The Art Nouveau Path: Valuing Urban Heritage Through Mobile Augmented Reality and Sustainability Education

Abstract

Cultural heritage is framed as a living resource for citizenship and education, although evidence on how in situ augmented reality can cultivate sustainability competences remains limited. This study examines the Art Nouveau Path, a location-based mobile augmented reality game across eight points of interest in Aveiro, Portugal, aligned with the GreenComp framework. Within a design-based research case study, the analysis integrates repeated cross-sectional student questionnaires (S1-PRE N = 221; S2-POST N = 439; S3-FU N = 434), anonymized gameplay logs from 118 collaborative groups, and 24 teacher field observations (T2-OBS), using quantitative summaries with reflexive thematic analysis. References to heritage preservation in students’ sustainability conceptions rose from 28.96% at baseline to 61.05% immediately after gameplay, remaining above baseline at follow-up (47.93%). Augmented reality items were answered more accurately than non- augmented reality items (81% vs. 73%) and involved longer on-site exploration (+10.17 min). Triangulated evidence indicates that augmented reality and multimodality amplified attention to architectural details and prompted debates about authenticity. Built heritage, mobilized through lightweight augmented reality within a digital teaching and learning ecosystem, can serve as an effective context for Education for Sustainable Development, strengthening preservation literacy and civic responsibility and generating interoperable cultural traces for future reuse.

1. Introduction

Over recent decades, a paradigm shift has taken place in the understanding of cultural heritage, which is now regarded as more than merely a static collection of monuments, artefacts, and traditions [1,2]. International frameworks increasingly recognize heritage as a living resource for identity, social cohesion, and sustainable futures. The 1972 United Nations Educational, Scientific and Cultural Organization (UNESCO) World Heritage Convention affirmed cultural and natural assets as part of a shared global legacy [3]. More recently, the Council of Europe’s Faro Convention [4] advanced a people-centered perspective that foregrounds the social, civic, and educational roles of heritage. These instruments promote a systemic vision that embeds conservation within broader urban agendas of sustainability, participation, and resilience, consistent with the Sustainable Development Goals (SDGs) [5], particularly SDG 4.7, which emphasizes Education for Sustainable Development by calling for the integration of sustainability competences, human rights, and global citizenship into formal curricula, and SDG 11.4, which calls for strengthened efforts to protect and safeguard cultural and natural heritage as part of strategies to foster inclusive, safe, adaptable, and sustainable urban environments. Heritage is thus understood not solely as a domain for specialist preservation, but also as a civic resource that cultivates values, competencies, and a shared sense of responsibility.

These policy shifts have been accompanied by similar academic debates. In heritage studies, researchers have argued that heritage is inherently pluralistic and values-based, shaped by social negotiations rather than just expert discourses [6,7,8]. Bandarin and van Oers [9] conceptualize urban heritage as a catalyst for community resilience and cohesion, while Champion and Rahaman [10] explore how digital storytelling and narrative coherence mediate meaning-making. Recent works by Moraitou and colleagues [11] and Katsianis and Gadolou [12] have further reinforced the idea that heritage gains significance when it is connected to everyday practices and collective futures. Accordingly, heritage is not only something to preserve but also a ‘living space for learning and participation’ [13] that should be actively mobilized. At the same time, Choay [14] has critically noted that heritage often remains detached from sustainability discussions in everyday life, risking aestheticization or commodification instead of fostering active civic engagement. This critique is grounded in the view that natural, cultural, tangible, and intangible heritage is unique and therefore non-renewable, a status that is frequently sidelined in sustainability discourse [15,16]. Bridging this remains a significant challenge for both research and practice.

Digital technologies applied to cultural heritage have evolved rapidly, changing how cultural assets are documented, preserved, and shared with the public. Initial research focused on photogrammetry and 3D modeling but has since expanded to include immersive media like augmented reality (AR) and virtual reality (VR), collectively referred to as extended reality (XR), alongside geographic information systems (GIS) and building information modeling (BIM) [17]. Concurrently, advances in semantic data models and interoperability standards, such as the International Committee for Documentation of the International Council of Museums Conceptual Reference Model (CIDOC-CRM) [18] and the Findable, Accessible, Interoperable, and Reusable Data Principles (FAIR) [19], underscore the necessity to organize digital assets ensuring their reuse, accessibility, and integration across diverse platforms [20,21]. The field is gradually shifting from isolated digital replicas to interconnected ecosystems that link heritage with infrastructures for knowledge production, education, and civic participation.

Within this context, digital mediation enables heritage to be rediscovered in situ [22,23]. Location-based applications and AR have been shown to draw attention to details that would otherwise go unnoticed, connecting contemporary streetscapes with historical imagery, and thereby transforming urban exploration into a coherent narrative rather than a series of isolated stops [10,24,25]. Evidence from other studies indicates benefits for both interpretation and preservation [2,26,27]. In parallel, semantic and cartographic approaches help maintain continuity across places and media [28,29,30], positioning AR not merely as a technical layer but as a mediational tool that shapes narrative and cultural meaning across contexts.

Building on these possibilities, AR has attracted particular interest for crafting situated experiences. By layering digital information onto in situ settings, it can foster careful observation and reflection on authenticity. Empirical studies report gains in interpretation and attention [31,32,33], but also recurring issues such as distraction and short-lived novelty. Most of this work has been conducted in museums or tourism contexts, with brief engagements and limited focus on long-term competence development in formal education [34]. Also, few studies track retention, transfer, or civic dispositions weeks after an intervention. Longitudinal or repeated cross-sectional studies are still rare, which constrains understanding AR’s effect on Education for Sustainable Development (ESD) when embedded in curricula and aligned with competence frameworks [34].

In the European context, with the European Sustainability Competence Framework [35], the GreenComp have consolidated a competence-oriented view of sustainability learning that integrates knowledge, skills, and attitudes. This framework is intended to guide curriculum innovation and assessment, although its educational implementation remains emergent and challenging [36]. Conventional classrooms often lack the immediacy and contextual relevance needed to mobilize sustainability values and civic responsibility. Heritage-contextualized environments, with their tangible, unique, and symbolic qualities, offer a promising, although still underexplored, setting for cultivating sustainability competences [37,38].

This work addresses that gap through a case study in which built heritage is used as a local context for AR-mediated, GreenComp-oriented learning [35]. Against this backdrop, the city of Aveiro, Portugal, presents an interesting case. Aveiro hosts one of the country’s most significant ensembles of Art Nouveau façades [39,40], recognized though its membership in the Réseau Art Nouveau Network (RANN), the establishment of the Aveiro’s Art Nouveau Museum, and sustained municipal initiatives that safeguard and promote this heritage through curated urban paths. Despite this visibility, educational initiatives have largely relied on guided visits and tourism-oriented narratives, with limited evidence of lasting effects on how young people value this built environment. The ornamental details of Aveiro’s Art Nouveau are inspired by local fauna and flora and expressed in monuments, like the ‘Obelisk of Freedom’ (Figure 1), and on façades through floral motifs and other inspirations on wrought iron works, tiles, and architectural details (as in Figure 2). These elements offer an underused resource for close observation and interpretation.

In line with Gruenewald’s [41] place-based pedagogy, the locally grounded features render Aveiro’s Art Nouveau heritage a distinctive context for competence-oriented sustainability education and a well-suited case study for this research.

This study examines the Art Nouveau Path, a mobile augmented reality game (MARG) developed within the EduCITY Digital Teaching and Learning Ecosystem (DTLE) (https://educity.web.ua.pt/, accessed on 4 November 2025), in order to explore its potential. The MARG links GreenComp competences [35] to site-specific tasks by combining AR overlays, multimodal media, and narrative dynamics across eight Art Nouveau heritage points of interest (POI) in Aveiro. Implementation activities were organized around collaborative students’ work in small groups (two to three students per group), promoting engagement with both the MARG and in situ built heritage. Students were invited to explore multiple architectural details and connected them, through storytelling, to local history, historical people, and sustainability competence development. During the implementation, students were accompanied by their teachers, who provided structured observations, complementing the gameplay logs and students’ questionnaires.

The broader research adopts a design-based research (DBR) approach and was implemented in three cross-repeated, cross-sectional phases: a baseline (S1-PRE), an immediate post-test (S2-POST), and a follow-up six to eight weeks later (S3-FU). The phases are named after the questionnaires completed by the students. Data sources include three GreenComp-based questionnaires (GCQuest) [42], in a version tailored for this MARG, available at the project’s Zenodo community page [43], gameplay logs comprising 4248 group-item responses, and ecological observations from 24 teachers (T2-OBS). By triangulating these datasets, this study examines whether a heritage-based AR intervention can, in this specific context, enhance engagement with cultural assets and promote sustainability competences.

The following research questions (RQs) were formulated to guide this study:

(RQ1) Can heritage serve as an effective context for ESD?

(RQ2) How do multimodality and augmented reality affect engagement and learning outcomes?

(RQ3) To what extent do students retain and transfer heritage-related sustainability competencies after gameplay?

Addressing these questions advances debate in three areas:

(i) The promotion of cultural heritage as an educational asset; (ii) the importance of digital heritage preservation; and (iii) the value of place-based semantic enrichment.

In parallel, this study examines the educational value of AR, moving beyond novelty to support long-term competence development. The findings therefore may have implications for the design of educational games and for broader strategies that embed cultural heritage within sustainable digital learning ecosystems.

In addition to its educational focus, this study aims to contribute to heritage preservation practice by providing lightweight digital documentation and structured descriptors of georeferenced cultural assets. This lightweight approach is guided by oriented digital records that are technically simple and low-cost, but sufficiently structured to remain reusable and interoperable [19,29]. Therefore, it adopts a pragmatic, FAIR-oriented approach to semantic organization and data stewardship, aiming to facilitate interpretive reuse, civic and educational engagement, and future interoperability with cultural knowledge bases [19].

Following the introduction, Section 2 presents a narrative thematic review of the literature and theoretical frameworks. Section 3 describes the methodological design, including the context, the participants, the instruments, and the DBR approach. Section 4 reports the findings. Section 5 discusses the pedagogical and methodological implications of these results. The final section synthesizes the key contributions, identifies limitations, and suggests paths for future research.

2. Theoretical Framework

This section reports a narrative, thematic literature review [44,45,46] conducted in line with established procedures for thematic synthesis, integrating inductive and deductive coding to ensure conceptual coherence across domains [47,48].

