Students’ emotions, beliefs, motivations, values, and attitudes, summarized under the

term affect, are crucial for learning mathematics, educational choices, and future life

trajectories. Over the past decade, research on affective constructs and their relations

to other educational outcomes has increased in mathematics education. This research

forum will present current developments and future directions in research on affect in

mathematics education. Contributions addressing overarching perspectives will

synthesize research from measurement, situational, and sociocultural perspectives.

Contributions focusing on specific contexts will address affect in problem solving and

problem posing, mathematical modelling and applications, and gender. The forum will

include audience discussion and invited commentary to advance research on affect.

Introduction: This thesis investigates whether repeated-session functional magnetic resonance imaging can support trustworthy decoding of affect-related representations within individuals and whether the learned mapping transfers across individuals. The underlying problem is that emotion-related decoding from neuroimaging can appear successful because of information leakage, session-specific confounds, or overinterpretation of model explanations rather than stable condition-related signal. The thesis therefore addresses a methodological need for leakage-aware evaluation and cautious interpretation in repeated-measures neuroimaging.

Research Question: The primary research question is whether repeated-session condition-level beta maps support stable within-person prediction of a three-class coarse affect target (positive, neutral, negative) under session-held-out evaluation, and to what extent the learned mapping transfers across individuals when applied without re-training. The thesis also asks whether class-specific coefficient patterns show recurring structure across validation repeats and stricter non-overlapping session partitions, while remaining bounded as model-behavior evidence rather than neural localization.

Method: The study uses a precision neuroimaging dataset with two deeply sampled individuals and condition-level beta maps derived from repeated task-based sessions. The final analysis included 19 sessions for the first participant and 23 sessions for the second participant, reflecting the participant-specific session counts available under the locked thesis scope. A locked confirmatory machine-learning protocol was applied using a regularized linear classifier, train-only preprocessing, leave-one-session-out validation for within-person evaluation, frozen cross-person transfer, nominal-reference comparison, fixed-prediction label-association permutation controls, and model-space explanation-stability analysis. Coefficient-pattern recurrence was examined both across leave-one-session-out folds and in a stricter supporting check using non-overlapping session partitions. The main supervised target was a three-class coarse affect label consisting of positive, neutral, and negative conditions.

Results: The results show within-person held-out-session decoding above the nominal three-class reference for both participants. Mean balanced accuracy was 0.811 for the first participant across 19 held-out sessions and 0.714 for the second participant across 23 held-out sessions. Crossperson transfer remained above the nominal three-class reference but was weaker, with balanced accuracy of 0.391 in one direction and 0.428 in the other. The neutral class was consistently harder to recover than the negative and positive classes. Leave-one-session-out coefficient patterns were highly similar, but these folds share most training data. In the stricter independent-partition check, within-subject coefficient similarity was modest but higher than cross-person similarity (full-pattern mean r = 0.191 versus r = 0.081).

Discussion: The thesis concludes that repeated-session beta-map representations can support stable within-person coarse-affect prediction under leakage-aware held-out-session evaluation, whereas direct transfer across individuals is possible but substantially weaker. The interpretability results provide supporting evidence for recurring participant-specific model structure, but the independent-partition analysis shows that this evidence is modest and should not be overinterpreted as strong stability or direct neural localization. The contribution is therefore methodological: a more trustworthy framework for evaluating within-person stability, cross-person transfer, and explanation robustness in repeated-session neuroimaging. Because the dataset contains only two individuals, the findings should be interpreted as evidence from two cases rather than as population-level generalization.

Introduction. In regulated quality assurance (QA) workflows, companies must demonstrate that work has been performed according to approved procedures. Procedure documents define requirements, while record documents provide evidence of performed activities. Reviewing this relationship manually can be slow, detailed, and difficult to scale, especially when evidence is incomplete, fragmented, or distributed across document sections.

Research questions. This thesis investigates how a retrieval-augmented and evidence-audited LLM framework can support requirement-level compliance assessment between procedure-derived requirements and record-document evidence. The project focuses on grounded requirement-level assessment, differences between non-retrieval, record-retrieval, and reference-assisted stages, and the role of deterministic evidence auditing in identifying supported, partial, missing, weakly matched, and contradicted evidence.

Methods. The thesis is designed as an industrial case study with a mixed methods evaluation. Document analysis is the primary method, supported by expert-informed review and quantitative comparison of framework outputs. A research prototype was implemented and evaluated using one procedure document, original company records, synthetic records, and one referenced supporting document. The framework includes requirement extraction, staged assessment, evidence grounding, deterministic audit, and metric-based evaluation.

Results. The results show that retrieval changes the quality of requirement-level assessment by increasing access to candidate record evidence and improving traceability to source documents. The record-retrieval stage provided a clearer connection between requirements and record evidence than the non-retrieval baseline. The reference-assisted stage was useful when requirement interpretation depended on the referenced supporting document. The deterministic audit helped separate grounded evidence from unsupported generated claims and made missing, partial, weakly matched, and contradicted cases easier to identify.

