(264) = -2.36, p = .049.
Concrete specific texts were rated as more informative by both 9–10-year-olds (EMM = 3.66, SE = 0.14) and 10–11-year-olds (EMM = 3.63, SE = 0.14) than children aged 11–12 (EMM = 3.12, SE = 0.13), t(265) = 3.57, p = .001 and t(265) = 3.40, p = .002, respectively. Finally, there were no significant differences across age groups for concrete-basic texts, all ps >. 07.
For Informativeness 2 (relevance), we found significant main effects of text type (F(5, 33.69) = 2.7898, p = .03), and age (F(2, 104) = 6.3992, p = .002). However, the interaction between text type and age does not reach significance (p = .71). Post-hoc contrast on type of types showed that concrete-general texts (EMM = 3.32, SE = 0.10) were perceived as including less relevant information compared to concrete-specific texts (EMM = 3.75, SE = 0.10), t (33.3) = -3.47, p = .017. No other contrasts reach significance; all ps > .13.
6. General discussion
The present study investigated how lexical properties affect children’s subjective evaluations of textual comprehensibility, focusing specifically on perceived clarity and informativeness. By examining the effects of lexical concreteness and specificity across different age groups, the study aimed to move beyond traditional readability metrics and provide a cognitively grounded account of how young readers perceive the complexity of written texts.
After confirming that clarity and informativeness are only partially correlated constructs, we examined the contribution of different lexical properties to each dimension separately. While clarity primarily reflects processing ease and lack of confusion, informativeness concerns the amount and relevance of the information conveyed. This conceptual distinction is further supported by the empirical findings of the present study, which show that the two dimensions are differentially sensitive to lexical parameters and follow partially independent developmental trajectories.
Lexical concreteness, as hypothesized, significantly influences children’s perceived clarity of written texts between the ages of 9 and 12. Texts containing a higher proportion of concrete words were consistently perceived as clearer than texts containing more abstract vocabulary. This effect emerged robustly across both measures of clarity (perceived ease of understanding and perceived confusion) and was independent of age. The absence of age-related interactions suggests that, once lexical representations are consolidated, the processing advantage associated with concreteness persists across late childhood. In the age range considered here, concreteness therefore appears to function as a stable lexical property that facilitates semantic access and reduces cognitive effort, independently of ongoing developmental changes. Importantly, reduced cognitive effort should not necessarily be interpreted as universally advantageous from an educational perspective. While concrete language may facilitate semantic access and increase perceived clarity, more cognitively demanding linguistic input, including abstract language, may in some contexts promote reflection, inferencing, curiosity, and deeper learning. Recent work, for instance, suggests that abstract language is often associated with exploratory and epistemically open communicative contexts (Mazzuca et al., Reference Mazzuca, Villani, Lamarra, Bolognesi and Borghi2025). Our findings, therefore, concern children’s subjective perceptions of clarity and informativeness during reading, rather than objective learning outcomes or long-term retention. This pattern is consistent with evidence showing that concrete concepts are typically acquired earlier than abstract ones and tend to remain more readily accessible in processing even after both types of concepts are established in the lexicon (Banks et al., Reference Banks, Borghi, Fargier, Fini, Jonauskaite, Mazzuca, Montalti, Villani and Woodin2023).
These findings are well aligned with a substantial body of psycholinguistic research showing that concrete and abstract concepts are processed differently in the human cognitive system. A large number of studies have demonstrated that concrete concepts are processed faster and more accurately than abstract ones in different tasks, such as lexical decision and word naming (e.g., Binder et al., Reference Binder, Westbury, McKiernan, Possing and Medler2005; James, Reference James1975; Kroll & Merves, Reference Kroll and Merves1986; Paivio, Reference Paivio1986; Schwanenflugel & Stowe, Reference Schwanenflugel and Stowe1989). This processing advantage has been attributed to the richer representational structure of concrete concepts, which are grounded in perceptual and sensorimotor experience. Within this framework, Dual Coding Theory (Paivio, Reference Paivio1971, Reference Paivio1986) provides a principled explanation for the observed effects. Concrete words benefit from both verbal and imagistic representations, whereas abstract words rely primarily on linguistic codes. This dual representational format facilitates semantic access and reduces processing effort, leading to faster comprehension and lower cognitive load. However, reduced processing effort should not be equated with universally superior learning conditions. Some degree of cognitive and inferential effort may support deeper processing, reflection, and stronger memory encoding. Educational research has suggested that overly simplified instructional materials may reduce depth of processing and negatively affect retention and learning outcomes (Land et al., Reference Land, Sanders and van den Bergh2009). More generally, certain forms of linguistic complexity may promote greater cognitive engagement and semantic elaboration, even when they are perceived as more effortful. In the context of text comprehension, this translates into increased perceived ease and reduced confusion, as readers can more readily construct coherent mental representations of the described content and associate it with contextual and situational information than abstract concepts (Davis et al., Reference Davis, Altmann and Yee2020; Schwanenflugel, Reference Schwanenflugel1991). Crucially, while concreteness significantly impacted the perceived clarity, it did not impact textual perceived informativeness. This dissociation suggests that concreteness primarily affects how information is processed rather than how much information is perceived to be conveyed. While concrete words facilitate semantic access and reduce cognitive effort, they do not necessarily increase the amount of information conveyed.