Given the interdisciplinary scope of this study, the theoretical framework was organized into five categories:

(1)

International frameworks for heritage preservation;

(2)

Art Nouveau as a cultural resource and as an urban identity asset;

(3)

Extended reality approaches applied to heritage, considering AR as the primary technology;

(4)

Semantic and geospatial logics for structuring and linking heritage data;

(5)

Smart heritage agendas towards the promotion of interoperability, openness, and long-term preservation.

Besides these categories, a transversal core focus is present in the broader research. This regards Education for Sustainability and competence development, with preservation-relevant data practices positioned as complementary and mutually reinforcing.

Searches were conducted in Scopus and Web of Science and supplemented with exploratory searches in Google Scholar to capture gray literature and institutional reports. Additionally, some literature was previously used in already published works [49,50].

The search period was April–September 2025, targeting works published between 2012 and 2025. Effective keyword combinations included (“augmented reality” OR “mobile augmented reality” OR “mobile augmented reality game” OR MARG) AND “cultural heritage” AND (education OR learning); “Art Nouveau heritage” AND education; (“narrative cartography” OR “spatial storytelling”) AND (mapping OR heritage); (“semantic data enrichment” OR CIDOC-CRM OR “cultural heritage ontology” OR “semantic trajectory”); (“digital heritage preservation” OR “cultural heritage interpretation”); (“valuing urban heritage” OR “heritage valorization”); (“sustainability education” OR “education for sustainable development” OR GreenComp); and (“digital teaching and learning ecosystem” OR DTLE) AND sustainability. Direct searches using “Art Nouveau” predominantly returned art-historical records. We retained only works intersecting education, AR/MARG, geoinformation/trajectory, or competence frameworks and excluded the remainder.

Studies were included if they (1) were peer-reviewed and indexed in Scopus or Web of Science, (2) addressed Education for Sustainable Development and/or sustainability competences, such as GreenComp, in formal or non-formal education, and (3) were clearly connected to at least one core focus of this paper. Exclusion criteria comprised (1) AR or XR studies focused solely on technical aspects without pedagogical framing or a link to ESD or competences, (2) tourism-oriented studies and museum studies lacking geoinformation or trajectory components, educational analysis, or in situ built-heritage context, (3) purely theoretical reflections without empirical or design-based components, (4) VR-only studies without a clear bridge to AR in educational heritage settings, and (5) duplicates or records thematically irrelevant to this study’s scope.

The database search retrieved 74 records. After de-duplication and abstract screening, 55 items were retained for full-text review. Applying the inclusion and exclusion criteria yielded 35 peer-reviewed articles. To ensure conceptual breadth, the database-derived corpus was complemented with purposively selected seminal conceptual works, key policy and institutional documents, and methodological references on mixed-methods integration, some of which were not directly retrieved through the initial database queries but are central to the conceptual and analytical framing of this study. The theoretical corpus comprises 75 sources, described here in three categories: (1) 52 peer-reviewed articles (4 from prior research outputs, namely [49,50], that concern this broader project), (2) 12 policy and institutional frameworks, and (3) 11 books and monographs. Methodological contributions that primarily guide the mixed-methods and integrative analytical strategy are not detailed in this subsection; they are, however, fully listed in Appendix A and in the reference list.

A hybrid thematic analysis was undertaken, integrating inductive and deductive coding. Following Boyd [46], multiple reasoning modes were iteratively applied to ensure conceptual coherence across the five previously identified domains. The policy frameworks and the reference works grounded the analysis in internationally recognized sources, while the authorship-related publications secured continuity with prior and broader research.

The following subsections examine the five domains in detail. They begin with international frameworks for heritage preservation and conclude with a broader synthesis.

2.1. International Frameworks for Heritage Preservation

Over the past thirty years, international policy has increasingly framed cultural heritage as a key element for promoting urban sustainability, fostering community participation, and strengthening resilience. The European Convention on the Value of Cultural Heritage for Society, commonly known as the Faro Convention [4], together with the UNESCO Recommendation on the Historic Urban Landscape (HUL) [51], exemplifies this shift. The HUL approach redefines historic cities as living, evolving socio-ecological systems rather than static relics, and it advocates for conservation practices tailored to local contexts so that heritage is integrated into urban planning, cultural economies, and broader societal well-being. Within this policy landscape, the SDGs [5], especially SDG 11.4 The New Urban Agenda, adopted in Quito in 2016 [52], further underscore the role of culture and heritage in supporting civic identity, memory, urban prosperity, and resilience to environmental and human-induced challenges.

The Faro Convention [4] is distinctive in explicitly recognizing cultural heritage not only as a human right but also as a shared societal resource, foregrounding participation, identity, and quality of life. Building on this foundation, the Council of Europe’s European Heritage Strategy for the 21st Century [53] translates these ideas into concrete priorities for heritage education, equitable access, and the sustainable management of cultural assets. More recent European initiatives, such as the Common European Data Space for Cultural Heritage [54] and the Twin it! 3D for Europe’s Culture campaign [55], further emphasizes the value of digitally documenting cultural assets in interoperable formats that support long-term preservation, openness, and reuse across sectors. These developments connect heritage preservation with the emergence of digital ecosystems and semantic infrastructures that make data sharing and knowledge exchange possible.

These policy developments are reinforced by academic research. Bandarin and van Oers [9] argue that the HUL approach [51] can turn heritage into a catalyst for sustainable urban development when it is embedded in broader socioeconomic and environmental strategies, while King [56] highlights how culturally grounded, digitally mediated participation and co-creation can expand heritage’s value beyond technical conservation, foregrounding identity, social cohesion, and civic inclusion. Avrami and colleagues [8] frame heritage management as a value-based, multi-stakeholder process that acknowledges contested meanings as an integral part of sustainable heritage practices. Ababneh’s [57] recent re-reading of the Venice Charter [58] underlines that contemporary conservation must integrate digital documentation, interoperability, and community-centered approaches.

In this study, the policy frameworks and academic research analyzed together reinforce the choice to work with the Art Nouveau façades in Aveiro and, more broadly, to use built heritage as a place-based context for competence-oriented ESD within a wider urban sustainability agenda.

2.2. Art Nouveau as a Cultural Resource and as an Urban Identity Asset

Emerging at the turn of the twentieth century, the Art Nouveau became one of the most emblematic artistic and cultural movements of European Modernism [59]. Conceived as a ‘Gesamtkunstwerk’, or total work of art [60], it integrated architecture, decorative arts, graphic design, and urban culture into a unified aesthetic language of floral motifs, undulating lines, symbolic figures, and organic compositions that expressed ideals of progress, modernization, and harmony between art, nature, and everyday life.

Studies about Art Nouveau emphasize its dual character [61,62]. On the one hand, it was cosmopolitan, circulating through cities such as Brussels, Paris, Vienna, Barcelona, and Riga; on the other, it remained regionally distinctive, absorbing local traditions, natural inspired motifs, materials, craft practices, and also aspirations.

However, the patrimonialization of Art Nouveau has been uneven [63]. Some cities, such as Brussels with its UNESCO-listed townhouses, Paris with its metropolitan façades, and Riga, have incorporated the style into their cultural identity and tourism strategies, while other contexts have remained comparatively marginal in both policy and research.

In Portugal, Aveiro concentrates one of the country’s most significant ensembles of Art Nouveau façades. This collection has been extensively documented by local researcher Amaro Neves in several works, published from 1980 to mid-2000 [64], and promoted by civic associations, such as the ‘Associação para o Estudo e Defesa do Património Natural e Cultural da Região de Aveiro’ [Association for the Study and Defense of the Natural and Cultural Heritage of the Aveiro Region] (ADERAV), as well as through municipal including the ‘Art Nouveau Museum’ and a dedicated Art Nouveau Route. The façades, with their wrought iron, tiles, and naturalist reliefs, function as anchors of urban identity by connecting the contemporary city to early twentieth-century histories of modernization and bourgeois prosperity.

Despite its value, some heritage researchers argue that treating such façades as static monuments reduces their cultural potential [65]. Choay [14] critiques the fetishization of heritage as isolated objects detached from social life, while Avrami et al. [8] emphasize heritage as a dynamic, plural, and contested resource whose meanings evolve through community engagement and values-based management.

Recent studies show that digital documentation, mobile applications, and gamified experiences can extend this integration by broadening access and fostering more sustained engagement with heritage sites [23].

In this study, the Art Nouveau Path conceptualizes Aveiro’s façades as living cultural resources rather than mere aesthetic surfaces for tourist consumption. Through the MARG’s narrative and multimodal media, such as AR contents, students are invited to attend to architectural motifs, inscriptions, and symbolic references, and to relate them to broader issues such as environmental change, water use, or civic memory. In this way, the façades are reactivated as urban identity assets that mediate between past and present, architecture and environment, and memory and belonging, providing a place-based foundation for competence-oriented ESD.

This orientation aligns with situated learning and place-based pedagogy, which argue that learning is deepened when anchored in real settings and local narratives [41,66]. These design choices depend on specific forms of digital mediation, and in particular, mobile AR, which are examined in the following.

2.3. Extended Reality for Heritage: Augmented Reality as a Primary Approach

XR technologies, particularly AR and VR, have become prominent in heritage documentation, reconstruction, interpretation, and education [26,31,32,33]. VR offers immersive reconstructions in simulated environments, whereas AR affords in situ mediation by layering digital content directly onto existing sites, façades, and artefacts. This makes AR especially suitable for visualizing and interpreting urban heritage within the constraints of everyday urban life, education, tourism, and conservation [9,26,51].

Empirical studies report that AR can enhance accessibility, engagement, and learning when narrative overlays are coherent with the physical context and supported by gamification and multimodal resources [10,32,67]. Other works show that AR can support conservation-related practices by visualizing restoration options, monitoring deterioration, digitally preserving historical layers, and reducing physical pressure on sensitive sites through digital alternatives for exploration [2,23,66,67,68,69,70].

Recent reviews position AR as part of a broader digital heritage toolkit rather than a temporary novelty. AR can promote inclusivity through multisensory channels of engagement [23], connect semantic and geospatial information via overlays that link to heritage information systems [29], and support active, inquiry-based learning in which participants develop knowledge through observation, questioning, and reflection [71]. For education, these affordances are particularly relevant when working with in situ heritage, since they allow learners to relate architectural details and historical narratives to contemporary concerns such as sustainability, identity, and civic responsibility.

