Supporting health literacy among older adults in basic education: Educators’ perspectives on the use of the KANSAS search engine

Abstract

Objective:

Providing basic health education to older adults with limited reading skills is increasingly important due to demographic change. Health literacy enables older adults to navigate healthcare more successfully and make informed decisions, but educators in adult basic education often struggle to find materials that are both relevant and suitable. This study explored two key aspects of this challenge. First, it investigated how adult educators from Germany perceived the relevance of health literacy for adults aged 65 years and above. Second, it examined the effectiveness of a language-adaptive search engine, KANSAS, in identifying linguistically appropriate health-related texts for this target group.

Design:

Mixed-methods design, combining a survey and an experimental comparison of perceptions and search results of two search engines.

Setting:

Data were collected via an online platform used by adult educators working in basic education contexts.

Method:

A total of N = 58 adult educators participated in a between-subjects study in October 2022, comparing the KANSAS search engine with a similar looking search engine without linguistic functionalities. Participants were asked to evaluate retrieved health-related reading texts regarding their linguistic suitability. Texts were analysed for readability, word and syllable count, and targeted grammatical features.

Results:

Many educators considered health literacy for older adults important in their professional context, although some reported limited experience working with members of this learner group. Texts retrieved via KANSAS showed significantly better readability and lower linguistic complexity. Although the frequency of targeted grammatical constructions did not differ significantly, participants rated KANSAS as more suitable for educators in identifying relevant linguistic features.

Conclusion:

KANSAS can support literacy educators by facilitating the selection of appropriate health texts. Combined with growing educator awareness of language-adaptive reading material, such tools may help integrate health literacy more effectively into various adult education contexts.

Keywords

Digital tools health literacy literacy older adult learners search engine

Introduction

Health literacy is essential for overall health and well-being. It involves the ability to access, understand and use health information to make informed health-related decisions (Institute of Medicine, 2004). While there exist numerous definitions of health literacy, from an educational perspective, it can be generally understood as a set of skills, such as reading and understanding health materials, communicating with healthcare providers, navigating healthcare systems and critically assessing health information to provide a basis for informed decision making (Nutbeam, 2008; Sørensen et al., 2012). Low health literacy has been linked to poorer health outcomes, including higher hospitalisation rates, increased costs and lower adherence to medication (Berkman et al., 2011).

Maintaining health and independence is a key concern for older adults. As they age, people typically attach more importance to their health and well-being (Kuhlmey et al., 2008). Despite this, older adults, especially those with a migrant background (Quenzel and Schaeffer, 2016), tend to have a lower health literacy than younger people (Sørensen et al., 2015). While health literacy is an important issue in all countries, this study focuses on the situation in Germany, with special attention paid to the perspectives of educators working in the context of adult basic education, and the potential of an educator-supporting digital tool to improve the quality and suitability of reading materials for older people, for health literacy purposes.

Health literacy, low literacy skills and older adults

Reading and understanding health information are basic components of health literacy (Sørensen et al., 2012) and are often referred to as functional health literacy (e.g. Nutbeam and Lloyd, 2021). Individuals with low literacy skills are nearly twice as likely to face challenges in comprehending and applying fundamental health information compared to the average population (Schaeffer et al., 2021). In Germany, many adults exhibit severe literacy deficiencies, as evidenced by representative studies (Grotlüschen et al., 2020; Rammstedt et al., 2024). While these studies have often focused on the working-age population, the proportion of poor readers is even higher among elderly people (Friebe et al., 2014; see also Schröter and Bar-Kochva, 2019, for an overview of the [low] reading competencies of different age groups in Germany). Furthermore, they tend to use simpler sentence structures and show a reduced ability to understand complex sentences (Cheung and Kemper, 1992; Kynette and Kemper, 1986; Light, 1993). As underlying causes of this, scholars have both discussed biological ageing leading to a reduced working memory (Naveh-Benjamin and Cowan, 2023) and cohort effects (Roman, 2004), with stronger support for environmental and social influences compared to general age-related cognitive decline (Kulmus et al., 2025). Importantly, health literacy challenges in older adulthood arise not only from limited reading skills but also from the interaction of age-related cognitive changes, numeracy demands and the increasing complexity of healthcare systems, which can overwhelm even otherwise experienced individuals (Kopera-Frye, 2016).

Health literacy education for older adults in the context of basic education

Older adults’ health literacy is important in contexts such as healthcare, community living, public health and digital services. While there have been several state-funded initiatives relating to health literacy in Germany (Jordan, 2023), there has been little systematic recording of the activities in which health literacy in adult basic education takes place. A key context for promoting health literacy can be found in adult basic education courses, where adults over the age of 65 make up one-fifth of participants in Germany (Christ et al., 2019). These courses offer accessible learning opportunities to satisfy older learners’ needs. General literacy courses are significantly more common than dedicated focused health literacy education programmes. Moreover, Integration and German as a second language courses¹ for migrants, with a focus on literacy, represent the most frequently offered format in adult basic education (Ortmanns et al., 2024). Integrating health topics into literacy, language or digital skills instruction may help build foundational competencies and practical health knowledge for older adults in a supportive environment. This can support learners not only to manage their health but also to strengthen social participation, confidence and independence – skills linked to more advanced forms of health literacy, such as interactive (active engagement in health communication) and critical (critical analysis and use of information for greater control over life) health literacy (Nutbeam, 2008). Educators have a key role to play in helping older adults to reach these goals through adult basic education courses (Comings et al., 1999).