Discussion. The findings suggest that retrieval-augmented and evidence-audited LLM frameworks can support QA documentation review, but their outputs should remain subject to human review. The project is limited by the size and scope of the original dataset, the use of synthetic records, document-processing choices, and the selected LLM backends. Future work should evaluate the framework on larger and more varied QA documentation, improve retrieval and audit methods, and study practical integration into regulated review workflows.

Relational data, which can often be modeled as networks, inherently involve dependenciesand therefore violate the statistical assumption of independent observations. As a result,traditional statistical methods risk producing misleading results when applied to data ofthis nature. The aim of this study is to examine how Latent Space Models (LSM) andrelated model extensions can be used to model non-independent network data by capturingstructural dependencies such as reciprocity, clustering, and homophily.The analysis is based on a directed network consisting of 199 countries, where ties betweencountries represent visa-free travel from one country to another. A suitable combination ofmodel components, with respect to cluster structure, latent dimensions, and random effects,is examined. The models are evaluated using multiple criteria, including AUC, BIC andgoodness-of-fit analyses, in order to assess their ability to describe the observed network.The results show that a model with three latent dimensions, two clusters, and both senderand receiver effects provides a substantially better representation of the network structurethan simpler models. The study further demonstrates that latent network models are wellsuited for the analysis of relational data where the independence assumption is violated.

This study aims to determine the factors that affect the proportion of students who achieve eligibility for public upper secondary schools at the municipal level in Sweden.

The study utilizes quantitative data, primarily sourced from the Swedish National Agency for Education’s (Skolverket) statistical database for the 2022/23 academic year.The data is analyzed using multiple regression analysis in two distinct stages. The firstmodel examines resource factors distributed within the school environment, factors overwhich the school’s principal authority has direct influence. The second model investigates how the relationships identified in Model 1 hold up when combined with demographic factors that are beyond the school authority’s control.

The findings demonstrate that the proportion of students who achieve eligibility for upper secondary school is primarily influenced by demographic factors. The municipality’s socioeconomic status has the most distinct effect on eligibility. The proportion of students with a foreign background and the proportion of students whose parents have a post-secondary education also have a clear, significant impact. Previous school performance is critical, the average results of the national exam in mathematics in Grade 3 have a strong correlation with achieving eligibility for upper secondary school in Grade 9. Resource factors, such as the proportion of licensed teachers and the perceived calm learning environment, demonstrate some importance but are overshadowed and rendered less decisive by the demographic factors in the full model.

This thesis investigates various spatial econometric model specifications to identify the most suitable one for capturing local economic activity in a Swedish context. The analysis covers the period 2012–2018 and relies on panel data for Sweden’s 290 municipalities. The economic activity of the municipalities, measured as Gross Regional Product (GRP) per capita, is analyzed with a specific focus on commuting spatial spillover effects. To capture these spillover effects, a commuting-based weight matrix is applied instead of a geographical one. The model selection is conducted using robust Rao’s Score test, and is confirmed by evaluating the actual model estimations. The results strongly indicate a dominant exogenous spatial dependence, whereas the endogenous dependence loses its significance when accounting for neighbouring characteristics (education, demography, etc.). Therefore, the thesis indicates that a spatial lag of X model (SLX) is the most appropriate specification for these data. The factors strongly associated with a municipality’s economic activity are the employment rate and age structure of the neighbouring municipalities, as well as the municipality’s own age structure and in- and out-commuting quotas. Overall, the thesis demonstrates that a municipality’s economic activity is influenced to a greater extent by the characteristics of neighbouring municipalities via the commuting network, rather than by the neighbours’ GRP per capita.

Quality adjustment in price index construction has long been a central issue in inflation measurement, as inadequate quality adjustments may introduce systematic bias into indices and thereby affect measured inflation. In official statistics, explicit quality adjustments are commonly performed using Hedonic regression models or judgmental methods. Although hedonic models have been applied for decades, they are associated with limitations such as assumptions regarding functional form and sensitivity to multicollinearity. Judgmental methods, on the other hand, may lack reproducibility and consistency across products and over time.

This study investigates two alternative approaches to quality adjustment: Elastic Net, as an extension of the traditional hedonic model through regularization, and Random Forest, a machine learning method that may address several limitations associated with linear models. In addition to comparing the predictive performance of these methods, the study examines how different qualityadjustment approaches influence price development and measured inflation in Sweden during the period 2024–2025.

The analysis is conducted for two product groups with distinct market characteristics: mobile phones, representing a high-technology product group, and microwave ovens, representing a more low-technology product group. Mobile phones are primarily quality-adjusted using Hedonic regression, with occasional judgmental adjustments to the model-based estimates, whereas microwave ovens use judgmental methods.