This conceptualization provides the backdrop for analyzing how multimodality and AR relate to students’ engagement with Aveiro’s Art Nouveau heritage and to their perceived sustainability competences in the empirical sections that follow.

2.4. Semantic and Geospatial Logics in Heritage Data

The sustainability of digital heritage depends not only on producing digital representations but also on how these are semantically structured and geospatially anchored. Semantic reference models such as CIDOC-CRM [18], recently standardized as ISO 21127:2023 [72], provide shared vocabularies for representing heritage entities, events, and relations across museums, archives, and built environments. Semantic and graphical knowledge models enable interoperability, expressive querying, and long-term reuse, allowing heritage data to circulate between institutions and platforms without losing context or meaning [11].

Geospatial infrastructures complement these models by linking heritage information to the territory and context. Archaeological and architectural studies increasingly use GIS-based resources to map sites, analyze spatial relationships, and construct narrative cartography [12]. Narrative cartography reconfigures maps as storytelling instruments that weave together historical, cultural, and social layers [25], often aligned with GIS platforms that bridge detailed architectural data with broader urban analysis [73].

Within broader linked open data agendas, semantic and geospatial forms of enrichment are treated as cultural tools rather than purely technical layers. Heritage datasets are connected to wider knowledge ecosystems instead of remaining in isolated repositories, and semantic trajectories and enriched metadata are used to organize visitor navigation and comprehension across locations and media [29]. In this view, semantics and spatial reasoning promote unity and continuity in how heritage is experienced and interpreted.

Even small-scale projects can align with these principles by implementing proportionate solutions. In the Art Nouveau Path, the MARG adopts controlled vocabularies for key attributes, such as architects, construction periods, materials, and motifs, together with consistent descriptors and georeferenced navigation. Rather than implementing full CIDOC-CRM [18] integration or complex GIS stacks, the project uses a lightweight schema that keeps each façade embedded in a coherent narrative while preserving the possibility of future alignment with richer semantic or GIS infrastructures. Here, lightweight digital preservation refers to a proportionate bundle of interoperable records that is less resource-intensive than full Heritage Building Information Modeling (HBIM) or digital twins, but still structured enough to support reuse, interoperability, and educational interpretation.

2.5. Smart Heritage Agendas, Interoperability, and Long-Term Preservation

Smart heritage has been used to describe the convergence of cultural heritage with digital ecosystems grounded in compatibility, openness, and cooperative governance. In the European context, initiatives such as the Common European Data Space for Cultural Heritage [54] and the Twin it! 3D for Europe’s culture campaign [55] express the expectation that cultural assets are documented with high quality, shared in interoperable formats, and reused across domains ranging from education to tourism and scientific research. These initiatives align with the principles of FAIR data management [19], which state that data should be findable, accessible, interoperable, and reusable, and together set a benchmark for the curation and long-term preservation of heritage datasets.

Digital preservation requires more than technical storage. Niccolucci [20] and Moullou et al. [21] argue that the sustainability of digital heritage depends on treating datasets as cultural resources supported by versioning, provenance records, and governance mechanisms that ensure continuity across technological changes. From this perspective, interoperability is not only a technical feature but a prerequisite for heritage data to operate within wider cultural ecosystems that connect archives, museums, built heritage inventories, and community-generated resources.

The Art Nouveau Path engages with these agendas through a lightweight but structured model of data production and preservation. Rather than creating full-scale digital twins or exhaustive repositories, the project generates a multimodal corpus that includes enriched visual records of façades (archival photographs, AR overlays, short videos, and narrative descriptions), gameplay logs that document group decisions in situ, GreenComp-based questionnaires administered before, immediately after, and some weeks after gameplay, and teacher validations and ecological observations. These materials are organized along the path and sustainability themes within a consistent storyline.

By foregrounding interoperability and openness as guiding principles, the broader project positions its datasets as cultural resources that extend beyond the immediate educational intervention. While modest in scale compared to national digitization campaigns, the Art Nouveau Path illustrates how smart heritage practices can be embedded in educational games and community projects.

2.6. Synthesis

The theoretical framework outlined above frames the Art Nouveau Path at the intersection of international preservation agendas, stylistic and urban identity discourses, technological innovation, semantic infrastructures, and smart heritage practices. Across these five strands stands a common premise emerges: cultural heritage is not as a static set of monuments but as part of dynamic cultural ecosystems that require mediation, documentation, and governance in both digital and material domains.

Within this perspective, international and European policy frameworks present heritage as a driver of sustainability, resilience, and participation, while initiatives such as the Common European Data Space for Cultural Heritage [54] promote interoperability, openness, and reuse. Art Nouveau, in turn, is understood as a cultural resource and urban identity asset whose locally inflected motifs provide a concrete setting for place-based education in Aveiro. The Art Nouveau Path draws on these strands by treating Aveiro’s façades and monuments as interactive anchors for competence-oriented ESD, inviting students to connect architectural details and local narratives with broader issues of environmental change, preservation, and civic memory.

The project also builds on research on AR for heritage education by using mobile AR as a mediational layer that links multimodal content and tasks to specific façades and points of interest. Narrative overlays, quiz prompts, and in situ challenges are designed to scaffold careful observation and interpretation without removing learners from the urban context. At the same time, lightweight semantic descriptors and georeferenced navigation connect the game to emerging digital heritage ecosystems, while the FAIR-oriented treatment of gameplay logs, questionnaires, and teacher observations positions the resulting datasets as reusable cultural resources.

In this way, the Art Nouveau Path operationalizes the five domains of the framework in a coherent manner. It uses built heritage as a meaningful context for ESD (RQ1), mobilizes multimodality and AR to support engagement and learning (RQ2), and structures semantic, spatial, and smart heritage elements so that students’ trajectories and questionnaire responses can be interpreted in relation to the retention and transformation of sustainability competences over time (RQ3). These conceptual choices underpin the methodological design detailed in the following section.

3. Materials and Methods

3.1. Research Design and Methodological Orientation

This study adopts a DBR approach to examine how a location-based MARG can position urban built heritage as a resource to be preserved, interpreted, and used for educational purposes in authentic settings. DBR combines cycles’ design, enactment, analysis, and redesign in real-world contexts, integrating cultural heritage content and technological affordances while maintaining ecological validity [74,75]. Within this broader program, the Art Nouveau Path functions simultaneously as an intervention and an object of study, with earlier iterations guided by explicit design principles and preliminary evaluations reported in previous works [49,50].

In practical terms, the DBR program has so far involved four iterative cycles. The first one regarded the MARG and its content design. After this, in an exploratory cycle, a prototype of the Art Nouveau Path was designed and piloted on-site to test the basic narrative structure, AR triggers, and quiz mechanics. The MARG was tested by different researchers. A third cycle focused on validation and curricular review with in-service teachers (T1-VAL and T1-R), leading to refinement in item wording, sequencing of POIs, and alignment with GreenComp-related learning goals. The present article reports on the fourth cycle, in which the revised MARG was implemented with school groups under authentic conditions and systematically documented through questionnaires, teacher observations, and gameplay logs.

Methodologically, this work is framed as an exploratory case study [76,77] centered on the Art Nouveau heritage in Aveiro. This design supports context-sensitive research in situations where the phenomena and their boundaries are not fully predefined, allowing research questions to be refined and propositions to be generated for subsequent cycles [76,77,78,79,80]. Rather than aiming at exhaustive 3D digitization or full Heritage HBIM pipelines, the project investigates a proportionate, lightweight approach to digital preservation oriented towards learning and sustainability awareness through the valorization of built heritage. This involves producing and curating interoperable cultural traces, including map-anchored narratives, AR overlays, archival media, structured descriptors, gameplay logs, and questionnaire data [11,73].

Given the multi-component nature of the wider EduCITY project, the present article delineates a specific analytical scope. The validation of the MARG with in-service teachers, which ensured curricular coherence and informed iterative design, has been published in previous works [49,50] and is referenced here only as contextual support. The focus of this study is on the implementation with students and their teachers in situ, the heritage-relevant data flows generated during gameplay, and the digital assets incorporated into the MARG. The three administrations of the GCQuest instrument (S1-PRE, S2-POST, S3-FU) are treated as a repeated cross-sectional design, without individual-level pairing across waves, and subsequent analyses are interpreted accordingly. In line with this design and the exploratory aims of this work, questionnaire results are reported using descriptive statistics only, without inferential statistical testing, and subsequent analyses are interpreted accordingly.

3.2. Context of This Study and Intervention

Aveiro, Portugal, offers a compact, walkable historic center in which early twentieth-century Art Nouveau façades are strongly concentrated. Their wrought iron balconies, tiles, and fauna and floral reliefs have been documented in local studies, municipal catalogues, and university research, making the city an appropriate setting for Art Nouveau heritage-based, place-anchored learning activities [39,40,61,62,81,82]. The organic vocabulary and craft traditions of this heritage align with discussions on cultural identity and environmental awareness that underpin the sustainability-oriented mediation explored in this study.

Within this context, the Art Nouveau Path was designed as a location-based MARG that turns the historic center into an information-rich itinerary. The game is structured around eight georeferenced POIs in the city center. At each POI, students access map-based navigation, on-site AR overlays anchored in architectural details, and a compact set of interpretive resources, including digitized archival photographs, brief narrative prompts, and semantic descriptors. Across the eight POIs, participants engage with 36 quiz-type items following a common structure of orienting cue, four-option multiple-choice questions, and immediate explanatory feedback (Figure 3). This structure is intended to sustain attention to material details while linking built heritage to multidisciplinary curriculum areas and to sustainability discourses framed by the GreenComp [35].

The in-app city map presents the POIs and path through the historic center, supporting wayfinding and situational awareness (Figure 4).

This supports wayfinding and situational awareness in ways consistent with mobile pedestrian navigation (MPN) systems that couple spatial guidance with contextual information.

An AR camera functions as an interpretive layer that overlays digital content on selected ornaments and façade elements (Figure 5).

Simple visual hints indicate the presence of markers and directional cues guide movement between locations (Figure 6 and Figure 7). These features were designed to stabilize navigation so that students can focus their cognitive resources on observation and interpretive tasks rather than on searching for the correct façades.

Conceptually, the intervention draws on narrative cartography by coordinating movement, observation, and meaning-making across the route rather than confining interpretation to a single site [12,25].