High-quality adult basic education requires an individualised approach that addresses the needs, backgrounds and learning preferences of adult learners basic education (Peuker et al., 2022) in order to foster psychosocial stability and good learning outcomes. However, it is yet unclear whether adult basic education educators see health literacy education for older adults as a relevant topic to engage with as part of their educational approach. Therefore, the first research objective in this study was to investigate perceptions of the importance of health literacy among adult basic educators, including in integration courses.

RQ1. How do educators working in adult basic education perceive the importance of health literacy for older adults in their professional practice?

The importance of high-quality educational materials

Suitable educational material is a core component of the individualised, learner-centred approach in adult basic education. Educators strive to find appropriate reading materials and adapt them to the specific needs, interests and language proficiency levels of particular target groups of learners (Weiss et al., 2018). Suitably selected reading material is central to fostering the development of functional health literacy and laying the foundations on which educators can further support the development of interactive and critical health literacy learning skills. The literature highlights several issues that need to be taken into account when selecting and adapting educational materials for adult basic education learners.

Content

Learning texts designed for adults with low literacy skills should be relevant to their immediate living environment (Genuneit, 2016). Particular health topics are of special importance to older adults (Kuhlmey et al., 2008). They include the prevention of age-specific diseases, knowledge of relevant physical and cognitive abilities, and information about the structure and how to access healthcare systems (Connell, 1999; Speros, 2009). Thus, incorporating relevant health-related learning content into adult basic education courses should foster older learners’ health literacy while simultaneously catering to their heightened interest in the subject matter.

Language proficiency

In order to be effective, learning materials should provide learners with developmentally appropriate input, neither overwhelming nor underwhelming them (Gass and Mackey, 2012; Krashen, 1977). To improve the understandability of health-related information for older adults in adult basic education, two complementary strategies can be applied. The first one focuses on simplifying texts to match low literacy skills. This involves reducing linguistic complexity by using fewer words, writing in short and concise sentences, and avoiding subordinate clauses, as generally recommended in both adult basic education (Weiss et al., 2018) and health education (Horner et al., 2000). Easy Read² is one way of reducing linguistic complexity, but there are various other approaches, and we do not promote the use of a specific method, since effective simplification depends on individual needs. Generally, adjustments to reduce linguistic complexity seek to lower the reader’s cognitive load and make information easier to understand (Hartley, 2004).

The second strategy seeks to complement the first by enriching texts in ways that stimulate and preserve language proficiency, which is important for communication, social interaction and cognitive function (Genuneit and Genuneit, 2011; Petersson et al., 2009). While individual literacy challenges may vary, in an investigation of healthy native English speakers aged between 50 and 90 years, Kynette and Kemper (1986) found that members of older age groups faced difficulties with particular grammatical elements, such as articles, possessive pronouns and verb tenses. Including such elements in reading materials and practice exercises can help reinforce and maintain essential language skills. Both strategies combined together can ensure that health information is accessible and cognitively engaging. While the first strategy provides a foundation by enabling basic comprehension, the second one promotes reading proficiency and intellectual stimulation, supporting both understanding and active participation in daily life (Manly et al., 2005; Petersson et al., 2009).

KANSAS – a linguistically informed search engine

While suitable health-related educational material is key to addressing the specific needs of older adults in adult basic education, finding appropriate reading texts remains a major challenge. Most available materials, including online health information, are of at least medium complexity (Daraz et al., 2018; Dittrich et al., 2019; McInnes and Haglund, 2011). Furthermore, educators must assess the suitability of the content and linguistic features simultaneously, making material research time-consuming and demanding. Thus, there is a need for innovative digital tools that help educators identify suitable learning material for older adults. This paper introduces KANSAS, a German search engine designed to support educators in finding appropriate and thematically diverse reading materials for use with low-literacy adults. KANSAS is an acronym and stands for KompetenzAdaptive, Nutzerorientierte Suchmaschine für Authentische Sprachlerntexte (competence-adaptive, user-oriented search engine for authentic language learning texts).

The KANSAS search engine

KANSAS is a free and open-access search engine.³ While it was developed to assist teachers of adult basic literacy education using German language texts (Weiss et al., 2018), it is available for use by anyone who wants to find readable material on any topic. As part of its development, the perspectives of practising literacy educators were integrated through iterative qualitative studies (Kholin et al., 2020).

Search modes

KANSAS offers three search modes: an unrestricted web search, a filtered web search limited to selected adult basic education-relevant websites and a corpus search. The corpus, known as the alpha corpus, is a collection of over 11,000 pre-analysed, open-access texts designed specifically for use in basic language education contexts. In earlier work, Dittrich et al. (2019) compared these modes in terms of content coverage, readability and learning text suitability. The unrestricted search offers the broadest coverage, supporting diverse and up-to-date content needs. The corpus search identified resources with the highest levels of readability and educational suitability. The filtered web search balances both aspects. Thus, each mode serves distinct educational purposes.

Linguistic complexity in KANSAS

KANSAS assesses linguistic complexity using alpha levels⁴ (Heinemann, 2011). They are based on specific text features such as text length, sentence length and structure, temporal structures and word length. These levels range from alpha level 1 to 6. Alpha levels 1 and 2 refer to basic skills such as letter and word recognition and are not directly related to understanding connected text. Alpha level 3 marks the entry point into text comprehension and roughly corresponds to the Common European Framework of Reference for Languages (CEFR) reading level B1. From levels 3 to 6, complexity increases progressively.

Filter functionality in KANSAS

KANSAS allows users to sort texts by specific constructions and preview them with highlighted grammatical elements. This feature builds on the signalling effect, which aims to enhance learning by drawing attention to the most relevant information (Van Gog, 2014). Signalling has been shown to reduce complexity, improve motivation, lower cognitive load and shorten learning time (Schneider et al., 2018).

