Abstract
Textbooks constitute important learning tools in foreign language education and should depict ethnic and gender diversity. This article responds to calls for intersectional textbook research and addresses the following question: How diverse are representations of wo/men of colour in comparison with white wo/men in sparsely researched Dutch language textbooks? We performed content analysis of six frequently used textbooks for learning Dutch, examining pictures, texts and (audio-recorded) dialogues. Results show an underrepresentation of People of Colour (PoC) in comparison with demographic statistics. In contrast with white wo/men, PoC are featured in a limited range of low-status roles (e.g. migrant), except for in one textbook. Countries and cities are Eurocentric and Netherlands-centric, there is little variety in terms of accented speech in dialogues, and non-native speakers utter sentences that frame them as dependent. These findings build on the body of (Anglocentric) textbook research from a cultural studies’ perspective, help determine the degree to which educational materials meet diversity demands in language education policies, and what alternative ‘realities’ could arise.
Keywords
I Introduction
Educational guidelines increasingly highlight the need for promoting diversity and inclusion (UNESCO, 2020), also in the context of foreign language education. The Common European Framework of Reference for Languages (CEFR) (Council of Europe, 2020, p. 278), for example, states that foreign language education should ‘protect linguistic and cultural diversity, promote plurilingual and intercultural education, reinforce the right to quality education for all, and enhance intercultural dialogue, social inclusion and democracy.’ Research supports these calls in terms of showing effects of pictures in educational materials or inclusive language use in educational settings on the achievements of different students. Good et al. (2010), for example, found that girls in American high schools achieved higher levels of science comprehension when viewing pictures of female scientists compared with pictures of male scientists. In another study, Good et al. (2020) showed that students of colour in America showed greater math/science comprehension in a lesson appreciative of multiculturality as compared with a lesson from a colour-blind perspective. White students’ performance remained stable across conditions.
One pathway towards more inclusive foreign language education, or education that ‘values the unique contributions students of all backgrounds bring to the classroom’ (UNICEF, 2022), is thus providing students with diverse visual and linguistic content in educational materials/textbooks (UNESCO, 2017). Textbooks are indeed an ideal platform to examine what perspectives on ethnic and gender diversity are being spread in foreign language education (Harklau, 1999; Osler & Starkey, 2000; Risager, 2021). Despite the contemporary plethora of (digital) learning materials, textbooks remain dominant tools for helping teachers structure their courses and for directing students towards the knowledge, values and norms that (do not) matter in particular speech communities. Moreover, they generate specific patterns of perceiving the world (Fuchs & Bock, 2018).
Textbooks for foreign language education aspire to equip their readers with the knowledge and skills needed to navigate certain speech communities, and they supply readers with texts and images in support of this goal (Atkinson, 2021; Derakhshan, 2021; Risager, 2018). While previous research has shown that many language teaching materials are shaped by nationalist perspectives, fewer studies have empirically approached educational materials from a cultural diversity perspective (Quist & Sas, 2011). Such research, inspired by cultural studies (Hall, 2017; Sefton-Green, 2011) and Critical Language and Race Theory (Crump, 2014; Kubota, 2020), may reveal various forms and degrees of cultural diversity in learning materials that are often employed with diverse learner groups in a globalized world (Keles & Yazan, 2023; Kuramalivedu, 2008). Moreover, such research can help determine the degree to which current educational materials meet policy demands that call for diversity in textbooks (Council of Europe, 2020; UNESCO, 2020). The present study adds empirical insights to textbook research from a cultural studies’ perspective, taking these diversity calls into account.
Our study focuses on sparsely researched textbooks for learning Dutch, which is an official state language in six countries – the Netherlands, Belgium, Surinam, Aruba, Curaçao and Sint Maarten – and has 24 million speakers. Dutch is taught as a foreign language in more than 130 countries worldwide (Koster, 2019; Taalunie, 2023a); for example, in the United Kingdom, Germany, the Czech Republic and Indonesia. The relatively large, diverse language community that should be represented in learning materials and the 15.000 foreign language students that are confronted with such representations justify our research focus (Taalunie, 2023b). We focus our analyses on six frequently used textbooks developed for the largest group of learners: beginning, adult learners of Dutch. 1 The sociocultural diversity in different countries where Dutch is spoken, however, poses challenges for textbook writers and publishers: What and whose ‘realities’ should be represented, and how? (Sunderland, 2015). Demographic statistics offer one such perspective. de Bruijn et al. (2020), for example, found that ethnic diversity in Dutch children’s books, another educational genre, did not match with current demographic diversity in the Netherlands: Although 25.2% of the population has a non-western migration background (Centraal Bureau Statistiek, 2021), white characters were overrepresented in comparison to People of Colour (PoC). Further, when considering gender as a factor, they found that girls of colour had less prominent roles in storylines of the examined books than did white boys and girls or boys of colour.
