Gender Stereotypes in the Physical Depictions of Girls in Turkish Children’s Literature: A Collocational Analysis

Collocational Analysis in Gender Studies

In corpus linguistics, collocation is generally defined as the habitual co-occurrence of two or more words (Firth, 1957) or, from a statistical viewpoint, as a lexical item’s significantly frequent co-occurrence with others in a given textual context beyond what would be expected by chance (Sinclair, 1987). More than mere proximity, collocation involves meaningful and structural associations between words, contributing to both the construction and reflection of meaning (Halliday, 1961). These co-occurring items, known as collocates, play a crucial role in shaping the meaning of the target word within specific contexts, and as such, collocation has come to be seen as “the key to meaning” (Firth, 1957).

Collocational analysis is fundamentally defined as the study of the tendency of words to occur together within a text (Stubbs, 2002). Such analysis can only be conducted using either ready-made corpora or specially compiled digital corpora designed for specific research purposes. Major online reference corpora often provide built-in collocation tools; however, researchers working with self-compiled corpora must rely on concordance software such as AntConc (Anthony, 2024) or WordSmith Tools (Scott, 2024). In both cases, the researcher enters a node word—the keyword determined by the research questions—into the system and can observe its occurrences within the corpus through the concordance interface.

Although the data are retrieved through computational methods, the decision regarding which collocates to evaluate ultimately depends on the researcher’s analytical judgment. While frequency-based approaches are commonly adopted in collocation studies, there is no full consensus on what exactly constitutes a collocational relationship. Depending on the research scope, questions, and corpus size, some scholars may even consider single observed co-occurrences as collocational. At this stage, it is crucial that the criteria for inclusion be clearly defined and applied consistently (Baker, 2014).

The next step after identifying collocates is interpretation, which shifts the analysis from a purely computational process to a human-centered one. To ground interpretations scientifically, researchers are expected to compare their findings with relevant statistical data and prior scholarship and to interpret them within established theoretical frameworks or approaches. This stage therefore requires thorough preliminary work and conceptual preparation (Baker, 2014). The collocation analysis method, by systematically examining which words co-occur with others in a given context, reveals not only surface-level frequencies but also implicit semantic, grammatical, and discursive associations between words. This analytical power has made collocation analysis a prominent tool across fields including corpus linguistics, lexicography, language education, literary analysis, and sociolinguistics. Since the early 2000s, this method has also been increasingly employed in studies that investigate how gender is represented and reproduced through language. By analyzing recurrent word associations in various types of texts, collocation analysis enables researchers to uncover the linguistic traces of gender-related stereotypes, normative roles, and ideological tendencies, thereby establishing itself as a powerful analytical framework in the field of gender and language.

One of the earliest gender studies to apply collocation analysis was conducted by Pearce (2008). In this pioneering study, Pearce (2008) examined the collocations of the words woman and man in the British National Corpus, finding that both genders are often stereotyped, with men being represented in terms of power, physicality, and violence, and women being depicted in terms of their relationships with men or their ability to have children. Subsequent studies have explored similar gendered norms, including those where men are evaluated based on their roles and status in society while women are analyzed in terms of their appearance and sexuality (Caldas-Coulthard & Moon, 2010); where women and men are associated with different actions (Herdağdelen & Baroni, 2011); and even where the polarity of emotions differs by gender (Norberg, 2012). Findings of women’s marginalization in certain contexts have also emerged (Aull & Brown, 2013). Among these contributions, Paul Baker’s (2014) work is particularly notable for demonstrating how corpus-based tools can be systematically applied to analyze gendered discourse across diverse linguistic contexts. In particular, his study of the words girl and boy using Sketch Engine exemplifies how collocation analysis can reveal implicit ideologies—especially through verb associations, adjective patterns, and syntactic roles such as subject and object.