The value of KANSAS in finding health literacy information for older learners

Prior research using KANSAS has demonstrated that adult basic educators perceive the search engine as a valuable tool for finding language learning materials and lesson preparation (Kholin et al., 2020; Mayer et al., 2023). In this study, we assessed how well KANSAS could identify health literacy materials for older adults with reading skills at alpha level 4 or below, meaning they can at best read short texts but still face difficulties (Grotlüschen et al., 2020). The specific research question we sought to explore was as follows:

RQ2. How effective is the KANSAS search engine in identifying health-related reading materials for low-literacy older adults, regarding readability and suitability ratings, compared to a conventional search engine?

It was our belief that the signalling features of KANSAS makes it easier to identify suitable learning materials than conventional search engines, enabling educators to make more informed decisions about their selection. Since KANSAS highlights specific grammatical constructions, educators can select texts with the desired linguistic properties or adapt them, where licensing permits. We were therefore interested in whether KANSAS can assist educators in identifying health-related reading texts to support both strategies (text simplification and text enrichment) to enhance older adults’ literacy development. Three specific hypotheses were developed to explore these issues.

Hypothesis 1. Regarding text simplification, educators using KANSAS are likely to select texts with higher readability compared to those obtained via conventional search engines.

Hypothesis 2. Regarding text enrichment, educators using KANSAS are likely to select texts with more specifically required grammatical constructions in contrast to the use of conventional search engines.

Hypothesis 3. Educators report higher satisfaction with KANSAS’s ability to find texts with higher readability and specifically required grammatical constructions, compared to conventional search methods.

Methods

To test the hypotheses concerning text readability, grammatical enrichment and educator satisfaction, we conducted a mixed-methods online study.

Participants and procedure

Participants

Participants were 58 adult basic educators from across Germany (n = 40 [69%] female, n = 18 [31%] male), recruited via newsletters and announcements on relevant professional websites. They completed the assessment online and received a €25 voucher in compensation. Regarding their qualifications, 13.8% (n = 8) of the participants held a bachelor’s degree, 43.1% (n = 25) had a master’s degree, 3.8% (n = 2) had completed a second state examination⁵ and 6.9% (n = 4) held a doctoral degree.

Most participants had formal qualifications in the teaching of literacy, German, or German as a second language. On average, participants had 7.54 years of professional experience (SD = 6.11), with 69% (n = 40) working as freelancers. They primarily taught Integration courses, that is, government-funded programmes that help immigrants learn German and understand the country’s culture, laws and everyday life (51.7%, n = 30), or literacy courses (48.3%, n = 28), averaging 19.10 teaching hours per week (SD = 12.22).

Experimental conditions

We conducted a between-subjects online experiment with two conditions, using two different search engines. In one condition, participants used KANSAS with full linguistic features and all three search modes. In the other condition, participants used a control search engine with a user interface that was designed to be similar to KANSAS but which lacked the advanced function of highlighting selected grammatical constructions, only providing text previews without these. The interface of the control search engine also contained a short info-text about alpha levels, without the option to adapt the search to ensure that all participants had the same knowledge basis on the readability criteria used in the study. Participants were randomly assigned to each of the two conditions.

Procedure

Participants received European data protection information and completed a questionnaire on demographics and vocational background. Next, they rated the personal relevance of the topic ‘Health literacy of people aged 65+’ before being randomly assigned to an experimental condition. They then watched a screencast introducing their assigned search engine, practised using it while undertaking a trial search and were then instructed to search for a reading text to support a fictional lesson aimed at improving health literacy in older adults.

They were invited to use a self-chosen keyword related to one of four predefined health topics (prevention, age-specific diseases, physical and cognitive abilities, healthcare system). To standardise the search process, participants were instructed not to change their keyword and to select a text corresponding to alpha-level 4. Although alpha-level 3 is less complex, it limits texts to five sentences, making it impractical for web searches (Vajjala and Meurers, 2013). Alpha-level 4 better allows for the selection of longer, more natural texts.

In both conditions, participants were asked to ensure the selected text included possessive pronouns, to represent the enrichment strategy. After exporting and saving the text, they rated the suitability of the search engine.

Measures and data analysis

Educator-rated importance of health literacy for older adults

This was assessed with the question ‘How relevant do you consider the topic of “health literacy among people aged 65+” within the context of your professional work?’, rated on a scale from 1 (not relevant) to 4 (very relevant). Participants were then asked to explain their rating using an open response format.

Calculation of linguistic properties

To evaluate the overall linguistic complexity of the retrieved texts, we selected measures independent of the alpha-levels metric to avoid merely assessing KANSAS’s direct functionalities and applied two established readability indices instead. First, we used the German language adaption of the Flesch Reading Ease Index (FRE; Amstad, 1978; Flesch, 1948). It calculates readability by considering average sentence length and average syllables per word, producing a score ranging from 0 to 100, with higher scores indicating easier readability. In addition, we applied the LIX readability index (Anderson, 1981, 1983; Björnsson, 1968), which evaluates global complexity based on sentence length and word complexity. Unlike the FRE score, LIX focuses on the proportion of long words and average sentence length to determine text difficulty. LIX scores under 25 indicate very easy-to-read texts, while scores above 50 are classified as very difficult. We also analysed the number of sentences, subordinate clauses, words and syllables per text. To examine whether participants included possessive pronouns as instructed, we counted their occurrence in each text relative to the overall word count. While both readability indices show perfect reliability due to their automated calculation, their validity varies across contexts. In general, they correlate with reading comprehension and grade levels (e.g. Anderson, 1983; Cunningham et al., 2018; Peterson, 1956), but the validity of the thresholds provided by the indices has been critically questioned in the health context (Carter et al., 2024). It should be noted that they capture text features rather than underlying reading skills.