Developed between 2023 and 2024 as part of a doctoral research program, the MARG was implemented with schools during February–June 2025. The intervention was supported by the EduCITY DTLE (available at https://educity.web.ua.pt/project.php, accessed on 12 September 2025) and its mobile app (version 1.3), available at https://educity.web.ua.pt/app.php (accessed on 12 September 2025). The EduCITY project was developed by a multidisciplinary research team and research units from the University of Aveiro. This context frames the participant groups and procedures described in the following subsections.

3.3. Participants

The broader study comprised four cohorts: in-service teachers who participated in the validation of the MARG (T1-VAL, N = 30); in-service teachers who conducted a structured curricular review (T1-R, N = 3); accompanying teachers who supervised fieldwork and completed structured observations (T2-OBS, N = 24); and students who completed GCQuest at three time points, namely baseline S1-PRE (N = 221), immediate post-intervention in situ S2-POST (N = 439) and follow-up S3-FU (N = 434).

In this work, the analysis focuses on student-generated evidence (questionnaires and gameplay logs) and on the accompanying teachers’ observations, while the T1-VAL and T1-R activities are treated as contextual inputs that informed the design of the intervention.

3.3.1. Teachers’ Validation (T1-VAL) and Curricular Review (T1-R)

Thirty in-service teachers (seventeen female, thirteen male) from central Portugal voluntarily participated in a validation workshop (T1-VAL). In addition, three in-service teachers with expertise in History, Natural Sciences, and Visual Arts/Citizenship carried out a structured curricular review (T1-R). These activities ensured the pedagogical coherence, curricular alignment, and refinement of the Art Nouveau Path prior to implementation [49,50] and correspond to one of the cycles in the broader DBR approach.

3.3.2. Students: Implementation Cohort

A total of 439 lower and upper secondary school students, aged 13–18, participated in the field study during regular school hours. The students were recruited through the Municipal Educational Action Program of Aveiro 2024–2025 (PAEMA) [83]. The sample comprised 19 classes spanning six grade levels: seventh (N = 19), eighth (N = 135), ninth (N = 156), tenth (N = 37), eleventh (N = 20), and twelfth (N = 72). Students played in groups of two to four. To ensure comparable gameplay conditions and comply with the General Data Protection Regulation (GDPR), each group used an EduCITY Android smartphone (version 1.3), which also supported collaborative work and shared interpretation. The MARG was implemented across 18 field sessions, forming 118 groups whose data were automatically and anonymously recorded in the app logs and later downloaded via the university network.

GCQuest was administered at three points in time: before the intervention (S1-PRE, N = 221), immediately after gameplay in situ (S2-POST, N = 439), and six to eight weeks later (S3-FU, N = 434). Because participation depended on school timetables and consent procedures, the three waves constitute partially overlapping cohorts rather than a strictly longitudinal sample. In this study, they are therefore treated as a repeated cross-sectional design and interpreted at that level.

As such, this student cohort should be understood as a municipal convenience sample within the PAEMA [83] rather than as a statistically representative sample of all students in the region.

3.3.3. Accompanying Teachers’ Field Observations (T2-OBS)

During the 18 on-site sessions, the 24 accompanying teachers were co-responsible for logistics and safety, as well as for completing the T2-OBS questionnaire. Their ecological observations provide complementary evidence on how students engaged with the MARG, the POIs, and the sustainability-related tasks in real conditions.

3.4. Instruments and Data Collection

This project used a mixed-methods strategy combining self-report questionnaires, structured field observations [84], automated gameplay logs [85], and curated digital heritage assets, as summarized in

Table 1. Overview of instruments and datasets used in this article.

Instrument (Code)/Dataset	Participants/ Scope	Focus	Data Type	Aim/ Use
S1-PRE	221 students (baseline)	Sustainability conceptions; heritage interest; AR readiness	25 Likert items + open-ended	Baseline diagnostic for context; qualitative prompts inform coding scheme
S2-POST	439 students (immediately post)	Engagement; narrative coherence; interest in Art Nouveau	25 Likert items + open-ended	Primary student self-report during implementation
S3-FU	434 students (6–8 weeks later)	Retention; transfer; sustained reflections	25 Likert items + open-ended	Follow-up perspective on persistence of effects
T2-OBS	24 accompanying teachers	On-site collaboration; AR use; attention to detail; wayfinding	Likert + checklists + open-ended	Ecological evidence under authentic conditions
Gameplay logs	Student groups (N = 118 groups)	Completion; correctness; distractor choice; path order; duration	App-generated data (group-level, anonymized)	Patterns of pacing and item-level performance; response and trajectory reconstruction
Digital assets and descriptors	Eight POIs (content layer, AR markers)	Archival images; AR overlays; narrative prompts; semantic tags	Multimedia and metadata	Cultural data outputs curated for reuse, with no personal data, and based on community and FAIR data sources

.

These instruments support methodological triangulation consistent with DBR process orientation, maintaining the analytic focus on in situ evidence generated during implementation.

3.4.1. Student Questionnaires (S1-PRE, S2-POST, and S3-FU)

The student questionnaires combined open-ended, dichotomous, and Likert-scale items. All administrations were in Portuguese, using the official EU GreenComp translation to ensure linguistic and conceptual fidelity. The open-ended items invited students to describe their views on heritage settings, attention to architectural details, responsibilities towards the urban environment, and their own understanding of sustainability, among other topics. The Likert section was adapted from the GreenComp framework [35] and had been validated in previous work through structural equation modeling (SEM) [85]. Although the three versions share a common structure, they were designed to function as independent yet conceptually linked instruments.

They were applied across three distinctive moments:

• S1-PRE (baseline), which was administered prior to gameplay and aimed to collect perceptions of sustainability, cultural heritage, and AR readiness;
• S2-POST, which was administered immediately after gameplay and aimed to assess engagement, perceived narrative coherence, interest in Art Nouveau, and the development of sustainability competencies;
• S3-FU, which was administered six to eight weeks after the MARG implementation and aimed to assess the value of the MARG in transferring knowledge to daily practices, concepts, and reflections on heritage preservation and the retention of sustainability competencies.

Each administration lasted approximately 20 min, and students completed the questionnaire individually. Coding procedures followed a protocol established by the authors and an EduCITY researcher, as detailed in previous works [49,50].

3.4.2. Teacher Observations (T2-OBS)

During the 18 field sessions, the 24 accompanying teachers completed the T2-OBS. This structured instrument combines Likert statements, checklists, and open-ended questions. The items targeted the following: (i) on-site collaboration and group dynamics, (ii) attentiveness to architectural details and heritage cues, (iii) interaction with AR content in authentic conditions, and (iv) navigation and pacing across the micro-itinerary. Emphasizing ecological validity, the instrument observes how learning processes unfold in situ.

3.4.3. Automated Gameplay Logs

The EduCITY application, in its version 1.3, produced anonymized group-level logs for each session, including start/end timestamps, completion status for the 36 quiz items (P1.1–P8.2), item-level correctness, and for incorrect responses, the selected distractor. In accordance with data minimization principles and GDPR compliance regulations, the logs do not contain any personal identifiers and were stored at the group (device) level exclusively. The temporal resolution facilitates session-level pacing and item-level outcomes; however, it does not permit the precise estimation of dwell time per POI.

3.5. Data Analysis

The data analysis integrates qualitative themes from open-ended answers and quantitative indicators, for example, item difficulty and completion. Integration followed a convergence-coding approach (convergence, complementarity, divergence) to produce meta-inferences about how AR-mediated, place-based tasks interacted with attention to façade details and sustainability sense-making. This integrative approach aligns with the established guidelines for integrating diverse research methods during the analysis and interpretation phases of research [86]. The S1-PRE, S2-POST, and S3-FU administrations are repeated cross-sectional, without individual pairing between waves.

3.5.1. Questionnaire Analysis

A mixed-methods analytical strategy was employed, organized in three layers: quantitative summaries of questionnaire items, reflexive thematic analysis of open responses and field notes, and trajectory-based analytics on gameplay logs, followed by an integrative synthesis.

The administration of the GCQuest tailored versions (S1-PRE, S2-POST, S3-FU) and dichotomous items was summarized using descriptive statistics, including means, standard deviations, and absolute and relative frequencies. The indicators reported include completion rates, item-level distributions, and aggregate scores by administration. In line with the exploratory nature of this case study, the non-probabilistic recruitment of classes through a local program, and the repeated cross-sectional design without individual pairing across waves, questionnaire data were intentionally treated as descriptive evidence for this specific implementation rather than as a basis for population-level inference. Consequently, no formal inferential tests or p values are reported for GCQuest comparisons across S1-PRE, S2-POST, and S3-FU, including for Likert-scale items. Applying simple hypothesis tests (for example, chi square or repeated-measures models) would have implied assumptions regarding random sampling, independence, and a panel structure that are not met in this dataset and could have overstated the generalizability of the results. Instead, descriptive patterns are used to characterize tendencies within the participating cohorts and to inform, and be informed by, the qualitative thematic analysis.

To report more synthetic indicators in Section 4, some of the dichotomous items were grouped into four analytic labels that mirror the key constructs of interest. These labels capture (i) interest in learning about sustainability through Art Nouveau heritage; (ii) interest in Aveiro’s Art Nouveau heritage; (iii) civic responsibility for local heritage; and (iv) the city understood as a shared resource. Each label corresponds to a specific set of items across S1-PRE, S2-POST, and S3-FU (for example, S1-PRE A.3.2 and S2-POST A.2.2 for interest in learning about sustainability through Art Nouveau, and S1-PRE A.3.3 and S2-POST A.2.3 for interest in Aveiro’s Art Nouveau heritage), as detailed in Appendix B (

Table A2. Mapping of analytic labels and questionnaire items and codes.

Analytic Label	Instrument	Item	Wording
Baseline counterpart of interest item	S1-PRE	A.3.2	“Do you think it is interesting to learn about sustainability from Art Nouveau heritage?”
Baseline counterpart of curiosity item	S1-PRE	A.3.3	“Would you like to learn more about Aveiro’s Art Nouveau heritage?”
Interest in learning sustainability through Art Nouveau	S2-POST	A.2.2	“Do you find it interesting to learn about sustainability from Art Nouveau heritage?”
Would like to know more about Aveiro’s Art Nouveau	S2-POST	A.2.3	“Would you like to know more about Aveiro’s Art Nouveau heritage?”
Civic responsibility for local heritage	S3-FU	A.2.4	“Did the activity lead you to reflect on the importance of preserving local cultural heritage?”
City as a shared resource	S3-FU	A.2.5	“Has the activity increased your curiosity or attention to the street, neighborhood, or locality where you live?”