These studies are important not only because they highlight the contribution of collocation analysis to gender studies but also because they demonstrate how different collocations can yield different results when analyzed across various corpora, emphasizing the significance of corpus selection in research (Herdağdelen & Baroni, 2011). Furthermore, it is noteworthy that the collocational features of girl are more frequently altered than those of boy (Macalister, 2011).

Methodology

This study examines how gender representations are constructed at the linguistic level through the physical descriptions of the girl figure in Turkish children’s literature, using a corpus-based collocational analysis. The procedure follows corpus-driven approaches to gender research (Baker, 2014) and is interpreted within the frameworks of subject positioning (Davies, 1993), performative repetition (Butler, 1990), and normalization (Foucault, 1977). These frameworks guide the interpretation of recurring collocational patterns as indicators of discursively stabilized gendered expectations.

Distribution of Physical Descriptors

The analysis of the Turkish Children’s Literature Corpus (TCLC, n.d.) shows that the lexical item kız (girl) occurs approximately 33,000 times, ranking 70th among the most frequent words in the corpus. This frequency is notably higher than its 141st position in general Turkish usage (Aksan et al., 2017), indicating the relatively high linguistic visibility of girl characters in children’s literature.

A total of 14,077 adjectives were identified as collocates modifying kız. Among these, 5,490 items (39%) were classified as physical descriptors according to the criteria outlined in the methodology. The remaining adjectives referred to non-physical domains such as personality, behavior, or evaluation and were therefore excluded from the present analysis.

The physical descriptors were grouped into semantic subcategories using the USAS semantic framework. Table 1 presents the distribution of these categories.

OPEN IN VIEWER

Table 1.

Semantic Subcategories and Frequency of Adjectives Collocating With the Word kız.

Subcategory	n	Ratio (%)
Age	3,303	60
Judgment of appearance	1,432	26
Body parts	459	8.4
Clothing, Accessories, and Make-Up	132	2.4
Weight	98	1.8
Height	42	0.8
Cleaning and personal care	24	0.4
Total	5,490	100

As shown in Table 1, age-related descriptors constitute the largest category, accounting for 60% of all physical adjectives (n = 3,303). These are followed by judgments of appearance, which represent 26% of the dataset (n = 1,432). Descriptors referring to specific body parts make up 8.4% (n = 459), while clothing (2.4%), weight (1.8%), height (0.8%), and personal care (0.4%) together account for the remaining 5.4% of the dataset.

This distribution shows that physical descriptions of the girl figure are concentrated primarily in two domains: developmental stage (age) and aesthetic evaluation. In the following sections, categories of hair, eyes, skin, weight, and height are analyzed together under the theme of “The Idealized Beauty Prototype” (“The Aesthetic Prototype: Beauty Judgments and Body Ideals” section) to provide a holistic view of recurring physical patterns.

Age-Based Representation of the Girl Figure

Age-related adjectives constitute the largest category of physical descriptors (60%, n = 3,303), indicating that developmental stage functions as a primary organizing principle in representations of the girl figure. Within the age-related set, descriptors cluster into three phases: childhood (66%, n = 2,180), adolescence (31%, n = 1,024), and adulthood (3%, n = 99). In addition to these size- and age-marking adjectives, numerical age adjectives provide a more explicit picture of the age ranges foregrounded in the corpus. As shown in Table 2, girl characters are predominantly portrayed within the 3 to 18 age range, while infancy (0–2 years) and adulthood are comparatively rare.

OPEN IN VIEWER

Table 2.

Developmental-Period Classification of Numerical-Age Adjectives Collocating With the Word kız.

Age range	n	Ratio (%)
Infancy (0–2 years)	3	4
Early childhood (3–6 years)	22	26
Childhood (7–11 years)	30	35
Adolescence (12–18 years)	26	31
Adulthood (20s–70s)	4	4
Total	85	100

The following subsections examine how physical description, spatial association, and attributed roles shift across these developmental stages.