Suitability of search engines

Participants rated the suitability of the search engine for identifying language-adaptive reading texts for older adults on a scale from 1 (not at all suitable) to 5 (very suitable). To cover linguistic features relevant to the simplification strategy, we chose low global complexity, characterised by few subordinate clauses, short sentences and few words. To represent the enriching strategy, we chose the presence of articles and possessive pronouns. Importantly, we included criteria that did not perfectly align with the functionality menu of KANSAS to avoid merely evaluating its built-in capabilities. To ensure the expert validity⁶ of these criteria, we asked participants to rate the importance of these linguistic features for health literacy reading materials for older adults on a scale from 1 (not relevant) to 4 (very relevant).

Data analysis

To answer RQ1, we descriptively analysed the results of the ratings of subjective relevance of health literacy in their professional practice and summarised the open-question answers. With respect to RQ2, we compared the readability (H1) and linguistic features (H2) of the texts retrieved, and educators’ ratings of the search engines’ suitability (H3) of the two groups, using t-tests, or alternatively, Welch tests, in the case of heterogeneous variances for independent samples. Normality of the tested data was confirmed graphically by Q-Q plots (see Ghasemi and Zahediasl, 2012).

Results

Subjective relevance of health literacy education for older adults

Among participants, 15.5% (n = 9) considered health literacy for older adults very important to their occupation, and a further 39.7% (n = 23) viewed it as fairly important. Conversely, 39.7% (n = 23) considered it fairly unimportant, and 5.2% (n = 3) stated it was not important at all. The subsequent open question revealed multiple reasons behind these evaluations.

For those who deemed the topic fairly or very important, health literacy for older adults played a significant role in their professional lives. Educators emphasised the need to teach language skills in the context of health, enabling older adults to understand healthcare information and make informed decisions. Challenges encountered included age-related health issues and difficulties finding healthcare providers who spoke the participants’ first languages. Participants believed that health literacy would become increasingly relevant due to demographic changes. They stressed the specific needs of participants attending literacy courses, many of whom had physical, mental or psychological health issues and some of whom had experienced trauma. Overall, these perspectives underscore the crucial role of health education for older adult learners in adult basic education to improve language skills, enhance health literacy and manage health-related challenges.

In contrast, participants who assessed the topic as (fairly) unimportant said they had limited experience working with older individuals in their classes. Yet, they too recognised its potential relevance due to the influence older adults have on their grandchildren or on children who might join the courses. Despite not having personal experience of the topic, some acknowledged the intergenerational nature of health-related issues. A few saw possibilities of offering courses specifically for seniors, recognising the age diversity in their classes and the need to address health topics differently depending on the age group. While the current target group of these educators was mainly younger learners, some acknowledged the potential future relevance of health-related adult basic education as participants age or when older individuals join their classes.

Efficacy of KANSAS

Readability and linguistic properties of the retrieved texts

Table 1 shows the results of t-tests/Welch tests regarding the analysis of linguistic characteristics in the texts. Regarding overall complexity, the readability indices differed significantly between the two groups. Texts retrieved from KANSAS demonstrated higher readability, as indicated by FRE, t(56) = 2.14, p < .05, and LIX, t(51.77) = −2.30, p < .05. Consequently, our data provide support for H1. In terms of complexity components, no significant difference was found in the number of sentences, t(55.69) = −1.23, p = .219, and subordinate sentence, t(56) = −2.21, p = .09. However, texts retrieved by KANSAS contained fewer words, t(51.82) = −2.24, p < .05, and syllables, t(53.06) = −2.36, p < .05. No significant difference was found in the relative number of possessive pronouns, t(54.09) = −0.61, p = .271. Thus, H2 was not confirmed.

Table 1.

Comparison of the search engines for linguistic properties.

Complexity	KANSAS		Control search engine		t-tests/Welch tests for linguistic properties
Complexity	M	SD	M	SD	t	df	p
FRE/Amstad scores	56.48	21.05	46.79	12.10	2.14	56.00	.018*
LIX scores	43.85	9.99	49.19	7.45	−2.30	51.77^a	.025*
Number of sentences	9.69	6.91	12.03	7.44	−1.23	55.69^a	.219
Number of subordinate sentences	1.31	1.22	2.00	1.79	−2.21	56.00	.091⁺
Number of words	88.38	58.25	129.03	77.99	−2.24	51.82^a	.029*
Number of syllables	175.17	125.89	264.59	159.96	−2.36	53.06^a	.014*
Specific grammatical constructions
Percentage of possessive pronouns	0.67	0.91	0.83	1.11	−0.61	54.09	.271

N = 58.

Welch test due to heterogeneous variances.

p < .10; *p < .05.

In post hoc analyses, we analysed the readability of texts which were retrieved from the alpha corpus in the KANSAS condition. As the content of the alpha corpus is static, the corresponding texts were easily identifiable. Overall, three texts from the KANSAS condition were retrieved from the alpha corpus, with an average FRE of M = 79.33 (SD = 1.15) and a LIX of M = 29.09 (SD = 7.47), which can be interpreted as (fairly) easy.

Subjective ratings of suitability

Table 2 shows descriptive statistics of the relevance and suitability rating scores and the results of t-tests for each criterion. Educators’ ratings indicated moderate to high relevance, M = 2.33 to M = 3.34, of the selected linguistic criteria for suitability of texts for older adults. Educators considered low global complexity, few subordinate clauses and short sentences to be the most relevant criteria. KANSAS scored higher for three out of five criteria, including text containing possessive pronouns and articles, t(56) = 3.08, p = .002, few subordinate clauses, t(55.59) = 2.02, p = .024, and low global complexity, t(54.49) = 1.91, p = .031. Thus, H3 was only partially supported.