).

The open-ended responses were then subjected to a reflexive thematic analysis, employing a codebook that integrated both deductive and inductive approaches and was aligned with the GreenComp framework [35] and based on previous works [49,50]. The authors and a second EduCITY project researcher independently coded the data using memos and an audit path. Disagreements were resolved through discussion. In accordance with the principles of reflexive thematic analysis [48], inter-coder coefficients were not computed.

3.5.2. Gameplay Log Analysis

Gameplay logs were analyzed by processing the app-generated, group-level records. The logs allowed us to analyze the following variables: (i) session duration distributions; (ii) completion and correctness by item/POI; (iii) distractor profiles for incorrect responses; and (iv) use of AR contents. When reporting item-level performance in this article, accuracy was defined as the proportion of collaborative groups selecting the correct option for each item. Considering the path dynamics, the MARG design incorporated contextual meaning, including the specific site, the designated task, and the prevailing motif [29]. This approach enabled interpretation that transcended the confines of raw coordinates, based on the AR markers. For instance, it facilitated the discernment of narrative cues that precede accurate responses or the identification of detours that cluster around façades or monuments [86].

In addition to item-level accuracy, a group-level AR score was constructed to summarize performance in the 11 AR-enhanced items distributed along the route. This score was computed as the sum of points obtained in those items (5 points per correct answer, 0 otherwise; theoretical range 0–55) and was used to distinguish low-exposure groups (AR score < 25) from higher-exposure groups (AR score ≥ 25) in the Ordinary Least Squares (OLS) contrasts reported in Section 4.3.2. The rationale for this threshold and additional robustness checks are presented in Appendix C.

3.6. Ethical Considerations

This study was conducted in accordance with the ethical protocols established by the University of Aveiro and in compliance with the GDPR. Participation in this study was voluntary, and all participants provided informed consent. Informed consent was obtained from all participating teachers; for students, parental or legal-guardian consent was additionally secured. No personally identifiable information was collected. The datasets are securely stored, and gameplay logs were recorded at the group/device level. In accordance with the principles of data minimization, this study did not collect gender or socioeconomic data. While this restriction limits the possibility of conducting subgroup analyses, it does support the principles of proportionality and ecological validity in an in situ educational intervention [87,88].

4. Findings

This section reports the findings in direct relation to the three research questions introduced in Section 1. It presents evidence from GCQuest questionnaires (S1-PRE, S2-POST, S3-FU), gameplay logs from 118 collaborative groups, and T2-OBS teachers’ questionnaires.

4.1. Cross-Source Overview and Analytic Rationale

This subsection provides a cross-source overview that frames the patterns relevant to all three research questions (RQ1, RQ2, and RQ3). It presents the cross-source results that establish the repeated cross-sectional pattern of change across S1-PRE, S2-POST, and S3-FU, while subsequent subsections contain more detailed thematic and specific analyses.

Prior to the MARG’s implementation, validations were conducted with 33 in-service teachers (T1-VAL and T1-R). These validations enabled us to assess and establish both curricular alignment and interpretive coherence for the Art Nouveau Path MARG across multiple curricular areas and to perform a specific History, Natural Sciences, Visual Arts, and Citizenship assessment. This process analysis outputs are exclusively used as inputs to model and implement dynamics, considering that comprehensive results have been previously presented [49,50].

The current work analysis focuses on student-generated evidence, in-app gameplay logs, and in-field teachers’ observations collected under the implementation. A total of 439 students, organized in 118 collaborative student groups, produced 4248 group-item responses across 18 field sessions, accompanied by 24 teachers who completed structured observation forms (T2-OBS).

The MARG’s implementation encompassed a total of 36 items, including 11 AR-based tasks and 25 non-AR tasks. The latter category included seven knowledge-check items, twelve multimedia prompts (static archival/photo/text without AR overlay), and six local-analysis items. The complete Art Nouveau Path MARG is available at https://doi.org/10.5281/zenodo.16981235.

Baseline conceptions positioned heritage preservation as marginal within sustainability. At the baseline (S1-PRE, N = 221), 28.96% of students (n = 64) explicitly associated sustainability with cultural heritage, while most answers were framed in exclusively environmental terms; 14.03% (n = 31) did not provide a clear response. These results frame S1-PRE as a heritage-oriented baseline rather than an outcome instrument.

Immediately following gameplay (S2-POST, N = 439), references to heritage preservation more than doubled to 61.05% (n = 268), a plus 32.09 percentage-point increase over baseline (Figure 8). In the subsequent assessment (S3-FU, N = 434), mentions declined but remained above baseline at 47.93% (n = 208), indicating an 18.97 percentage-point increase.

Dichotomous indicators show growth across waves. At baseline, related proxies aligned with items of interest increased significantly in S2-POST, and in S3-FU, the construct was explicitly measured by A.2.4 (“civic responsibility for local heritage”, 50.92%) and A.2.5 (“city as a shared resource”, 53.92%). Figure 9 summarizes these patterns; the full mapping of labels and items is in Appendix B.

Open-ended responses corroborated and nuanced these shifts. Thematic analysis confirmed a reweighting of sustainability discourses toward heritage preservation. Three non-mutually exclusive categories were salient and evolved consistently across phases: (i) preservation and care of the built structure, 28.96% (n = 64) at baseline, 61.05% (n = 268) after the game, and 47.93% (n = 208) at follow-up; (ii) heritage within sustainable urban development, 22.17% (n = 49), 43.96% (n = 193), and 35.94% (n = 156); and (iii) environmental frameworks, 57.92% (n = 128) at baseline, 30.98% (n = 136) after the game, and 41.94% (n = 182) at follow-up (see Figure 10). This coexistence suggests a pattern of integration rather than substitution. Illustrative examples include “Sustainability also means not letting the old façades fall apart; they are part of our city.” [S2-POST]; “Taking care of buildings and not just nature, both matter for the city to last.” [S2-POST]; and “We should maintain these houses; it is sustainable because it preserves culture and avoids waste.” [S3-FU].

Teacher observations corroborated the same trend. In 58.33% of the observation forms (14 of 24 T2-OBS), teachers identified spontaneous discourse on preservation, often prompted by overlays juxtaposing archival photographs with current façades. Micro-dialogues such as “we should protect this” and “it would be unfortunate if this broke” were frequently recorded.

Item-level performance illustrated the same heterogeneity. Using the gameplay logs from 118 groups, the weakest preservation-framed item was POI5.4 at 58.47% correct (69/118). In contrast, the set POI3.5, POI4.3, POI5.5, and POI5.6 yielded 93.22% (n = 110/N = 118), 92.37% (n = 109/N = 118), 89.83% (n = 106/N = 118), and 76.27% (n = 90/N = 118), respectively, with a cluster mean of 87.92% and SD of 7.90, informing the prioritization of on-site interpretive prompts where students historically confuse restoration with repainting, as presented in Figure 11.

These descriptive results suggest patterns of short- to medium-term change in this case study of the observed changes, without implying formal statistical significance. The salience of preservation increased from 28.96% at baseline to 61.05% immediately after the intervention and then stabilized at 47.93% in the follow-up. These gains encompassed a triad of objectives: preservation, civic responsibility, and the conceptualization of the city as a shared resource. Notwithstanding the deterioration of post-test gains over time, preservation remained above the baseline, which is consistent with short- to medium-term internalization in this cohort.

4.2. Art Nouveau as a Cultural Resource and Urban Identity Asset

This subsection presents findings that concern primarily RQ1 (“Can heritage serve as an effective context for ESD?”), by examining how students and teachers framed Art Nouveau façades as cultural resources and urban identity assets.

The Art Nouveau Path posits that Art Nouveau façades and monuments did not emerge as inert monuments, but rather as cultural anchors that served to mediate identity, memory, and belonging [14,87,88,89]. A thorough examination of the extant data, encompassing student questionnaires, open-ended responses, gameplay logs, and teacher observations, unveils four recurrent categories that manifest with a high frequency. The categories encompass the recognition of architectural elements, the establishment of a connection to civic identity, the cultivation of affective pride and belonging, and the conceptualization of heritage as a living resource.

4.2.1. Thematic Categories from S1-PRE, S2-POST, and S3-FU

Open-ended responses in S1-PRE resulted in four interconnected categories that remained visible on S2-POST and S3-FU. The category labels and prevalence are presented in

Table 2. Prevalence of thematic categories on students’ conceptions of heritage across questionnaires.

Category	Description	S1-PRE (N = 221)	S2-POST (N = 439)	S3-FU (N = 434)	Illustrative Example
C1. Recognition of architectural details	Identification of façades and decorative motifs	31.22% (n = 69)	71.98% (n = 316)	61.06% (n = 265)	“I never noticed the balconies before; now I look for them when I walk in the city.” [S3-FU]
C2. Connection to Aveiro and civic identity	Buildings framed as markers of local distinctiveness and memory	18.10% (n = 40)	43.96% (n = 193)	38.02% (n = 165)	“These houses are Aveiro; if they disappear, the city changes.” [S2-POST]
C3. Affective pride and belonging	Emotional investment in heritage as part of self and community identity	12.22% (n = 27)	33.03% (n = 145)	29.03% (n = 126)	“I want to show my parents what I learned; it makes me proud.” [S3-FU]
C4. Heritage as a living resource	Heritage understood as a civic asset, not a static monument	8.14% (n = 18)	22.10% (n = 97)	19.12% (n = 83)	“Heritage is not only in museums; it is here in our streets.” [S2-POST]

.

These categories reveal a layered shift: from noticing architectural motifs (C1) to embedding them within civic identity (C2), affective pride (C3), and the recognition of heritage as a dynamic resource for sustainable cities (C4).