Intersectionality, or the joint analysis of ethnicity and gender, has become a well-known concept since it was coined by Crenshaw (1989). It is therefore striking, as noted by Niehaus (2018), that few studies have operationalized this concept in (language) textbook research (but see Knudsen, 2006). We do find studies that examined these categories separately, mainly in English language textbooks (Alter et al., 2021; Yamada, 2011), but also in German, Burmese, Taiwanese, Iranian and, only recently, Dutch language textbooks (Baghdadi & Rezaei, 2015; Ho, 2021; Ilett, 2009; Koster & Litosseliti, 2021; Treadwell, 2014). Research into ethnic diversity in language textbooks has revealed that ethnic minorities in countries of reference are generally underrepresented and are depicted in low-status roles (Chisholm, 2018). Treadwell (2014), for example, found in her corpus of five Burmese reading textbooks only 19 references to 10 of the 135 ethnic minorities in Burma, which she classified as stereotyping. Further, gender research (Mills & Mustapha, 2015) into language learning materials suggests that women are underrepresented as compared to men, especially in domains such as work life. Koster (2020), for example, found that women are backgrounded in Dutch language textbooks, as female professional names appear less frequently than male professional names, and gender-neutral professions refer more often to men than to women.
Against this background, our research question is as follows: How diverse are representations of wo/men of colour and white wo/men in Dutch language textbooks? Based on previous studies, we expect that our analyses will reveal that (male) white characters are overrepresented as compared to (female) characters of colour. We expect PoC to be featured in limited range of roles with low-status (e.g. jobseeker, waiter, domestic worker). We also expect to find little content that supports enhancement of ‘intercultural dialogue’ by presenting non-European perspectives, and little ‘linguistic diversity’ (Council of Europe, 2020), in terms of variation on standard Netherlandic-Dutch in dialogues. It is not our aim to criticize individual textbook authors or publishers, nor to impose fixed norms. Rather, we aim to build on (Anglocentric) theoretical and educational debates by documenting what ‘realities’ are represented in frequently used textbooks in an under researched context, that is Dutch foreign language education, and discuss what alternative ‘realities’ could arise.
II Theoretical background
Risager (2018) describes five theoretical perspectives in representation research in language textbooks: national studies, citizenship education studies, postcolonial studies, transnational studies and cultural studies. The present article contributes to the latter field, which theorizes about how we categorize, identify and describe complex communities (Sefton-Green, 2011). We document how this is done in Dutch language textbooks, which we view as a medium of social observation within a network of media (Fuchs & Bock, 2018), and to what degree this genre features diversity in terms of describing complex ethnic and gender identities. Our article thus builds on empirical research in educational and media studies that match cultural studies’ theory (Block & Gray, 2017; Ros i Solé et al., 2020). Here, culture is not viewed as fixed but as constructed in discourse.
Along the same post-structural lines, we define our central intersection under investigation. We view ethnicity as ‘a doing – a dynamic set of historically derived and institutionalized ideas and practices that allow people to identify, or be identified, with groupings of people on the basis of presumed, and usually claimed, commonalities [. . .]: language, [. . .] nation [. . .] physical appearances’ (Markus & Moya, 2010, p. 22). We also define gender as a social construction, which is non-stable and performed (Butler, 1990). This article, however, does make use of binary categories such as PoC and white individuals, female and male. We rely on this ‘strategic essentialism’ (Spivak, 1978) as an aid to document potential imbalances, which may open up discussions about alternatives. Moreover, it is not unlikely that many textbook users employ these same binary categories while using their textbooks (Alter et al., 2021).
1 Ethnic diversity in textbooks
Whereas the study of representations of race or ethnicity in educational media has a long tradition (Chu, 2015; Harklau, 1999; Ho, 2021), few studies have examined ethnic representations in intersection with factors such as gender (Chisholm, 2018). In the following, we discuss relevant studies whose methodological foci, based on pilot analyses, were integrated into our analyses.
Chisholm (2018) noticed that most textbook studies on ethnicity combine both quantitative and qualitative methods, such as the study by Ilett (2009) on 17 German language textbooks (2004–2009) for both beginners and advanced learners in the USA. Due to their prominence, the author focused on images (e.g. pictures, drawings, tables) and accompanying texts (e.g. ‘culture boxes’ and ‘chapters’). In a quantitative analysis, he categorized images as ‘marked’, whereby ethnic minorities are presented as such, for example, by designations like ‘immigrant’, or ‘unmarked’, whereby ethnic minorities are not presented as such but are recognizable through physical appearances (Markus & Moya, 2010). Results showed that textbooks for beginners displayed diversity more frequently through pictures than text, but the opposite was true in textbooks for advanced learners. The ratio of marked vs. unmarked pictures increased with level of proficiency (78.6% marked pictures at the advanced level). Possibly, this trend reflects the conviction that students can only converse about diversity when they have gained sufficient knowledge and skill of the target language and culture. Along these lines, the current study also presents a quantitative analysis of images and accompanying texts, whereby we categorize using Ilett’s (2009) marked–unmarked distinction. In extension, we also analyse gender and, thus, count how many women and men of colour, and white women and men appear on (un)marked images in our sample.