Table 2.

Comparison of the search engines for suitability ratings.

Educator-rated linguistic criteria	Relevance ratings^a		Suitability KANSAS^b		Suitability control search engine^b		t-tests/Welch tests for suitability ratings
Educator-rated linguistic criteria	M	SD	M	SD	M	SD	t	df	p
Low global complexity	3.34	0.63	3.69	1.00	3.14	1.18	1.91	54.49^c	.031*
Few subordinate clauses	3.14	0.66	3.55	1.05	2.97	1.14	2.02	55.98^c	.024*
Short sentences	3.10	0.64	3.59	1.24	3.21	1.11	1.22	55.37^c	.113
Few words	2.67	0.65	3.52	1.21	3.14	1.12	1.23	55.68^c	.111
Articles and possessive pronouns	2.90	0.69	3.72	1.06	2.93	0.88	3.08	56.00	.002**

N = 58.

Rating on a scale from 1 (not relevant) to 4 (very relevant).

Rating on a scale from 1 (not suitable) to 5 (very suitable).

Welch test due to heterogeneous variances.

p < .05; **p < .01.

Discussion

This study investigated how educators perceived the relevance of health literacy for older adults in adult basic education and assessed the effectiveness of the KANSAS search engine in helping them to find suitable health-related texts for low-literate older learners. Using a mixed-methods design with open-ended questions and a randomised online experiment, the study explored educators’ views and tested the tool’s suitability in a standardised way.

Summary and critical interpretation of the results

With respect to RQ1, the results of this study showed that while not all of the participants currently taught older adults, over half viewed health literacy as (fairly) important in their professional context. Many anticipated its growing relevance due to demographic change. Even among those who did not yet consider health literacy as important, its potential future significance was acknowledged, reflecting Germany’s demographic trend towards an increasingly ageing population with the associated health implications (Federal Statistical Office, 2023). Some educators noted that health literacy was also relevant for family members in caregiving roles, suggesting caregiver health literacy might become a relevant subdomain within adult basic education. However, the overall question asked was a broad one, and the study would have benefitted from a more in-depth inquiry. Educators’ prior experiences of health literacy could have been more systematically explored, for example, by directly asking whether they had encountered the topic in practice and in what form.

Regarding RQ2, KANSAS outperformed a conventional search engine. Participants retrieved texts with higher readability, mainly due to lower word and syllable counts. Nevertheless, KANSAS texts still fell into the category of ‘fairly difficult’, while those from the control search engine were ‘very difficult’ (Amstad, 1978; Anderson, 1981; Björnsson, 1968). This highlights the key challenge of high online text complexity, even with a tool designed to enhance web searching. Results also indicated that using the alpha corpus led to the identification of texts with higher readability. While too few participants used it to allow statistical testing, our findings align with earlier research showing the alpha corpus’s potential to retrieve texts with lower linguistic complexity (Dittrich et al., 2019).

KANSAS was also rated more suitable for finding texts with several linguistic criteria: low global complexity, few subordinate clauses, and the identification of articles and possessive pronouns. Although texts retrieved via KANSAS did not contain more possessive pronouns, participants perceived the search engine to be better at identifying them – likely due to its highlighting feature. Regarding subordinate clauses, the results show that although frequency did not differ, KANSAS received higher suitability ratings. These results may be due to the visible global complexity evaluation (alpha-level) in the KANSAS condition: Upon seeing the complexity evaluation, which is salient in the KANSAS condition, participants may have assumed that finding low complexity texts might be easy. In contrast, short sentences and low word counts did not receive significantly different suitability ratings across the two conditions, although actual occurrences differed significantly. These features may be less visually salient in KANSAS. Overall, when comparing the two conditions, participants’ subjective ratings of suitability did not always align with the actual occurrence of the text features as instructed. A within-subjects design might have improved interpretability by allowing direct comparison between both tools. Nonetheless, KANSAS was generally perceived as better at identifying relevant linguistic features, supporting simplification and enrichment strategies.

A key limitation to the study was the lack of a direct educator rating of KANSAS’s overall suitability for health literacy education, which remains important for future research. In addition, the online format of the assessment prevented observation of how participants used the tool, as some may have skipped the intro video or ignored alpha-level filters, potentially affecting the results.

Implications for future research and practice

In this study, adult basic education teachers appeared to recognise the relevance of health literacy for older adults, but this awareness had not yet translated into systematic implementation of specific educational practices. Building professional capacity could be helpful, for example through professional development and training on the topics of greatest relevance to older learners and health/basic literacy, as well as specific didactic methods, to ensure that educators are equipped to respond to the evolving needs of this growing learner group.

Enhancing KANSAS with didactic guidance for specific learner groups would be a practical next step. As KANSAS is currently designed for general literacy and not explicitly for health literacy education, more in-depth qualitative research is needed into educators’ prior experiences, instructional strategies and support needs when teaching health literacy to older adults. Moreover, since KANSAS was specifically designed to identify German language reading materials and utilises Germany-specific readability measures (i.e. alpha levels; Heinemann, 2011), the findings from this study cannot be generalised to other European or international contexts. However, the underlying principles (e.g. highlighting complexity and grammatical constructions) have been investigated in other language contexts as well (e.g. Chinkina and Meurers, 2016) and can be applied to any language.