4.2.2. Quantitative Patterns from S1-PRE, S2-POST, and S3-FU

At the baseline (S1-PRE, N = 221), only 31.22% (n = 69) of students spontaneously mentioned façades or decorative details when reflecting on sustainability and the city, with most answers centering on natural resources. Following the conclusion of the gameplay phase (S2-POST, N = 439), this proportion increased to 71.98% (n = 316), with frequent mentions of architectural details (41.00%, n = 180), wrought-iron balconies (35.99%, n = 158), and tiles (28.02%, n = 123). In the subsequent study (S3-FU, N = 434), the qualitative coding of open-ended responses revealed that 61.06% (n = 265) explicitly mentioned architectural details (Noticing category), indicating substantial recall of specific features. In contrast, in the closed item A.2.1 [Do you still remember any details, buildings or areas of the city you visited during the game?], 81.94% (n = 356; N = 434) of the participants reported remembering details in general, indicative of broader declarative recall. These complementary data signify both generic and concrete forms of memory retention, as presented in

Table 3. Student references to architectural features across S1-PRE (N = 221), S2-POST (N = 439), and S3-FU (N = 434), with triangulation from gameplay logs and teacher observations.

Dataset/Source	N (Total)	Coded Units (n)	Outcome Metric	Descriptors
S1-PRE	221	69 coded responses	31.22% mentioning at least one architectural feature	General façades; few decorative details
S2-POST	439	316 coded responses	71.98% mentioning architectural details	Various details (n = 13); balconies (n = 158); tiles (n = 123)
S3-FU	434	265 coded responses	61.06% mentioning at least one recalled feature	Recall of at least one feature
Gameplay logs	4248 group-item responses	118 collaborative groups	83.00% mean item accuracy (AR-related subset)	Lower-accuracy motifs linked to distractors
Teachers’ observations (T2-OBS)	24 teachers	19 forms rated high/very high	79.17% observed increased attention	Attention increased when AR overlays highlighted details

.

The S3-FU results confirm both generic and concrete recall of architectural features above baseline, consistent with gameplay accuracy and teachers’ observations.

4.2.3. Triangulation with T2-OBS and Gameplay Logs Regarding Art Nouveau POIs

The reliability of these findings was further substantiated by teacher observations, which corroborated the observed patterns. In 62.50% of T2-OBS forms (15 of 24), teachers reported overhearing students framing façades as “ours” or “belonging to Aveiro.” A substantial proportion of the teachers (54.17%, 13 out of 24) reported observing affective reactions, including enthusiastic pointing, photographing, or verbalized expressions of pride.

The findings were reinforced by gameplay logs (N = 118 collaborative groups; 4248 group-item responses), which also revealed tensions. At POI4 (Old Agricultural Cooperative), the term “aesthetic repainting” was conflated with “authentic tile preservation,” resulting in an accuracy rate of 69.49% (82 of 118 correct). This ambiguity exemplifies how students actively negotiated the boundary between façades as surface appearance and façades as cultural heritage.

The affective and cognitive traces documented across S1-PRE, S2-POST, and S3-FU responses, in conjunction with gameplay logs and T2-OBS field notes, substantiate the hypothesis that Aveiro’s Art Nouveau façades, when mobilized through AR mediation, functioned as living resources rather than static monuments.

4.3. Extended Reality for Heritage: Augmented Reality as the Primary Approach

This subsection addresses RQ2 (“How do multimodality and AR affect engagement and learning outcomes?”), focusing on the mediating role of AR in students’ interactions with façades and monuments.

The incorporation of AR within the Art Nouveau Path has emerged as a pivotal catalyst, superimposing interpretive content directly onto the urban landscape and profoundly influencing the way students interacted with façades. The extant evidence, derived from post- and follow-up questionnaires, gameplay logs, and teacher observations, demonstrates three interrelated dynamics: (i) a heightened level of attention is allocated to architectural elements. The integration of AR overlays prompted students to observe elements of the built environment that would have otherwise gone unnoticed, including architectural details, wrought-iron balconies, and tiles; (ii) the implementation of a multimodal and game-based design approach in the educational environment has been demonstrated to promote two key elements: motivated exploration and sustained engagement. In this case, students were encouraged to move attentively through the path, thereby prolonging their focus and generating enthusiasm; and (iii) the tensions between productivity and authenticity, as well as the challenges posed by distraction, were particularly pronounced. While augmented reality (AR) technology enhanced interpretive depth, it also introduced moments of ambiguity, particularly when students confused surface renovation with heritage preservation.

The significance of AR in its value to function not merely as a technological augmentation but rather as a mediational layer with the potential to profoundly reshape the perceptual and affective experience of heritage is underscored by these dynamics. In this regard, façades emerge as living anchors of sustainability discourse.

4.3.1. Thematic Categories from Open-Ended Answers (S1-PRE, S2-POST, S3-FU)

The reflexive thematic analysis of the open-ended sections of the GreenComp-based questionnaires was conducted, which resulted in the identification of three recurrent categories that articulate how students perceived the role of AR in the Art Nouveau Path. Percentages indicate the proportion of students whose responses were coded in each category at baseline (S1-PRE), immediately after gameplay (S2-POST), and at follow-up (S3-FU), as presented in

Table 4. Prevalence of thematic categories on students’ perceptions of AR across S1-PRE, S2-POST, and S3-FU.

Category	Description	S1-PRE (N = 221)	S2-POST (N = 439)	S3-FU (N = 434)	Illustrative Example
C1. AR as amplifier of observation	AR overlays highlighted unnoticed architectural details (façades, balconies, tiles)	22.17% (n = 49)	46.92% (n = 206)	38.94% (n = 169)	“With the AR camera, I noticed the flowers in the windows; before I would just pass by.” [S2-POST]
C2. AR as motivator of curiosity	AR perceived as fun, novel, and motivating, increasing willingness to explore	18.10% (n = 40)	41.00% (n = 180)	32.95% (n = 143)	“It was fun, like a game, not a class, and it made me want to walk more.” [S2-POST]
C3. AR as double-edged experience	AR produced moments of distraction or ambiguity, shifting focus away from the real object	9.05% (n = 20)	20.96% (n = 92)	17.97% (n = 78)	“Sometimes I only looked at the phone and forgot the real house.” [S3-FU]

.

These categories illustrate AR’s role as both an amplifier and a disruptor: it heightened observation and curiosity while also introducing tensions between digital mediation and the embodied experience of place.

4.3.2. AR as a Key Element: Quantitative Patterns from S2-POST, S3-FU, and Gameplay Logs

In the S2-POST (N = 439), 67.88% (n = 298) of students explicitly cited AR as the element that “helped them notice things better” or “helped them see details.” At the S3-FU (N = 434), 52.07% (n = 226) of participants restated that AR had a transformative effect on their perception of buildings. As presented in

Table 5. Student references to AR as a key element shaping perception of façades (S2-POST and S3-FU).

Dataset	N (Total)	n (Coded)	% Mentioning AR as Key Element	Illustrative Example
S2-POST	439	298	67.88%	“Helped me notice things better”; “Helped me see details.”
S3-FU	434	226	52.07%	“Changed how I look at buildings.”

, these references underscore the significance of AR in influencing the perception of façades and monuments.

Gameplay records corroborated these self-perceptions. AR items had a mean group-level accuracy of 81.00% versus 73.00% for non-AR items, based on gameplay logs. The ranges are reported: 81.00% [95% CI 78.75–83.01%; N items = 11; N groups = 118] and 73.00% [95% CI 71.39–74.59%; N items = 25; N groups = 118]. At the group level, the average exploration time was 42.77 min (SD 6.08) versus 32.60 min (SD 6.77).

Table 6. Gameplay metrics by AR versus non-AR items and group exploration time.

Gameplay Metric	AR-Based Items/Groups	Non-AR Items/Groups	Difference
Mean group-level accuracy (%)	81.00	73.00	+8.00 pp
Mean exploration time (minutes, SD)	42.77 (SD = 6.08)	32.60 (SD = 6.77)	+10.17 min

presents a comparative analysis of gameplay metrics.

The AR score was modeled as continuous in OLS with fixed session effects and robust errors with small-sample correction. The results were qualitatively similar to the bivariate contrast, as can be analyzed in Appendix C.

4.3.3. Triangulation with T2-OBS and Gameplay Logs Regarding AR

Teachers’ observations further substantiated AR’s catalytic function. In 83.33% of T2-OBS forms (20 of 24), teachers reported that overlays increased students’ attention to façades, often describing learners pointing, comparing digital and material details, and verbally negotiating meanings. Concurrently, 37.50% (9 of 24) of respondents expressed concerns regarding excessive reliance on screens, providing comments such as “some groups focused too much on the phone instead of the building.”

This duality is corroborated by the gameplay records. AR tasks demonstrated greater accuracy, 81.00% [95% CI 78.75–83.01%; N items = 11; N groups = 118], compared to 73.00% [95% CI 71.39–74.59%; N items = 25; N groups = 118] in non-AR items. The error patterns that emerged from this analysis indicated a notable instance of conceptual confusion, characterized by the erroneous attribution of decorative repainting as original Art Nouveau motifs. These indicate that AR overlays enhance both attention and interpretive depth, but they also introduce risks of screen-centric engagement. The convergence of teacher field notes and gameplay logs underscores the ambivalent role of AR in education. It may be used as an interpretive amplifier that requires meticulous scaffolding to harmonize digital mediation with embodied observation.

4.4. Immediate Post-Game Perceptions from S2-POST

This subsection deepens the response to RQ1 and RQ2 by analyzing immediate post-game perceptions of learning through Art Nouveau heritage and augmented reality.

The triangulation of data from teachers’ observations (T2-OBS) and gameplay logs serves to corroborate the categories derived from the student data. This instrument was designed to capture students’ perceptions immediately following gameplay, integrating Yes/No indicators, open-ended reflections, and Likert-scale items.

The data indicates a significant shift in the way façades and monuments are perceived, interpreted, and incorporated into broader sustainability discourses. The students exhibited an elevated level of proficiency in recognizing architectural elements and demonstrated an augmented capacity to articulate the interrelationships between preservation, civic responsibility, and the concept of the city as a shared resource.

4.4.1. Dichotomous Items on S2-POST

The binary responses in the S2-POST (N = 439) dataset indicate a high level of approval for the learning approach and a selective curiosity regarding Aveiro’s Art Nouveau heritage. Two key items illustrate these patterns. Most of the respondents expressed interest in the prospect of learning sustainability through Art Nouveau. However, the level of interest in exploring Aveiro’s Art Nouveau specifically was, in comparison, more moderate. As presented in

Table 7. Dichotomous items from S2-POST: Interest in learning sustainability through Art Nouveau and interest in Aveiro’s Art Nouveau.