Ilett (2009) also qualitatively analysed marked and unmarked images in his sample, which, according to him, are reflective of two tendencies: stressing difference vs. considering diversity as a given. Both these tendencies may have positive or negative effects. Stressing diversity, for example, may inform readers about diversity in Germany or celebrate differences, but, on the other hand, it may support an ‘us vs. them’ dichotomy instead of a complex understanding of identities. Considering diversity as a given may have the positive effect of framing diversity as neutral and natural. On the other hand, it may gloss over differences and support an assimilation ideology (see also Agirdag, 2020). To determine whether Ilett’s findings can be replicated in another context, we highlight marked and unmarked cases in our sample and discuss to what degree they reflect these tendencies.
Markus and Moya (2010) stress that signifiers of different ethnicities may arise not only via physical appearances but also in national affiliations. Yamada (2011) examined the occurrence of nations in 15 English language textbooks (1987–2002) used in Japanese secondary education. She counted how often and which countries were described in texts and also categorized ethnic groups based on physical appearance (e.g. skin and eye colour, hairdo, facial features) and cultural aspects (e.g. country of origin, language, religion) described in texts and images. She observed that Western countries were more frequently named than Asian countries (see also Lee, 2009; Joo et al., 2020), though more recent editions have featured a slight increase of Asian countries. The USA and Japan were the most frequently mentioned countries, but ethnic diversity within these countries was hardly represented nor discussed. She did find a few ambiguous representations of nationalities as well as some interactions (e.g. international marriages), and Africa tended to be treated as a country and not as continent. Along these lines, the present study also analyses countries as well as cities to examine whether the tendencies found by Yamada also appear in our sample.
2 Gender from left to right
A long tradition of research has examined gender representations in (language) textbooks (see Angouri & Baxter, 2021; Mills & Mustapha, 2015; Niehaus, 2018). An interesting textbook study that addressed gender inequality in German textbooks at more subtle, spatial levels was performed by Moser and Hannover (2014), who were inspired by spatial positioning theory (Suitner and Maass, 2016). Following Suitner and Maass (2016), in (patriarchal) cultures that read from left to right (e.g. such as German or Dutch cultures), the most important, or male, information is represented on the left side. In linguistics, this phenomenon is also known as ‘male firstness’ (Motschenbacher, 2013). Moser and Hannover examined this hypothesis in a corpus with 18 first language German and mathematics textbooks. Though they found no spatial arrangement biases in the pictures in their corpus, we aim to test whether they do occur in our sample. Moreover, we aim to extend the gender hypothesis to include ethnicity: White men appear more frequently on the left side in pictures than do women of colour.
3 Intersectionality: Ethnicity × gender
Though few studies have examined intersectionality in language textbooks, Alter et al. (2021) conducted a relevant study that examined several social categories separately. The authors studied three recent English language textbooks intended for use in German secondary education and analysed inequalities in terms of gender, sexuality, people with physical and mental variations on body norms, race and class. The authors found that men appear more frequently than women, notably in portraits of single individuals, famous persons or as homogenous male groups. They also found that white characters constitute the norm, and that PoC appear less frequently. In particular, they observed that women with headscarves or individuals with black skin tend to appear in relation to discourses about racism, social problems (e.g. poverty) or occupational orientation. Osler and Starkey (2000), in the context of French mixed-media language learning materials, also observed that ‘the presence of ethnic minorities is often associated with disadvantage, problems and tensions’ (Osler & Starkey, 2000, p. 208; see also Bori & Petanovic, 2017). The authors also observed that for young women of African parentage, issues of identity are presented as especially problematic. They argued that texts about these women thematize a site of conflict between elements within the minority community who wish to restrict or control women and the forces of modernity. We are alert to whether such tendencies can also be found in our data.
Treadwell (2014) examined ethnicity and gender in a qualitative study of five textbooks for reading education in Burmese primary schools. She argued the books contain gender representations of both older and younger women and men in diverse professional roles and activities, which contrasts with previous findings that women are underrepresented (Mills & Mustapha, 2015). However, she also observed that the large ethnic, linguistic and religious diversity in Burma is hardly present in her corpus. The Buddhist, Burmese-speaking majority is the norm, and minority religions (e.g. Christianity and Islam) and languages are not mentioned. The few references to ethnic minorities have a marginalizing effect, she argued, since these stereotypical images are invariably accompanied by a dominant national discourse, suggesting that national identity supersedes minority identity (for similar results in Taiwanese textbooks for learning English and Chinese, see Ho, 2021). As in the study by Treadwell, the present study focuses on both gender and ethnicity, and we use diversity in these areas in the Netherlands as reference frame, as demonstrated by demographic statistics (for a similar approach, see de Bruijn et al., 2020). Demographic statistics offer one perspective on ‘realities’ in the Netherlands, and one which might affect textbook content. We aim to give some insight into the degree to which these or other ‘realities’ are reflected in our Dutch textbook corpus.