Ultimately, our study demonstrates that tools such as KANSAS can help in the selection of linguistically better suited educational materials. However, the core issue of there often being too high a level of text complexity in the available materials remains. While some forms of generative AI (GenAI) have the potential to produce simplified texts, they are not yet well suited for the targeted retrieval of health-related materials for this age group. GenAI also lacks source transparency and is more error-prone than traditional search engines (Memon and West, 2024). It may also reproduce harmful stereotypes in health information (Van Kolfschooten and Pilottin, 2024). Crucially, although GenAI can improve readability (Rodler et al., 2025), its health-related outputs still require oversight by medical professionals, and this responsibility clearly cannot be taken on by adult basic education educators only. Moreover, GenAI tools currently lack robust filtering based on linguistic and didactic criteria, limiting their precision, reliability and curricular relevance to health literacy education. To address this gap, we see a need for larger dedicated text corpora with both linguistically simplified and thematically health literacy-relevant content. Open licencing is also essential, as it allows educators to legally adapt and reuse materials.

Participants’ evaluations in this study confirm that low text complexity is seen as necessary for conveying health-related content to older learners. KANSAS was rated positively for key features, particularly for functionalities based on visual cues and filtering options. However, the acceptance of such tools in health literacy and adult basic education remains an open research question. While an earlier study (Kholin et al., 2020) indicates high usability of KANSAS, these findings need to be reassessed for educators the context of health literacy for older adults. Future research should also examine search behaviour in more naturalistic settings, for example, through think-aloud protocols. In addition, the perspectives of older adults regarding their needs, reading strategies and experiences with digitally retrieved texts must be systematically explored to ensure health literacy-supporting tools genuinely meet the needs and requirements of this group of learners.

Limitations

This study has several methodological and substantive limitations. First, the relatively small sample size and the exclusive focus on educators working in Germany limit the generalisability of the findings. Second, the between-subjects design prevented direct comparison of both search engines by the same participants; a within-subjects design could have yielded clearer insights into perceived differences. Third, the online setting did not allow observation of actual search behaviour, making it unclear how consistently participants used available filters and indicators. Fourth, the study focused on educators’ perceptions and selected text features but did not assess learning outcomes or the perspectives of older learners themselves. Finally, KANSAS relies on German-specific readability measures, restricting transferability to other linguistic and educational contexts.

Conclusion

This study highlights both the relevance and challenges of integrating health literacy into adult basic education for older, low-literate learners. While educators acknowledge the growing importance of the topic, there is a need for greater support in addressing the complex intersection of ageing, literacy skills and health-related content. Tools such as KANSAS demonstrate potential in facilitating access to linguistically appropriate materials, yet the overall complexity of available online texts and the lack of domain-specific corpora remain major obstacles.

Looking forward, future research should focus on developing specific training programmes for educators focusing on both older learners’ needs and health literacy, and expanding open-access, health-focused text repositories aligned with low literacy levels. Further empirical studies are required to understand the needs and learning strategies of older adults and to examine how educators can use different digital tools in real teaching contexts. In particular, greater insight into learners’ perspectives is essential. Future research should therefore investigate how older learners engage with educational materials and whether this engagement supports the development of interactive and critical health literacy beyond functional skills. By addressing these gaps, we can take the next step forwards towards enabling equitable access to health information and resources for older adults.

Footnotes

Acknowledgements

We thank the adult basic education educators who participated in this study for sharing their time, experiences and perspectives. We also acknowledge Detmar Meurers and his research group for their collaboration and long-standing support in the development of the KANSAS search engine, including key contributions grounded in computational linguistics expertise. At the time of the collaboration, the group was based at the University of Tübingen but it is now affiliated with the Leibniz-Institut für Wissensmedien (IWM), Tübingen.

Funding

The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: this research was funded by the Bundesministerium für Bildung und Forschung (now the Bundesministerium für Forschung, Technologie und Raumfahrt) as part of the AlphaDekade initiative.

AI statement

Artificial intelligence tools (ChatGPT-5) were used exclusively for language refinement purposes, including checking grammar, improving readability and ensuring stylistic consistency. They were not used for content creation, data analysis, interpretation or conceptual development. The authors accept full responsibility for the accuracy and originality of the content of this paper as published.

ORCID iDs

Mareike Kholin

Sophie Buchwald Pähler

Hannes Schröter

Notes

References

Amstad

(1978) Wie verständlich sind unsere Zeitungen? Dissertation, Universität Zürich, Zürich.

Anderson

(1981) Analysing the readability of English and non-English texts in the classroom with LIX. In: Paper presented at the seventh Australian reading association conference, Darwin, NT, Australia, August. https://files.eric.ed.gov/fulltext/ED207022.pdf

Anderson

(1983) Lix and rix: Variations on a little-known readability index. Journal of Reading 26(6): 490–496.

Berkman

Sheridan

Donahue

, et al. (2011) Low health literacy and health outcomes: An updated systematic review. Annals of Internal Medicine 155(2): 97–107.

Björnsson

(1968) Läsbarhet. Stockholm: Liber.

Brooks

Davie

Duckett

, et al. (2001) Progress in Literacy: Do Learners Learn? London: Basic Skills Agency.

Carter

Nayak

Vembenil

(2024) The accuracy of readability formulas in health content: A systematic review. Journal of Health Care Communication 9(1): 9002.

Cheung

Kemper

(1992) Competing complexity metrics and adults’ production of complex sentences. Applied Psycholinguistics 13: 53–76.

Chinkina

Meurers

(2016) Linguistically aware information retrieval: Providing input enrichment for second language learners. In: Proceedings of the 11th workshop on innovative use of NLP for building educational applications (eds Tetreault

Burstein

C Leacock

, et al.), San Diego, CA, June, pp. 188–198. Stroudsburg, PA: Association for Computational Linguistics.

10.

Christ

Horn

Ambos

(2019) Angebotsstrukturen in der Alphabetisierung und Grundbildung für Erwachsene in Volkshochschulen 2018. Ergebnisse der alphamonitor-Anbieterbefragung des DIE. Bonn: Deutsches Institut für Erwachsenenbildung – Leibniz-Institut für Lebenslanges Lernen e.V.