Turning to our second research question, the results provide clear support for our second hypothesis, showing that lexical specificity plays a significant role in shaping children’s judgments of informativeness. Texts containing more specific lexical items were perceived as more informative than texts based on more general vocabulary. Crucially, however, and in contrast to lexical concreteness, specificity did not exert a significant effect on children’s perceived clarity of written texts, either in terms of perceived ease of understanding or perceived confusion.
Overall, lexical specificity does not directly affect the perceived complexity of a text, at least from a subjective and metacognitive standpoint. One possible explanation for this pattern is that lexical specificity does not directly operate by making texts easier or harder to process in absolute terms, but rather by shaping how precisely the referent of interest is identified. More specific terms may increase semantic content and informational load, yet at the same time, they delineate the intended referent more sharply, thereby constraining interpretation. As a result, the increased semantic richness associated with specificity does not translate into a perception of increased perceived difficulty, nor into a systematic reduction of processing effort. In our results, the effects of specificity emerge not on perceived ease, but selectively on perceived confusion, and only for younger readers. For them, general terms may leave interpretative gaps that they are less able to resolve inferentially, leading to higher levels of confusion. Conversely, specific lexical choices reduce referential ambiguity and support interpretation by narrowing the range of possible meanings.
As predicted, specificity primarily affected informativeness judgments. Across analyses, specific texts were generally rated as more informative than general ones, particularly with respect to perceived quantity of information, and this effect was especially pronounced among younger children. These findings support our second hypothesis and indicate that specificity enhances perceived informativeness by sharpening referential precision.
At the same time, the effects of specificity on informativeness were strongly modulated by age. Younger children (9–10 years) consistently judged specific texts as more informative than general ones, suggesting that increased referential detail helps them extract and value information. In contrast, older children (11–12 years) systematically assigned lower informativeness ratings across all specificity levels, including specific texts. This developmental shift could indicate a change in evaluative criteria: whereas younger children may equate informativeness with explicitness or detail, older children appear to adopt a more selective standard, judging texts as informative only when they provide conceptually relevant content. This interpretation is further supported by the complementary pattern observed across the two informativeness measures. Older children tended to assign lower ratings for quantity of information, but higher ratings for relevance of the information, suggesting a transition from quantity-based to relevance-based evaluations of informativeness. Thus, age-related differences in informativeness judgments do not reflect reduced comprehension, but rather more refined metacognitive standards. At the same time, we acknowledge that developmental differences in informativeness ratings may partly reflect differences in how children interpret the notion of “information” itself. The construct may therefore not be fully stable across age groups, representing a potential limitation of construct validity in the present task.
An additional finding worth discussing is that basic texts did not emerge as uniformly more informative than either general or specific texts. Instead, their perceived informativeness was developmentally constrained: basic-level texts were rated as more informative by younger children but not by older ones.
This pattern suggests that while basic-level categories may be cognitively privileged for conceptual access, as proposed for instance in classic works by Rosch (Reference Rosch, Rosch and Lloyd1978), this privilege does not straightforwardly extend to subjective judgments of informativeness, especially as readers become more experienced. As school-children grow older, basic-level labels may increasingly be perceived as insufficiently informative.
Finally, our third research question addressed the effect of concreteness and specificity combined on the perceived clarity and informativeness of the texts. In relation to clarity, results revealed that texts combining abstract and general vocabulary were consistently experienced as less clear, whereas texts characterized by concrete and specific lexical choices were perceived as easiest to understand, particularly by the youngest and oldest children. This suggests that abstractness and generality jointly increase processing demands, while concreteness paired with specificity provides complementary support for comprehension.