Whereas the studies above have predominantly focused on images and texts in textbooks, fewer studies have analysed how women and men with diverse ethnic backgrounds speak in a subgenre particular to language textbooks: dialogues. It is important to study dialogues, since they constitute models for actual use in the classroom (Jones et al., 1997). Along the lines of how ethnicities are displayed, Markus and Moya (2010) identified language as a key identifier of ethnicity. In the context of textbooks for English, a language with more non-native than native speakers (Crystal, 2003), considerable debate has been conducted as to whether varieties other than just British and American English should be presented in learning materials (Jenkins, 2002). Kivistö (2005), however, noticed that these debates have had limited effects on English language textbooks. She found hardly any varieties in English textbooks used in Finland, which aligns with Hillard (2014), who studied English textbooks for global use. She found that the textbooks tend to overemphasize standard native speaker English accents, and her sample included few tracks containing non-native accents. Two of the books purposely focus on North American accents, and no regional dialects appeared. The present study analyses to what degree varieties of standard Netherlandic-Dutch are present in Dutch language textbooks. We also aim to analyse the dialogues presented in the textbooks, specifically regarding whether significant differences occur in the number of types and tokens that women and men of colour vs. white individuals utter, since previous research (e.g. Jones et al., 1997) has shown that this distribution may differ when it comes to gender. Koster and Litosseliti (2021), for example, found that women in chapters about ‘family’ utter more types and tokens than men in dialogues in recently published textbooks (2007–2018). From a gender diversity perspective, this absence of male voices must be viewed in a critical manner.
III Materials and methods
To address the question of how diverse representations of wo/men of colour and white wo/men are in Dutch language textbooks, we employed a mixture of content and linguistic analysis of textbook texts and images. In particular, we selected six of the currently most frequently used textbooks for Dutch foreign language education (Ingrid Degraeve, personal communication, September 12, 2019): Contact 1 (2018), Help! 1 (2019), Nederlands in gang (2018), Routledge intensive Dutch (2006), Taal vitaal nieuw (2019) and Vanzelfsprekend (2020). All textbooks were targeted at adult (educated) beginners, which constitute the largest group of Dutch foreign language learners. The first three textbooks were published in the Netherlands, whereas Routledge was published in the United Kingdom, Taal vitaal in Germany and Vanzelfsprekend in Belgium. Contact, Taal vitaal and Vanzelfsprekend were printed in colour, whereas the other textbooks appeared in black-and-white ink. A complementary study that interviewed some authors of these textbooks (Koster & Iding, 2023) suggests they were predominantly white and female, carried the Dutch nationality, and spoke Dutch as their mother tongue. In total, we examined 1,772 textbook pages. Unlike the other books, both Help! and Vanzelfsprekend are characterized by key figures that reappear in almost every chapter. In Help! the main character is Maria Borsato, and in Vanzelfsprekend, they are Paolo Sanseverino and his friends. Both are introduced as Italian exchange students studying in the Netherlands and Belgium, respectively.
We performed quantitative analyses of textbook contents, focusing on visuals and their accompanying texts (based on Ilett, 2009); our focus on visuals is justified by their saliency and omnipresence. Here, an ethnically and gender diverse research team with seven trained students and five principal investigators coded images. Each book was analysed by one principal investigator, who was assisted by one or two students. Ambivalent cases were discussed within these sub teams until they were solved in joint discussion (as in Park et al., 2021). Within a final meeting with all principal investigators and students, the most difficult cases were discussed once more until solved. We counted how many pictures of people could be found and analysed the following aspects of the represented people: skin colour, gender, clothing/activity and position. To compare human representation in the textbooks with demographic statistics, we calculated a baseline measure of gender and ethnic diversity in the Netherlands in the period 2006–2020 based on data by Centraal Bureau Statistiek (CBS). We replicated methodology by de Bruijn et al. (2020), calculating men and women within the population with ‘Western’ and ‘non-Western’ backgrounds, taking these as a proxy for white individuals and PoC. Due to limited space, we refer interested readers to a detailed description of the data and this procedure in the IRIS database. 2
Considering the multimodal nature of textbook representations, we also analysed the texts accompanying pictures. Of the images depicting people, 44.52% (N = 256) were accompanied by texts (e.g. titles, articles, vocabulary etc.). Based on information in these texts, we categorized images of PoC in the categories ‘marked’ and ‘unmarked’ according to Ilett (2009) (see Section II). We also examined further ethnic markers in the texts accompanying pictures, such as countries and cities. Second, we analysed all dialogues with audio files in the textbooks, since they constitute important models for classroom conversations (Jones et al., 1997). We examined those conversing based on the following aspects: names, gender, variety of Dutch spoken, and the uttered number of types (e.g. unique words) and tokens (e.g. all uttered words). We qualified dialogues as conversations between two or more individuals with at least two conversational turns.