11.

Comings

Parrella

Soricone

(1999) Persistence among Adult Basic Education Students in Pre-GED Classes (NCSALL Report No. 12). Cambridge, MA: National Center for the Study of Adult Learning and Literacy. Available at: http://www.ncsall.net/index.html@id=652.html (accessed 30 June 2025).

12.

Connell

(1999) Older adults in health education research: Some recommendations. Health Education Research 14(3): 427–431.

13.

Cunningham

Hiebert

Mesmer

(2018) Investigating the validity of two widely used quantitative text tools. Reading and Writing 31(4): 813–833.

14.

Daraz

Morrow

Ponce

, et al. (2018) Readability of online health information: A meta-narrative systematic review. American Journal of Medical Quality 33(5): 487–492.

15.

Dittrich

Weiss

Schröter

, et al. (2019) Integrating large-scale web data and curated corpus data in a search engine supporting German literacy education. In: Proceedings of the 8th workshop on NLP for computer assisted language learning (eds Alfter

Volodina

Borin

, et al.), Turku, 9 September, pp. 41–56. Linköping: LiU Electronic Press.

16.

Federal Statistical Office (2023) 15th coordinated population projection for Germany. Available at: https://service.destatis.de/bevoelkerungspyramide/index.html#!y=2028anda=20,65andl=enandg (accessed 20 July 2023).

17.

Flesch

(1948) A new readability yardstick. The Journal of Applied Psychology 32(3): 221–233.

18.

Friebe

Schmidt-Hertha

Tippelt

(eds) (2014) Kompetenzen im höheren Lebensalter. Ergebnisse der Studie ‘Competencies in Later Life’ (CiLL). Bielefeld: W. Bertelsmann Verlag.

19.

Gass

Mackey

(2012) The Routledge Handbook of Second Language Acquisition. New York: Routledge.

20.

Genuneit

(2011) Analphabetismus im Alter. In: Bothe

(ed.) Funktionaler Analphabetismus im Kontext von Familie und Partnerschaft. Münster: Waxmann, pp. 43–61.

21.

Genuneit

(2016) Analphabetismus und Alphabetisierung im Alter. In: Löffler

Korfkamp

(eds) Handbuch zur Alphabetisierung und Grundbildung Erwachsener. Münster: Waxmann, pp. 250–263.

22.

Ghasemi

Zahediasl

(2012) Normality tests for statistical analysis: A guide for non-statisticians. International Journal of Endicrinology and Metabolism 10(2): 486–489.

23.

Grotlüschen

Buddeberg

Dutz

, et al. (2020) Low literacy in Germany: Results from the second German literacy survey. European Journal for Research on the Education and Learning of Adults 11(1): 127–143.

24.

Hartley

(2004) Designing instructional and informational text. In: Jonassen

Driscoll

(eds) Handbook of Research on Educational Communications and Technology. New York: Routledge, pp. 917–947.

25.

Haynes

Richard

Kubany

(1995) Content validity in psychological assessment: A functional approach to concepts and methods. Psychological Assessment 7(3): 238–247.

26.

Heinemann

AMB

(2011) Alpha-Levels, lea.-Universum, Item-Entwicklung und Feldzugang – zu den Grundlagen des förderdiagnostischen Instruments. In: Grotlüschen

Kretschmann

Quante-Brandt

, et al. (eds) Alphabetisierung und Grundbildung, Band 6. Literalitätsentwicklung von Arbeitskräften. Münster: Waxmann, pp. 86–107.

27.

Horner

Surratt

Juliusson

(2000) Improving readability of patient education materials. Journal of Community Health Nursing 17(1): 15–23.

28.

Institute of Medicine (2004) Health Literacy: A Prescription to End Confusion. Washington, DC: The National Academies Press.

29.

Kholin

Gladkova

Weber

, et al. (2020) Digitale Tools in Alphabetisierungskursen als Brücke zwischen Standardisierung und Binnendifferenzierung am Beispiel von KANSAS. Alfa-Forum 97: 28–33.

30.

Kopera-Frye

(2016) Health literacy 101. In: Kopera-Frye

(ed.) Health Literacy among Older Adults. New York: Springer, pp. 1–16.

31.

Jordan

(2023) Gesundheitskompetenz/health literacy. In: Bundeszentrale für gesundheitliche Aufklärung (BZgA) (Hrsg.) Leitbegriffe der Gesundheitsförderung und Prävention. Glossar zu Konzepten, Strategien und Methoden. Available at: https://doi.org/10.17623/BZGA:Q4-i065-3.0

32.

Krashen

(1977) Some issues relating to the monitor model. On TESOL 77: 144–158.

33.

Kuhlmey

Hitzblech

Schnitzer

(2008) Alter und Gesundheit: Vorstellungen in unterschiedlichen Generationen. In: Böcken

Braun

Almhof

(eds) Gesundheitsmonitor 2008. Gesundheitsversorgung und Gestaltungsoptionen aus der Perspektive der Bevölkerung. Gütersloh: Bertelsmann Stiftung, pp. 210–230.

34.

Kulmus

Buddeberg

Kosinski

, et al. (2025) Still different: Adults with low literacy skills and different age cohort experiences in eastern and western German federal states. Studies in the Education of Adults. Epub ahead of print 11 March. DOI: 10.1080/02660830.2025.2475542.

35.

Kynette

Kemper

(1986) Aging and the loss of grammatical forms: A cross-sectional study of language performance. Language and Communication 6: 65–72.

36.

Light

(1993) Language changes in old age. In: Blanken

Dittmann

Grimm

, et al. (eds) Linguistic Disorders and Pathologies: An International Handbook. Berlin: De Gruyter, pp. 900–918.