For informativeness, the combined analyses reinforced the central role of specificity. Concrete-general texts were repeatedly judged as the least informative, particularly by younger children, indicating that concreteness alone is insufficient to support informativeness judgments in the absence of referential precision. Conversely, concrete-specific texts consistently received high informativeness ratings, highlighting the additive value of specificity in enriching perceived informational content. No significant effects related to abstract (specific or general) concepts on informativeness were observed. When age is considered, the situation becomes more complex, as overall, the results suggest age-related differences in how children perceive the informativeness of texts. Younger children (9–10 and 10–11 years) generally evaluated texts as more informative than older children (11–12 years), particularly when texts contained abstract vocabulary. These findings suggest that children’s perceptions of textual informativeness evolve with age, possibly reflecting the development of more critical or selective evaluation of information in older readers. From an educational perspective, the higher perceived relevance of concrete-specific texts indicates that combining concreteness with specificity may facilitate children’s perception of informative content. Designing instructional materials that balance concreteness and specificity may therefore support children’s engagement with and understanding of written texts.
7. Conclusions
In this study, we investigated the relationship between two lexical variables (concreteness and specificity) and measures related to text readability and comprehensibility in children. Given that many traditional readability formulas rely primarily on quantitative parameters, it is crucial to complement these measures with speakers’ subjective evaluations of textual clarity and informativeness. Focusing on the perceptions of learners during a critical phase of language and literacy development can have a substantial impact on educational practice, as it may help identify and overcome some of the most common obstacles to learning.
Overall, the findings demonstrate that clarity and informativeness represent distinct, though related, dimensions of textual quality.
Lexical concreteness emerged as a robust and age-independent predictor of perceived clarity, consistently enhancing ease of comprehension and reducing confusion. In contrast, lexical specificity primarily influenced judgments of informativeness, particularly with respect to the perceived quantity and relevance of information, and showed a more pronounced developmental modulation.
Taken together, these results indicate that different lexical properties support different aspects of text comprehension: concreteness facilitates semantic access and processing ease, whereas specificity contributes to referential precision and informational richness. Importantly, the interaction between lexical properties and age highlights developmental changes in how children evaluate texts, suggesting a gradual shift from quantity-based to relevance-based standards of informativeness.
From an applied perspective, these findings have direct implications for educational practice and text design. Adjusting lexical choices to increase concreteness may improve clarity across age groups, while carefully calibrating specificity can enhance informativeness, particularly for younger readers. More broadly, the results underscore the importance of considering multiple, cognitively grounded dimensions of readability when developing instructional materials for children.
Future studies should also investigate whether the present findings extend to more ecologically valid reading contexts, including authentic educational and narrative materials characterized by richer discourse structures and more natural pragmatic contexts, as well as disentangling the respective contributions of lexical specificity, familiarity, frequency, and perceived lexical difficulty to children’s evaluations of texts. Such approaches would further clarify the mechanisms through which lexical properties shape text comprehension and support the development of more effective, developmentally sensitive readability models.
Data availability statement
All data and scripts are available at https://osf.io/3hvjn/.
Acknowledgments
The authors would like to thank Paola Vernillo and Andrea Amelio Ravelli, members of ABSTRACTION research group (GRANT AGREEMENT: ERC-2021-STG-101039777), for comments and discussions. The authors would like to thank all the students and teachers of the IC7 public school in Imola (BO) for their participation in this study.
Author contribution
Conceptualization: L.O., C.V., T.L., A.L., G.R., M.M.B.; Data curation: L.O., C.V., T.L., A.L., G.R., M.M.B.; Funding acquisition: M.M.B.; Investigation: C.V., A.L., M.M.B.; Methodology: L.O., C.V., T.L., A.L., G.R., M.M.B.; Project administration: M.M.B.; Supervision: M.M.B.; Writing – original draft preparation: L.O., C.V., M.M.B.; Writing – review & editing: L.O., C.V., T.L., A.L., G.R., M.M.B. A one-page Accessible Summary of this article in non-technical language is freely available in the Supporting Information online and at https://oasis-database.org.
Funding statement
L.O., C.V., T.L., G.R. and M.M.B. were funded by the European Research Council funded project ABSTRACTION, (grant agreement: ERC-2021-STG-101039777). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council Executive Agency. Neither the European Union nor the granting authority can be held responsible for them. Adele Loia was funded by PNRR – PRIN 2022 (2022EPTPJ9) “WEMB: Word Embeddings from Cognitive Linguistics to Language Engineering and back”.
Competing interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Open access