IV Results
1 Individuals depicted in images
We hypothesized that white men would appear most often and women of colour least often in the textbook sample. In total, the corpus contained 575 images with people (interval: 16–206 images, average: 95 images per book), depicting 1,364 individuals (see Table 1). Table 2 features the frequencies of white wo/men and wo/men of colour for each textbook. Figure 1 shows the division for the complete sample, and the sample without Contact (PoC, N = 205, 84.71% of 242 PoC in the total sample) as this textbook seemed an outlier. We first performed a chi-squared test without Contact revealing significant differences, with χ2(3) = 79.16, p < .001. Comparison of observed values and the calculated population proportions (see Table 3) showed that white men, χ2 = 24.61, were overrepresented, whereas men of colour, χ2 = 26.35, and women of colour, χ2 = 27.68, were underrepresented. Second, we ran a chi-squared based on the sample with Contact. Again, we revealed significant differences between observed and expected values, with χ2(3) = 23.20, p < .001. However, white women, χ2 = 8.21, were now slightly underrepresented and women of colour, χ2 = 10.89, were overrepresented.
Distribution of pictures and dialogues across the sampled textbooks.
Gender, skin colour and markedness of depicted individuals in the individual textbooks.
Gender and skin colour of depicted individuals in the full sample and the sample without Contact.
Values for chi-square analyses with full sample (A) and the sample without Contact (B).
We also hypothesized that wo/men of colour appeared less often on the left side in group pictures with a mixed ethnic and gender profile (51 pictures; see Table 1) than white wo/men. Again, Contact seemed an outlier, since 37 (72.55%) of the pictures that displayed both wo/men of colour and white wo/men appeared in this textbook. In total, we identified 5 women of colour, 10 men of colour, 17 white women and 20 white men in left-hand positions on pictures. A chi-squared test, based on population proportions, showed that there was a significant difference between the observed and expected values for men of colour, χ2(3) = 11.04, p = 0.02. Ten men of colour appeared in left-hand positions, whereas, based on the odds ratio, 3.65 should have appeared according to the population statistics. The other frequencies were reflective of societal tendencies. In sum, when not considering Contact, our results support the hypothesis that white men appeared most often and women of colour least often. This does not reflect demographic ‘realities’ in the Netherlands. Contact, however, displayed significant ethnic and gender diversity, featuring slightly more women of colour than men of colour, but compared with demographic statistics, too few white women. On pictures with mixed ethnic and gender profiles, mostly stemming from Contact, we see men of colour in left-hand positions more often than predicted.
2 (Un)marked images
We hypothesized that PoC in the textbooks would be featured in a limited range of roles with low status (e.g. jobseeker, waiter, domestic worker). We also aimed to examine whether PoC were presented in a marked (e.g. by designations like ‘immigrant’) or unmarked manner (e.g. not described through designations like ‘immigrant’, but recognizable as ethnic minority through physical appearances). Table 4 shows the marked and unmarked pictures featuring PoC per textbook. Of these 93 pictures, we categorized 14 (15.05%) as marked, and 79 (84.95%) as unmarked, for example, a picture of female physician of colour in Contact (p. 125). This suggests that unmarked presentations of PoC are the default in Dutch textbooks. However, again, Contact seemed to drive this effect. If we removed the data from Contact from our dataset, the balance changed, since 9 (37.5%) from the 24 images could be categorized as marked, and 15 as unmarked (62.5%) (see Figure 2).
Markedness for depicted People of Colour (PoC).
Distribution of marked versus unmarked images of People of Colour (PoC) in the full sample and the sample without Contact.
Help! and Routledge both contained only one image with PoC, depicting a woman of colour, both marked. In Help!, a black girl is described as living in the underdeveloped ‘third world’ and in need of help. Routledge showed a young, Muslim woman, who is framed as member of an ethnic minority community in an accompanying textual discussion. In this discussion, five youngsters discuss the appropriateness of her headscarf (pp. 238–239). They defend different positions concerning ‘integration’ of ethnic minorities, varying from assimilationist perspectives, that require minorities to adapt to the ethnic majority in the Netherlands, to pluralistic views, that appreciate diversity of identities. On pages 239–241, the textbook authors build on this discourse through vocabulary lists, a fill in the gaps assignment, and a speaking and writing exercise. Taal vitaal featured seven marked images of PoC, and the accompanying text stressed non-Dutch nationalities and mentions that depicted PoC migrated from another country to the Netherlands. The three marked images in Contact also mentioned non-Dutch nationalities or migration histories. By analysing nouns and substantiated adjectives in the texts accompanying marked pictures, we identified the following functions ascribed to PoC: Aside from student, tourist and jobseeker, PoC were described as migrant (worker), asylum seeker, cook, colonized and recipient of development aid. In sum, we found that most PoC were depicted and described in both unmarked and marked manners, indeed, in a limited range of low-status functions, perhaps with the exception of tourist and student.