37.

Manly

Schupf

Tang

, et al. (2005) Cognitive decline and literacy among ethnically diverse elders. Journal of Geriatric Psychiatry and Neurology 18(4): 213–217.

38.

Mayer

Jambor-Fahlen

Kholin

(2023) Einstellungen von Lehrpersonen zu digitalen Medien und die effektive Nutzung einer Suchmaschine (KANSAS). Lernen und Lernstörungen 12(2): 95–106.

39.

McInnes

Haglund

(2011) Readability of online health information: Implications for health literacy. Informatics for Health & Social Care 36(4): 173–189.

40.

Memon

West

(2024) Search engines post-ChatGPT: How generative artificial intelligence could make search less reliable. arXiv [preprint]. DOI: 10.48550/arXiv.2402.11707.

41.

Naveh-Benjamin

Cowan

(2023) The roles of attention, executive function and knowledge in cognitive ageing of working memory. Nature Review Psychology 2: 151–165.

42.

Nomura

Nielsen

Tronbacke

(2010) Guidelines for Easy-to-Read Materials, IFLA Professional Reports, No. 120. The Hague: International Federation of Library Associations and Institutions.

43.

Nutbeam

(2008) The evolving concept of health literacy. Social Science & Medicine 67(12): 2072–2078.

44.

Nutbeam

Lloyd

(2021) Understanding and responding to health literacy as a social determinant of health. Annual Review of Public Health 42(1): 159–173.

45.

Ortmanns

Lux

Bachem

, et al. (2024) Volkshochschul-Statistik. 62. Folge, Berichtsjahr 2023. Bielefeld: wbv.

46.

Pähler

Kholin

(2023) KANSAS: Die kompetenzadaptive, nutzerorientierte Suchmaschine für authentische Sprachlerntexte (Manual). Available at: https://kansas-suche.de/kansas/manual.pdf

47.

Peterson

(1956) Comparison of Flesch readability scores with a test of reading comprehension. Journal of Applied Psychology 40(1): 35.

48.

Petersson

Ingvar

Reis

(2009) Language and literacy from a cognitive neuroscience perspective. In: Olson

Torrance

(eds) The Cambridge Handbook of Literacy. Cambridge: Cambridge University Press, pp. 152–181.

49.

Peuker

Johannsen

Langemack

, et al. (2022) Grundbildung in der Lebenswelt verankern: Praxisbeispiele, Gelingensbedingungen Und Perspektiven. Bielefeld: wbv.

50.

Quenzel

Schaeffer

(2016) Gesundheitskompetenz vulnerabler Bevölkerungsgruppen. Bielefeld: Universität Bielefeld.

51.

Rammstedt

Gauly

Kapidzic

, et al. (2024) PIAAC 2023. Grundlegende Kompetenzen Erwachsener im internationalen Vergleich. Münster: Waxmann Verlag.

52.

Rodler

Cei

Ganjavi

, et al. (2025) GPT-4 generates accurate and readable patient education materials aligned with current oncological guidelines: A randomized assessment. PLoS ONE 20(6): e0324175.

53.

Roman

(2004) Illiteracy and older adults: Individual and societal implications. Educational Gerontology 30(2): 79–93.

54.

Schaeffer

Berens

Gille

, et al. (2021) Gesundheitskompetenz der Bevölkerung in Deutschland – vor und während der Corona Pandemie: Ergebnisse HLS-GER 2. Bielefeld: Interdisziplinäres Zentrum für Gesundheitskompetenzforschung (IZGK), Universität Bielefeld.

55.

Schneider

Beege

Nebel

, et al. (2018) A meta-analysis of how signaling affects learning with media. Educational Research Review 23(1): 1–24.

56.

Schröter

Bar-Kochva

(2019) Keyword: Reading literacy. Reading competencies in Germany and underlying cognitive skills. Zeitschrift für Erziehungswissenschaft 22(1): 17–49.

57.

Sørensen

Pelikan

Röthlin

, et al. (2015) Health literacy in Europe: Comparative results of the European Health Literacy Survey (HLS-EU). European Journal of Public Health 25(6): 1053–1058.

58.

Sørensen

Van den Broucke

Fullam

(2012) Health literacy and public health: A systematic review and integration of definitions and models. BMC Public Health 12: 80.

59.

Speros

(2009) More than words: Promoting health literacy in older adults. Online Journal of Issues in Nursing 14(3): 5. Available at: www.nursingworld.org/MainMenuCategories/ANAMarketplace/ANAPeriodicals/OJIN/TableofContents/Vol142009/No3Sept09/Health-Literacy-in-Older-Adults.aspx (accessed 30 June 2025).

60.

Van Gog

(2014) The signaling (or cueing) principle in multimedia learning. In: Mayer

(ed.) The Cambridge Handbook of Multimedia Learning. Cambridge: Cambridge University Press, pp. 263–278.

61.

Van Kolfschooten

Pilottin

(2024) Reinforcing stereotypes in health care through artificial intelligence–generated images: A call for regulation. Mayo Clinic Proceedings: Digital Health 2(3): 335–341.

62.

Vajjala

Meurers

(2013) On the applicability of readability models to web texts. In: Proceedings of the second workshop on predicting and improving text readability for target reader populations, Sofia, August, pp. 59–68. Stroudsburg, PA: Association for Computational Linguistics.

63.

Weiss

Dittrich

Meurers

(2018) A linguistically-informed search engine to identify reading material for functional illiteracy classes. In: Proceedings of the 7th workshop on NLP for computer assisted language learning, Stockholm, 7 November, pp. 79–90. Linköping: LiU Electronic Press.