Item Code	Examples	Yes (n, %)	No (n, %)	Total (N)
A.2.2	“It is interesting to learn about sustainability through Art Nouveau”	432 (98.45%)	7 (1.55%)	439
A.2.3	“I would like to know more about Aveiro’s Art Nouveau”	415 (94.53%)	24 (5.47%)	439

, the responses from students are summarized.

The results indicate, in comparison, a distinction. Almost all students’ population demonstrated a recognition of the value of integrating sustainability with Art Nouveau, as evidenced by the findings of this study. The results indicated that 98.45% of the students (n = 432) acknowledged the potential of façades as effective educational entry points. Less, but also the large majority, 94.53% (n = 415), of respondents expressed a desire to continue exploring Aveiro’s Art Nouveau specifically. These results demonstrate an overall comprehension of the broader conceptual link between heritage and sustainability, and their engagement with the local case study.

4.4.2. Triangulation Between T2-OBS, S2-POST, and Gameplay Logs

The observations made by the teachers supported the thematic categories that had been previously identified in S2-POST. In 62.50% of the T2-OBS questionnaires (15 of 24), teachers reported dialogues of students overheard framing façades as “ours” or “belonging to Aveiro.” A substantial proportion of teachers, specifically 54.17% (13 out of 24), reported observing affective reactions, including enthusiastic pointing, photographing details, and verbalizing pride.

Gameplay logs (N = 118 collaborative groups; 4248 group-item responses) demonstrated consistently high engagement with detail-recognition tasks (mean group-level accuracy = 83.00% based on gameplay logs). However, deeper data analysis and cross-checking also unveiled conceptual ambiguities at POI4 (Old Agricultural Cooperative), where respondents exhibited confusion between “aesthetic repainting” and “authentic tile preservation,” resulting in 69.49% group-level accuracy (82 of 118 collaborative groups selected the correct option).

4.5. Follow-Up Retention and Transfer (S3-FU)

This subsection responds directly to RQ3 (“To what extent do students retain and transfer heritage-related sustainability competencies after gameplay?”) by examining follow-up evidence on retention and transfer.

Post-game evidence (from S2-POST) indicates that AR-mediated activities activated façades as cultural interfaces rather than passive backdrops, thereby reframing them as markers of civic identity and everyday cultural resources. A comprehensive review of extant data collection tools, namely, the qualitative data, observational studies, and log files, reveals that this place-anchored mediation has been embedded within the discourse on sustainability.

The coding of responses to A.2.1.1 (“Do you remember a detail, building, or area from the game?”) in the follow-up dataset (S3-FU, n = 434) yielded three main categories.

Table 8. Repeated cross-sectional comparison of A.2.1. categories (S2-POST vs. S3-FU).

Category	S2-POST (N = 439)	S3-FU (N = 434)	Illustrative Example
Noticing heritage details	76.08% (n = 334)	56.68% (n = 246)	“The building details stayed in my memory; I still look for them in the city.” [S3-FU]
Experiential learning and transfer	15.72% (n = 69)	31.80% (n = 138)	“Now, when I walk with friends, I explain what Art Nouveau is and why it matters.” [S3-FU]
Unclear or other	8.20% (n = 36)	11.52% (n = 50)	“I do not remember exactly, but I know it was interesting.” [S3-FU]

provides a comparison of the prevalence with the immediate post-game phase (S2-POST), thereby offering a repeated cross-sectional perspective on the evolution of heritage engagement over time.

A repeated cross-sectional analysis yielded observations indicating an attenuation in detail recognition, concomitant with an augmentation in experiential transfer. This finding may signal a transition from a focus on detail recognition to the cultivation of civic and everyday sensibilities [90,91].

Regarding these results, the following presents students’ illustrative examples regarding this repeated cross-sectional comparison: “We should maintain these houses; it is sustainable because it preserves culture and avoids waste.” [at S3-FU], linking preservation practices with both cultural continuity and resource efficiency; “I want to show my parents what I learned; it makes me proud.” [at S3-FU], reflecting intergenerational transfer and affective pride in local heritage; and “Since the game, I pay more attention to the façades when I walk in my neighborhood.” [S3-FU], signaling behavioral change, extending attentiveness beyond the game context into daily life.

4.6. Teachers’ Observations and Micro-Dialogue Report

This subsection provides complementary evidence for RQ1 and RQ2, drawing on teachers’ in situ observations and micro-dialogues about preservation, engagement, and the use of AR.

Teachers frequently recorded student micro-dialogues such as “we should protect this” and “it would be a pity if this broke”, particularly at façades where AR overlays juxtaposed archival photographs with contemporary views. Logs identified ambiguity at POI4, where distractors blurred the distinction between repainting and authentic tile preservation, yielding 69.49% accuracy (82 of 118). At POI6 (Art Nouveau Museum), an element of hesitation regarding the definition of ‘preservation’ was also observed; analysis of gameplay logs revealed that the item with the lowest accuracy in relation to preservation, POI5.4, attained 58.47% group-level accuracy, while the preservation set exhibited an average group-level accuracy of 82.05%, according to the gameplay logs. These patterns are in alignment with the accounts provided by the teachers, thus demonstrating the way students navigated the boundary between surface appearance and cultural heritage.

4.7. Triangulation Across Questionnaires, Logs, and Teacher Observations

This subsection integrates these strands of evidence to synthesize how the findings jointly inform RQ1, RQ2, and RQ3.

The triangulation of questionnaires, gameplay logs, and teachers’ observations indicates that the Art Nouveau Path activated façades as cultural interfaces rather than static backdrops [46]. Students expanded sustainability to include preservation, noticed and remembered architectural details, and reported sharing and revisiting beyond the activity. AR and multimodality strengthened attention and accuracy [92], while also producing authentic debates about what counts as preservation [32]. These patterns are consistent with a desirable difficulty profile, although explicit “civic responsibility” is measured only in S3-FU (A.2.4), while in S1-PRE and S2-POST, analyses used interest proxies (see Appendix B).

5. Discussion

This study explored whether the Art Nouveau Path, a location-based MARG, could promote urban heritage while cultivating sustainability competencies. Three research questions guided the analysis: RQ1, whether heritage can serve as an effective context for Education for Sustainable Development; RQ2, how multimodality and AR affect engagement and learning outcomes; and RQ3, to what extent students retain and transfer heritage-related sustainability competences after gameplay.

5.1. Heritage as a Context for ESD (RQ1)

The results of the study suggest that, in this case study, built heritage may function as a meaningful context for sustainability learning. As detailed in Section 4.1, the proportion of students explicitly linking sustainability to heritage increased markedly from baseline to the immediate post-test and remained higher than baseline at follow-up. In S3-FU, “civic responsibility for local heritage” was measured explicitly only in A.2.4, whereas the baseline and post-test results draw on interest proxies rather than on the same construct; see mapping in Appendix B. Teachers’ reports in T2-OBS corroborate these tendencies, with preservation dialogues documented in over half of the responses. These shifts are consistent with people-centered heritage agendas that conceptualize heritage as a living social resource [4,5,51] and with values-based management perspectives [8,9,56]. In the present work, light-touch, in situ tasks appear to have operationalized these principles at the street level for the participating classes, embedding preservation within students’ sustainability conceptions in a tentative but observable way. Although such alignment was anticipated, it has rarely been documented with repeated, cross-sectional descriptive evidence in ESD-with-heritage studies [93,94].

5.2. The Role of Multimodality and AR in Heritage Learning (RQ2)

Multimodal approaches, especially the combination of archival materials and AR overlays, appear to have promoted attentive observation and sustained engagement. As shown in Section 4.3, AR-supported items were consistently more accurate than non-AR items and were associated with longer exploration times at the group level. These findings align with reviews that position AR as an interpretive amplifier when narratively coherent with place [23,31,32] and with narrative cartography arguments that situated media bind movement, noticing, and meaning-making across stops [25].

At the same time, well-documented tensions emerged. Several T2-OBS notes and S3-FU reflections point to episodes of distraction or overreliance on the screen. Authenticity negotiations were particularly visible at POI4 (Old Agricultural Cooperative) and POI5.4, where some students conflated aesthetic repainting with authentic tile preservation. Rather than hindering learning, these tensions likely constituted productive difficulty, prompting students to contrast current surfaces with archival evidence, the very type of comparison that AR enables in context [29]. In this sense, AR operated as both an interpretive lens and a preservation representation, documenting and re-contextualizing façade features without the need for extensive digitization pipelines [2].

5.3. Retention, Transfer, and Long-Term Disposition (RQ3)

Follow-up evidence indicates a shift from immediate, detailed observations to everyday application and civic sense-making. The S3-FU reports that students shared insights with family and peers and paid more attention to façades in daily life. Meanwhile, the T2-OBS documents revisits with parents or peers in 54.17% of narratives. This trajectory aligns with the goals of competence-oriented ESD and the emphasis on attentiveness, belonging, and stewardship in GreenComp [39]. Items that were challenging in situ (like at POI5.4) resurfaced in follow-up interpretations, suggesting that momentary interpretive struggles helped consolidate durable memory traces. This effect appears to be potentiated by AR-supported comparison [23,32].

5.4. Integrative Implications

Synthesizing evidence across the GCQuests (S1-PRE, S2-POST, and S3-FU), gameplay logs, and T2-OBS distils four cross-cutting implications linking heritage-based pedagogy, AR-mediated multimodality, and smart-heritage governance:

• Heritage as a driver of ESD: by embedding GreenComp-aligned tasks within authentic façades and micro-narratives, the intervention locally enacted global frameworks [4,5,51], indicating that urban heritage can catalyze sustainability competences related to responsibility, identity, and participation;
• Multimodality and AR as interpretive scaffolds: the coupling of archival images, concise prompts, and spatial overlays supported noticing and retention while inviting debate over authenticity, consistent with narrative cartography and semantic/geospatial integration [25,29];
• Game design as a central feature: being essential to mitigate screen-centric behavior while keeping the benefits of in situ comparison [57,70,95];
• Bridging documentation with civic activation (smart heritage): even with a lightweight system, the Art Nouveau Path generated interoperable cultural traces, like multimedia records, descriptors, gameplay logs, and survey artefacts, consistent with FAIR stewardship and smart heritage governance [19,20,21] at a local, exploratory scale.

At this level, the case suggests a proportionate path through which schools and municipal partners could contribute modest, school-generated datasets to emerging national or international data spaces in the future, without full HBIM or digital twin requirements, while primarily strengthening preservation literacy and community engagement in this specific city. These contributions remain local and do not imply formal changes in heritage governance structures.