3 Countries, nationalities and cities
National affiliations can also be markers of ethnicity, and from a diversity perspective, we would expect to find a plethora of non-European and European countries and cities. 3 Our analysis showed that only 21 of 195 possible different countries were mentioned in texts accompanying pictures (e.g. through nouns such as Belgium and adjectives such as Dutch). The Netherlands was mentioned most frequently in all textbooks (43 of 143 instances in total, 30.07%; see Table 5). Belgium was mentioned less frequently (N = 14, 9.79%) and only in three of six textbooks. Suriname was mentioned twice in Taal vitaal, but other countries where Dutch is an official language (e.g. Curaçao, Aruba, Sint Maarten) were not described. Next to Suriname, South Africa and Indonesia were the only countries with a history of Dutch colonization appearing in the corpus (N = 1; N = 4 in two books). We found that a limited range of European countries (e.g. Germany, Italy, Spain; 112 instances, 78.32%, eight different countries) was more frequently mentioned than a cluster of non-European countries (e.g. Chile, Syria, Cuba; 27 instances, 26.68%, eight different countries). Focusing on cities (162 instances), we found that Dutch cities (14 different cities, mentioned 36 times, 22.22%) and Belgian cities (11 different cities, 90 instances, 55.56%) appeared most frequently in the corpus. Nine other European cities (24 instances, 14.81%) and 7 non-European cities (9 instances, 5.56%) were mentioned. The Belgian cities almost exclusively appeared in the Belgian textbook, Vanzelfsprekend. Both larger as well as smaller cities were mentioned (e.g. Amsterdam vs. Houten; Brussels vs. Oostende). All in all, from a global citizenship stance, the countries and cities in the textbooks can be characterized as quite Eurocentric and Low-Countries-centric, with some exceptions.
Count of (non-)European countries and cities cited in text accompanying the picture sample.
Notes. * indicates appearances in more than one textbook.
4 Dialogues
Finally, we hypothesized that little linguistic diversity in dialogues, a subgenre within language textbooks, would appear. We were also interested in examining whether female, male or (if present) gender-neutral characters contribute equally to dialogues, since Koster & Litosseliti (2021) documented imbalances. The corpus contains 253 dialogues appearing in both written and spoken form, ranging from 23 dialogues (Nederlands in gang, 2018) up to 94 (Vanzelfsprekend, 2020) (see Table 1). Based on textual references, we identified the dialogue speakers as follows: 344 women, 383 men (e.g. indicated by fe/male names or pronouns) and 25 instances of gender-neutral individuals; e.g. indicated by gender-neutral nouns, such as kind (‘child’); professor (‘professor’), klant (‘customer’), buur (‘neighbour’), or loketbediende (‘counter clerk’). For 21 of the 25 gender-neutral individuals, we subsequently inferred a gender using the pitch of the speakers in the audio files; this resulted in the total number of gendered dialogue participants as being 347 women and 401 men. We could not determine whether the dialogues included PoC or white individuals, since few dialogues were accompanied by pictures. We thus focused our type–token analysis on utterances of women and men. As shown in Table 6 and Figure 3, women and men spoke approximately as diversely, as defined by the types (unique words) uttered, and equally as much, as defined by the tokens (all words) uttered. In Vanzelfsprekend and Help!, men occupied slightly more verbal space, whereas the other textbooks were more balanced.
Type–token distribution of wo/men and gender-neutral characters speaking in (audio-recorded) dialogues per textbook.
Distributions of tokens in dialogues over women and men.
Further analysis of the audio files showed that standard Netherlandic-Dutch is overwhelmingly represented in the textbooks. One exception is the textbook published in Belgium, which has Belgian-Dutch as the default. Though we found one instance of Belgian-Dutch in the textbooks published outside of Belgium, we found no examples of varieties spoken in the Netherlands, Belgium, Surinam (e.g. Surinamese-Dutch), Aruba, Curaçao and Sint Maarten, and only one example of accented speech: Italian-Dutch. We noticed that a prominent function of the few speakers of varieties or with nationalities other than Dutch or Belgian seems to be learning phrases that indicate dependency. Such speakers were described in introductory texts or within dialogues, such as French and Americans in Vanzelfsprekend and an Austrian couple in Taal vitaal. These speakers uttered phrases like ‘Could you repeat this?’ (Help!, p. 223; our translation) or ‘I do not understand, what do you mean?’ (Nederlands in gang, p. 143; our translation). In sum, we found that women and men of mostly unidentifiable colour are equally represented in dialogues and produce equally as many utterances. We found virtually no language varieties or accented speech as the audio files feature speakers of either Netherlandic-Dutch or Belgian-Dutch.
V Discussion
This study aimed to contribute to textbook research from a cultural studies perspective (Risager, 2018), by examining textual and visual representations of wo/men of colour and white wo/men in six frequently used textbooks in an under researched context: Dutch language education. We now discuss findings in relation to existing literature (e.g. UNESCO, 2020, supported by Good et al., 2020) and consider their implications.