The Aesthetic Prototype: Beauty Judgments and Body Ideals

Physical descriptors related to appearance cluster around a limited set of facial and bodily features that collectively form a recurring beauty prototype in the corpus. Rather than appearing as isolated traits, these features tend to co-occur, producing a relatively homogeneous image of the girl figure across the corpus.

Marginal Physical Categories

Compared to the two sections discussed above, the girl figure co-occurs less frequently with adjectives related to other aspects of physical appearance, including clothing (51%, n = 80), accessories (29%, n = 45), personal care (15%, n = 24), and make-up (5%, n = 7). Within these categories, no distinctive pattern has been identified beyond clothing types and colors.

In clothing-related descriptions, dresses constitute the most frequently mentioned category (28%, n = 22), followed by smaller groups such as outerwear, trousers, school uniforms, and skirts.

Color adjectives occur frequently in clothing descriptions. The most common color categories are white (23%, n = 15), red (17%, n = 11), and blue (15%, n = 10), while pink appears less frequently (9%, n = 6). In many instances, color terms function not merely as modifiers of specific garments but as central elements in the visual portrayal of the girl figure:

Cross-Cultural Patterns and the Turkish Context

Comparative research on gender representations in children’s literature indicates that gender norms tend to display recurring structural patterns across cultural contexts, even when their specific manifestations vary (Adam & Harper, 2023). Moattar’s (2010) comparison of children’s books from Sweden and Iran illustrates this continuity: while Swedish narratives more frequently foreground female agency and Iranian texts more often position girls within domestic and marriage-oriented roles, both contexts retain underlying androcentric narrative structures.

The patterns identified in the Turkish Children’s Literature Corpus largely align with these international tendencies: the girl figure is predominantly constructed through physical descriptors emphasizing age, appearance-based evaluation, and bodily norms across an innocence–beauty–marriageability continuum. The equation of beauty with goodness and narrative reward has similarly been documented in Western fairy tales and children’s books (Baker-Sperry & Grauerholz, 2003; Hamilton et al., 2006), while the positive portrayal of thinness and tallness, coupled with the negative framing of larger body types, corresponds to findings in content analyses of children’s media across cultural settings (Brugeilles et al., 2002; Herbozo et al., 2004). The Turkish corpus thus appears to reflect widely observed representational conventions rather than an isolated national pattern.

Subject Positioning, Performativity, and Normalization

From the perspective of Davies’s (1993) subject-positioning framework, the recurring collocational clusters identified in this study establish a limited range of position options for the girl figure. The systematic co-occurrence of adjectives referring to beauty, slenderness, height, and light-colored features stabilizes expectations concerning how a girl should look and be evaluated. These patterns contribute to the discursive organization of normative subject positions. The repeated association between physical attractiveness and positive moral traits can be interpreted, in line with Butler (1990), as performative reinforcement. In Foucauldian terms, such regularities constitute processes of normalization, operating through the recurring presentation of a narrow aesthetic standard as natural and desirable (Foucault, 1977).

Body-related descriptors reinforce this pattern: the distributional asymmetry that privileges thinness and tallness in positive contexts, while associating larger body types with evaluative or corrective environments, resonates with feminist analyses of bodily discipline and the beauty regime (Bartky, 1990; Bordo, 1993)—a regulation that emerges through patterned co-occurrence and evaluative framing.

Demographic Divergence and the Construction of a Narrow Prototype

The corpus reveals marked divergence between literary prototypes and Turkish demographic characteristics. While dark hair and brown eyes predominate in the Turkish population (62% and 70%, respectively; Nalçaoğlu, 2012), the corpus disproportionately foregrounds light-colored features—blonde hair (58%) and blue eyes (56%)—in positive evaluative contexts. Blonde hair is thus overrepresented nearly 15-fold, and blue eyes more than fourfold, relative to population prevalence. Similarly, thinness and tallness are privileged despite demographic realities in which substantial proportions do not conform to these body types (TURKSTAT, 2019).