For heritage professionals, these results may suggest that AR-based paths can deepen interpretation and sustain civic value without intensive digitization. For educators, they show how competence-oriented design converts the city fabric into a cross-curricular learning space. For policymakers, they illustrate how small-scale, interoperable interventions can enact international and policy agendas by turning façades and monuments into catalysts of sustainable citizenship.

6. Conclusions

This study explored how the Art Nouveau Path fosters sustainability competencies by analyzing gameplay logs, student questionnaires, and teacher observations.

6.1. Main Conclusions

Initially, the heritage context proved to be an effective strategy for ESD. In the S2-POST study, which included 439 participants, an overwhelming majority (98.45%) found the subject of sustainability as explored through Art Nouveau to be intriguing. A substantial majority (94.53%) indicated an interest in further exploring the topic. T2-OBS (N = 24) documented spontaneous “preservation dialogues” in 58.33% of instances, substantiating heritage’s function as a mediational tool for sustainability education.

Secondly, the incorporation of multimodality and AR significantly improved user engagement, thereby promoting a constructive challenge. Analysis of the records (N = 118 groups; 4248 responses) revealed 81.00% [95% CI 78.75–83.01%; N items = 11] on AR items versus 73.00% [95% CI 71.39–74.59%; N items = 25] on non-AR items. The mean group-level accuracy in the ‘architectural detail’ subset was 83.00%, indicating a consistently high performance across these items. High AR exposure was associated with a longer route time, with a mean gap of 10.17 min (42.77 vs. 32.60). This association is statistically significant in the OLS specification with session fixed effects and small-sample robust errors (p = 0.004; see Appendix C). In S3-FU (N = 434), 81.94% (at question A.2.1.) recalled building details, with 61.06% explicitly mentioning features (at question A.2.1.1.), indicating desirable difficulty and the need to balance challenge with support.

Thirdly, the retention and transfer of knowledge extended beyond the confines of the classroom. In the S3-FU dataset, 69.35% (at question A.1.2.) reported increased sustainability in their actions, 68.20% (at question A.1.3.) shared ideas with colleagues or family members, and 79.03% (at question A.2.1.) demonstrated a closer attention to architectural details. These indicators describe self-perceived behavioral tendencies rather than externally observed civic actions.

6.2. Design Implications for Heritage-Based MARGs

The effectiveness of heritage MARGs depends on intentional multimodality. At each POI, pair a concise historical record (photograph, short video, or audio) with a single, unambiguous prompt and, when relevant, a precise AR overlay that directs attention to the feature under examination. Juxtaposing past and present supports authenticity judgments and reduces ambiguity.

Where feasible, include lightweight metadata (source, date, author) to connect in situ interpretation with documentation practices and to enable asset reuse. Low-load micro-checks, such as a one-item quiz or a brief justification, reinforce dual coding of verbal and visual information and discourage passive consumption.

It is essential to maintain a coherent narrative throughout the MARG. Replace isolated stops with a coherent arc of opening, discovery, and synthesis or feedback. Recurring motifs gain meaning as they reappear across stops. Sequenced tasks, in which a clue from one site informs interpretation at the next, build continuity and sustain inquiry.

Implementation activities need to be pre-organized. For group work, define a clear synthesis point, such as a square, a familiar street, or a museum entrance. There, each group produces a micro-narrative—for example, a captioned photograph or a 30 s audio note that explicitly links identity, authenticity, and preservation to the contemporary city.

It is important to establish a transfer system that is operational daily, extending beyond the designated route. Follow-ups such as identifying and documenting the heritage motifs in the neighborhood or collecting family memories related to specific buildings foster intergenerational bridges and strengthen belonging. A small reusable observation card listing materials, motifs, and conservation clues functions as a heritage-literacy aid. Brief periodic self-reports and, when appropriate, geotagged photographs help monitor the persistence of attentive looking and the occurrence of “preservation dialogues.”

Also, it is important to guide the experience from naked-eye observation to AR and back to naked-eye comparison to counter screen-centricity and privilege in situ checks of materiality. Overlays should remain restrained and legible, avoiding graphical noise and superfluous animation. Introduce desirable difficulty through graded hints and plausible distractors, paired with a brief “why?” prompt (for example, “Why is this a restoration rather than a repainting?”) to turn ambiguity into interpretive learning.

Last, in the cross-field of heritage and sustainability domains, the experience must be anchored in international frameworks such as the UNESCO Historic Urban Landscape approach [51], the Faro Convention [4], and SDGs [5] by using consistent descriptors, identified sources, and reusable formats, even with a lightweight stack. Schools and municipalities can generate interoperable cultural traces, including multimedia records, descriptors, interaction logs, and survey artefacts, that are compatible with, and potentially reusable within, emerging national and international data spaces, even if they do not in themselves constitute formal entries in those infrastructures or governance arrangements. In this view, the practices explored here primarily enhance preservation literacy and community attachment at the local level, illustrating how carefully designed AR itineraries can enrich interpretation and sustain civic value with proportionate technical investment.

6.3. Limitations

The evidence presented herein should be interpreted considering several limitations. Firstly, it should be noted that this is a single-case study in one city focused on a specific heritage typology, Art Nouveau façades. This limitation constrains the generalizability of the findings across places, audiences, and heritage categories. In addition, the student cohort constitutes a local convenience sample of classes participating in the PAEMA [83], so the findings are not statistically representative of wider student populations. Secondly, the conditions surrounding the sampling and implementation process introduce limitations on the external validity of this study. Specifically, classes that were joined through a municipal program during school hours, along with factors such as weather, crowding, and route logistics, were not systematically controlled during the experimental process. Thirdly, the findings are contingent on self-reported data and structured observations; student questionnaires and teacher field notes are vulnerable to social desirability, recall bias, and inter-observer variability. Furthermore, the GCQuest questionnaire data for S1-PRE, S2-POST, and S3-FU were analyzed using descriptive statistics only, without formal inferential hypothesis testing between waves. This choice reflects the exploratory character of this study, the non-probabilistic sampling, and the repeated cross-sectional design without individual pairing, and it reduces the strength of claims that can be made about differences across administrations. As a result, apparent changes over time should be interpreted as case-specific patterns rather than as statistically confirmed effects. Fourthly, gameplay logs were collected at the group level, which lacked per-student micro-interactions, dwell time per POI, and fine-grained hint-use sequences. This restricted the modeling of attention and individual pathways. Fifthly, the decisions regarding anonymity were in favor of data minimization; however, they impeded panel matching across administrations and precluded moderation analyses by demographics or prior interest. Sixth, the follow-up period was of a relatively brief duration, spanning approximately six to eight weeks following gameplay. Consequently, the extent of long-term retention and the development of civic behaviors remain uncertain. Accordingly, references to civic responsibility in this work should be understood as referring to students’ and teachers’ self-reported dispositions in this specific implementation, rather than to directly observed behavioral change or formal civic participation outcomes. The design of this study did not include a comparison path that was delivered without the application or without AR. Without a non-AR arm or a crossover condition, the unique contribution of AR and multimodality cannot be isolated from novelty, place-based inquiry, or teacher mediation. Therefore, the estimates are associational rather than causal. The eight POIs pertains to this study’s placement within an exploratory DBR cycle. It is important to note that there was no pre-registered analysis plan and no multiplicity adjustments typical of confirmatory trials. That aspect of this study necessitates a cautious interpretation of statistical signals. The ninth issue pertains to the delivery of instruments and content in Portuguese, accompanied by a GreenComp-aligned adaptation. However, the evaluation of cross-language invariance and broader transferability remains to be conducted. In the context of data stewardship, a lightweight approach was adopted, characterized by the consistent labeling and georeferencing of assets. However, the implementation of full CIDOC-CRM [18] mapping or HBIM pipelines was not undertaken, a choice that consequently restricts the immediate interoperability with high-fidelity conservation workflows. As such, the contributions to smart heritage and heritage governance documented in this study should be read as local and exploratory, illustrating how a municipal-scale educational route can generate reusable traces, rather than act as a substitute for comprehensive national data spaces or formal governance infrastructures. Consequently, the constraints imposed by the devices and settings may have exerted a significant influence on the observed engagement patterns. In urban public spaces, groups shared a single mobile device, which is subject to factors such as screen glare, ambient noise, and connectivity. These factors have the potential to influence pacing, attention, and the balance between screen-focused and object-focused observation.

6.4. Future Paths

Subsequent iterations should explicitly instrument the route as a research site. The integration of enhanced log analytics, accompanied by fine-grained temporal traces, in conjunction with core multimodal learning analytics, will facilitate a more precise characterization of the evolution of attention and interpretation across points of interest and media. These traces should be systematically triangulated with self-report measures and behavioral indicators, such as on-site actions and hint usage, to strengthen claims about transfer and reduce reliance on perception-based evidence alone.

Where sampling and design conditions permit, future work should also complement these descriptive patterns with confirmatory inferential analyses of GCQuest responses across administrations, explicitly triangulated with gameplay logs and teacher observations to assess the robustness of the observed changes across waves.

To expand external validity, future research should implement the design across a broader array of heritage typologies, including industrial, vernacular, and natural contexts. This scaling is best achieved through structured co-creation with teachers, students, and heritage professionals, using iterative design studios to calibrate multimodality, narrative coherence, and cognitive challenge to local curricula and conservation constraints.

Finally, the project’s cultural outputs, namely multimedia assets, structured descriptors, and interaction logs, should be curated for interoperability and deposited in European digital heritage infrastructures. These initiatives facilitate cross-border educational reuse of 3D and AR resources, advance cultural sustainability objectives, and demonstrate how lightweight, school-based interventions can contribute meaningfully to continental preservation ecosystems.

From a methodological perspective, future research endeavors should be designed to include a non-AR comparison or a crossover design, with matched classes and identical prompts on paper. The instrumentation of the application should be adapted to facilitate the capture of anonymized individual taps, dwell time, and hint sequences. Furthermore, the duration of the follow-up period should be extended to range from three to six months, incorporating brief micro-surveys and optional geotagged traces. The collection of minimal, ethics-approved demographics is essential for the testing of moderation. Additionally, pre-registration of confirmatory analyses and adjustment for multiple comparisons are necessary. Finally, the mapping of descriptors to CIDOC-CRM classes should be conducted, while piloting a thin GIS layer to assess the efficacy of plug-and-play interoperability.