1 From underrepresentation to fundamental equality?
We asked how diverse representations of wo/men of colour and white wo/men in Dutch language textbooks are. Our frequency analyses of individuals in pictures showed a white, male bias, and an underrepresentation of PoC in comparison with demographic statics, except for in the textbook Contact. Our findings align with the white, male bias found by Alter et al. (2021) in English language textbooks in Germany. The underrepresentation of PoC also aligns with findings for Dutch children’s literature by de Bruijn et al. (2020), suggesting the replication of bias across educational genres. This seems an important topic for further study, including other digital, audiovisual learning materials (see Ott, 2022). Yet, in terms of ethnic and gender diversity, it was encouraging to see that the authors of Contact chose another direction, and displayed women and men of colour in numbers and left-hand positions (in contrast with findings by Moser & Hannover, 2014) that exceeded Dutch demographic statistics.
This opens up interesting debates about the degree to which textbooks should indeed orient towards demographic statistics within certain linguistic societies of interest, or whether more fundamental balanced representations of ethnic and gender diversity, as in Contact, should become the norm. We argue that the latter should not be excluded, as this aligns with the prescriptions of UNESCO (2020), and the first option seems to give rise to the following dilemmas. Indeed, authors could mirror the most recent demographic statistics of certain countries, but which ones is obviously the question, since, in our case, one could opt for the Netherlands, but one could just as well choose for Belgium, Surinam, Aruba, Curaçao and Sint Maarten or a mixture of these. In addition, one could debate whether demographics of the home country of the intended students also need to be taken into consideration. Opting for representing ‘future’ fundamentally equal worlds is therefore an interesting route to take and Contact sets an example of a textbook moving in this direction.
We also analysed how PoC were represented in our sample. In the sample without outlier Contact, our finding for markedness matched with Ilett’s (2009) findings for German language textbooks, in which 36.64% of marked images were found at the introductory level. Future studies for Dutch language textbooks should document whether this tendency also holds at intermediate level, or whether, as Ilett reports, the level of markedness increases in books at the intermediate level (e.g. 78.57%). As Ilett discusses, it could be beneficial to find both type of representations in a balanced manner, reflective of ‘stressing difference’ vs. ‘considering diversity as given’ ideologies, as both types of ideology may have positive or negative effects in terms of forming complex views on ethnicity, or views that ‘enhance intercultural dialogue, social inclusion and democracy’ (Council of Europe, 2020). Indeed, future research that documents effects of different types of representation on students is needed to shed light on this matter. The discussion exercise in Routledge intensive Dutch mentioned in Section IV.2, which includes several ideological perspectives on integration, may constitute concrete experimental material in such studies.
The limited range of functions ascribed to PoC in our sample aligns with findings for textbooks on TEFL (teaching English as a foreign language), where PoC tend to be linked to social problems (e.g. poverty) or occupational orientation (Alter et al., 2021). We did not find examples of PoC in discourses of racism, as the authors above did, which may be explained by the authors’ focus on a chapter that discussed apartheid in South Africa, and the fact that no comparable chapters were present in our sample. Comparable to French learning materials (Osler & Starkey, 2000), we did find a woman of colour at a site of conflict between customs within the minority community and forces of modernity, as shown in an exercise that encouraged discussion about her headscarf. Our finding that PoC appear in stereotypical, low-status roles with little agency also aligns with representations of ethnic minorities in materials for language education in a Burmese (Treadwell, 2014), Catalan (Bori & Petanovic, 2017) and Taiwanese (Ho, 2021) context, which stresses that this is an issue worldwide and justifies the intermingling by institutions such as UNESCO (2020). To prevent a situation in which PoC or ethnic minorities are perceived only as migrants, the principle of counter-examples (UNESCO, 2017) could be fruitful to reduce this bias; that is, for every waiter of colour, textbooks should also show an engineer of colour, etc. In addition, more elaborate stories about PoC or ethnic minorities stressing their complex, multidimensional identities are needed.
2 Eurocentrism and the doctrine of standard language
We also aimed to examine whether textbooks in our sample enhance ‘intercultural dialogue’ by presenting non-European perspectives, and ‘linguistic diversity’ (Council of Europe, 2020), in terms of variation on standard Netherlandic-Dutch in dialogues. We therefore documented ethnic diversity in terms of the countries and cities described in the texts that accompanied pictures of humans. This distribution could be characterized as Eurocentric and Netherlands-centric, with some exceptions. This result may be unsurprising, given that the textbooks were published in the Netherlands, in Belgium, or for learners in the Anglophone world and predominantly designed by (white, female) authors who have Dutch nationality and speak Dutch as mother tongue (see Koster & Iding, 2023). Our findings are in line with Yamada (2011), who found Eurocentric tendencies and American dominance in English language textbooks. Though in our corpus Africa was not treated as a country, as Yamada (2011) found, it is noticeable that individual countries or cities in Africa hardly appeared. This also holds for the continents off America and Asia. In terms of representing diversity and enhancing ‘intercultural dialogue’, the Dutch language textbooks in our sample should thus pay attention to a larger part of the world, including contexts outside the Netherlands and Belgium where Dutch is spoken. The fact that countries formerly colonized by the Dutch or the Belgian were hardly described seems to be an interesting issue to thematize in future research from one of the other research angles that Risager (2018) presents: a postcolonial studies analysis of language textbooks. This topic has already been researched for TEFL textbooks, which do feature historical information about (post)colonial relations, though from a narrow, British point of view (Risager, 2018). Future studies could examine whether this upholds for language textbooks of other imperial countries, and, perhaps even more interesting, in language textbooks for languages spoken in formerly colonized countries.