Comparable tendencies have been documented in studies of Chinese children’s books, where lighter skin tones and patterns described as “white preference” have been observed in award-winning titles (Yang et al., 2022). The Turkish data therefore align with broader cross-cultural patterns in which lightness and specific facial features function as recurring aesthetic markers. Beyond physical traits, the corpus shows that beauty becomes increasingly associated with marriageability in adolescence and adulthood. Similar linkages between female maturation and romantic or marital trajectories have been reported in East Asian contexts (Charlesworth et al., 2021), while analyses of Western fairy-tale corpora likewise demonstrate the centrality of romantic resolution (Baker-Sperry & Grauerholz, 2003). Recent research suggests a partial shift in some Western narratives toward representations emphasizing agency and self-sufficiency (Clarke et al., 2024).

Implications for Self-Worth and Bodily Diversity

Although the present corpus-based design does not permit direct claims about reader effects, prior research in body image and objectification theory provides a relevant interpretive framework. Studies indicate that repeated exposure to narrow aesthetic ideals is associated with appearance-based self-evaluation, body dissatisfaction, and diminished self-worth among children (Dittmar et al., 2006; Fredrickson & Roberts, 1997; Tiggemann & Slater, 2004).

In the present corpus, physical appearance is not presented as a neutral descriptor but is systematically linked to moral evaluation. Beauty co-occurs with virtue-related traits, while ugliness appears in negatively evaluated contexts (Section 2.3.3), establishing a dual framework in which children whose bodies diverge from the prototype may encounter implicit messages that position them as inadequate on both aesthetic and moral dimensions. Given that the idealized features identified in the corpus—blonde hair, blue eyes, slenderness, and tallness—are not representative of the broader demographic profile of Türkiye, this narrowing has implications for the representation of bodily diversity.

Rather than promoting diversity or self-acceptance, the corpus constructs a narrow ideal that implicitly positions many Turkish children as deviations from what may be perceived as a culturally foreign standard, communicating implicit messages about desirability, belonging, and social value to those whose characteristics align with Türkiye’s majority demographic profile. This normalization operates not through explicit instruction but through the repeated naturalization of a limited aesthetic range as self-evidently desirable—a process through which children’s literature actively participates in the reproduction of beauty norms rather than merely reflecting them (Bartky, 1990).

Vulnerability, Constrained Agency, and Developmental Trajectories

In addition to aesthetic norms, the corpus reveals recurring patterns in which girls are positioned as vulnerable figures requiring protection. Childhood representations emphasize dependency, while later stages link physical maturation to marriageability and behavioral regulation (Sections 2.2.1–2.2.3). This developmental sequencing—innocence, appearance-centered evaluation, and relational positioning—constructs a trajectory in which physical appearance and relational status are foregrounded over autonomy and capability.

National statistical reports document the continued presence of early marriage and protection-related concerns regarding girls in Türkiye (TBMM, 2010; TURKSTAT, 2021). While the corpus design does not establish causal connections between literary representation and social outcomes, the recurrence of such motifs suggests that children’s literature may participate in the discursive circulation of gendered expectations, naturalizing early marriage as a normative developmental outcome. This positioning functions as normalization: the repeated presentation of limited subject positions as natural and inevitable constructs a girlhood defined less by possibility than by constraint—a trajectory oriented toward fulfillment of gendered social roles rather than autonomy.

Directions for Future Research

The present findings open several avenues for further investigation. Future research could extend the present design by examining additional linguistic dimensions, such as verb-based agency and syntactic positioning, and by integrating visual analysis in multimodal studies of picture books. Longitudinal comparisons across time periods, publishers, and genres would help determine whether the identified patterns are stable or undergoing transformation. Reader-response studies examining how children interpret and negotiate these representational norms would provide valuable insight into reception processes and the extent to which literary prototypes influence self-concept and identity formation.