Finally, we examined to what degree ‘linguistic diversity’ was featured in our textbook sample. We found that women and men of mostly unidentifiable colour are approximately equally represented in dialogues and give equally as many (and as diverse) utterances. This contrasts with Koster and Litosseliti (2021) who found a gender bias in favour of women in language textbook chapters about family. The focus of Koster and Litosseliti (2021) on the family context may explain why, in the large ranges of dialogues about different themes, we did not find significant differences. In addition, new research with language textbook authors (Koster & Iding, 2023) shows that authors consciously choose to balance female and male voices to reduce potential confusion for language learners. Despite this general tendency of equality, it remains recommendable to watch out for inequality with themes where gender plays an important role (e.g. family, profession etc.). We found virtually no language varieties or accented speech, as the audio files accompanying the dialogues feature speakers that speak standard Netherlandic-Dutch, with only two (female) exceptions. This lack of variety is in line with research into English textbooks (Hillard, 2014; Kivistö, 2005). Future studies should examine whether this pattern is found in educational materials for learning other languages as well, or whether textbooks with linguistic diversity exist, and what experiences teachers and students have or desire with such content (for perceptions of linguistic diversity by teachers of Dutch, see De Wilde, 2019). In addition, we noted that ‘other’ characters are framed as dependent through their utterances (e.g. ‘Could you repeat this?’). To our knowledge, this finding has not yet been thematized extensively in the literature, but it matches with earlier findings showing that ethnic minorities in French and Catalan textbooks are represented in dependent, low-status and working-class roles (Bori & Petanovic, 2017; Osler & Starkey, 2000). Next to more obvious forms of exclusion, this, and other fine-grained forms of exclusion, should be thematized in specific diversity guidelines for educational materials such as the one by UNESCO (2017), to avoid repetition of biased representations and to work towards more diverse representations of humans in language textbooks.
3 Follow-up studies and implications
The present study was conducted in response to calls for textbook research on intersectionality, building on existing theoretical and educational debates. Future studies could further examine multimodal, micro-level foci (see Sunderland, 2015) not included in this study, such as the (biased) distribution of characters of colour vs. white individuals on group images and portraits (and accompanying texts) or image placement following certain themes or topics (see Park et al., 2021). There is also scope for non-binary investigations of gender representations in textbooks. Examining intersections with other categories, such as class and age are also needed, since previous research suggests that middle-class, young or middle-aged characters are overrepresented in English language textbooks (Gray & Block, 2014).
Further, there is significant scope for research in this area involving different (e.g. non-Indo-European; on Arabic foreign language textbooks in Iran, see Baghdadi & Rezaei, 2015) languages to document to what degree and in what forms bias appears in textbooks worldwide, what progress is made, and how different language communities can learn from each other when it comes to reducing bias. Educational media that complement textbooks in class should also be examined in order to determine to what degree they replicate or counter bias in textbooks. In particular, given the ever-increasing digitalization of education, as accelerated by the Covid-19 pandemic, online audio-visual materials should be examined (Christophe et al., 2018; Ott, 2022). Indeed, more research is also needed to determine what verbal strategies teachers have to reduce bias in given educational materials, and what role teacher training can play (see Koster & Litosseliti, 2021).
There is also scope for studies that aim to involve publishers and textbook authors (Bläsi, 2018; Koster & Iding, 2023; Otto, 2018), and to give insight into what keeps them from adhering to diversity guidelines such as those by UNESCO (2017), so that research findings can be transferred into more inclusive and diverse textbooks. Indeed, complementing research from other theoretical perspectives, such as textbook studies from a postcolonial point of view (Ho, 2021; Kubota, 2020; Risager, 2021) is needed. And finally, more studies documenting effects of biased versus equal content on students in foreign language classes (as in Good et al., 2010, 2020) on achievement or related concepts are necessary to solidify the justification of UNESCO’s (2020) and the Council of Europe’s (2020) recommendations, and to work towards more diverse representations of humans in language textbooks.
Footnotes
Acknowledgements
We give thanks to participating student-assistants, and thank Gerdi Quist, Gunther de Vogelaer and two anonymous reviewers for their insightful comments.
Declaration of conflicting interests
Christine Sas is both author of the present article as well as author of the examined textbook Routledge intensive Dutch (2006). We included this material since it is one of the most frequently used textbooks for Dutch as a foreign language (DFL) education, and one of few textbooks explicitly directed at English-speaking DFL learners. Since another research team member was responsible for analysing Routledge intensive Dutch, the integrity of this research is not challenged.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Dutch Taalunie, project number 453100.