The results indicate that Elastic Net did not provide a clear improvement over traditional hedonic models for either product group. Random Forest consistently achieved the highest predictive performance and the lowest prediction errors, particularly for mobile phones where the improvement relative to linear models was substantial. For microwave ovens, the improvements were less pronounced. The choice of quality-adjustment method had a considerable impact on index development at the product-group level. However, the effects on overall inflation were limited, although differences remained observable after aggregation to the total CPI. Among the evaluated methods, Random Forest produced index series most closely aligned with the officially reported inflation rate.

Since no objective benchmark for true quality adjustment exists, it is not possible to determine which method is most accurate. Nevertheless, the findings suggest that Random Forest may represent a promising alternative or complement to current quality-adjustment methods in Swedish official statistics, particularly for product groups characterized by rapid technological development.

 In this thesis, looksmaxxing is examined as a subculture and how TikTok functions as an apparatus through which ideals of masculinity are produced. These ideals are analyzed through a diffractive reading with romance literature. Netnography and diffractive analysis illuminate resonances and disruptions between the materials. The analysis reveals an ideal of masculinity through a cultural archetype in the form of the alpha male. The alpha male emerges within an affective atmosphere that conveys a suggestive and dystopian mood. The ideal produced by TikTok resonates with the alpha male that appears in the novel’s narrative world. This is particularly evident in his appearance, with a pronounced jawline, hunter eyes, and a tall, muscular body. On TikTok, appearance is understood to correlate with health through a physiognomic understanding of the body. Furthermore, behavior and relationships with others emerge as central to the alpha male’s positioning. The alpha male controls, intimidates, and outclasses others. There are points of disruption between TikTok and the novel in the alpha male’s treatment of others regarding consent and vulnerability. The conclusion is that masculinity can be understood as an intra-active phenomenon produced and legitimized through material and discursive expressions and the affect they generate. Taken together, the analysis shows how contemporary ideals are produced and disseminated on TikTok, while simultaneously resonating with another subcultural context. These subcultures thus enable and constrain how young people can express gender.

Driven by the depletion of aquatic environments and the urgent need to sustain a growing population with sustainable protein sources, research on aquaculture has gained momentum in recent years. This need is amplified by threats to food security stemming from crises such as pandemics, geopolitical tensions, and climate change. There's a clear ambition to increase aquaculture production, yet how this translates into practice remains a critical question.

This thesis broadens research on resilience, food sovereignty, and food security by examining how global dynamics shape a localized food sector – Swedish Rainbow Trout cultivation. Using a case study that combines semi-structured expert interviews, a survey of farmers, and participant observation at the National Aquaculture Conference in Mora, the study addresses (1) how resilience is operationalized within the sector, (2) how farmers' autonomy manifests in food chains, and (3) what risks must be managed to enhance food security. The findings demonstrate that for aquaculture to expand in Sweden, a dual strategy is required. First, a “push” approach through production-boosting measures is needed to simplify operations for primary producers, as permit processes, disease risks, and inadequate infrastructure create structural bottlenecks that undermine economic viability. This must be coupled with a “pull” approach that increases market demand through community connection and local branding. Consequently, this study provides a crucial reality check for 2030 targets, illustrating how localized sectors can overcome structural bottlenecks in food systems to align ambitious policy with practical production realities.  

Introduction: Organizations use competence mapping to support workforce planning, employee development, and internal mobility. Much of this information is found in internal documents such as role descriptions. These texts are written in natural language, vary in style, and are difficult to analyze in a consistent way. NLP methods offer ways to structure this material, but different representations capture different aspects of the text, which shapes the results.

Research Question: This thesis examines how different NLP text representations influence competence mapping based on internal organizational role descriptions. The focus is on how representation choices affect the identification, structuring, and communicability of competence‑related information.

Method: The study is conducted as a case study at a global manufacturing and service company and uses 89 internal role descriptions from a Nordic and Baltic regional unit. Two approaches are compared under the same conditions: a vocabulary‑based method (TF‑IDF) and an embedding‑based method (Sentence‑BERT). Both are applied using unsupervised clustering. The results are evaluated using internal clustering metrics, a robustness check with BERTopic, and a post‑hoc comparison with a formal competency framework.

Results: The two approaches produce clearly different outcomes. Sentence‑BERT generates more coherent and stable role groupings based on semantic similarity, even when roles are described with different wording. TF‑IDF produces more fragmented clusters that mainly reflect similarities in phrasing and documentation practices rather than shared meaning. The BERTopic analysis supports the stability of the Sentence‑BERT clusters. Both methods recover most of the formal competency framework, with only a small number of competencies remaining weakly represented.

Discussion: The findings show that competence mapping with NLP is not only a technical task but also a representational and organizational one. Where coverage is limited, this reflects the boundaries of what role descriptions capture in practice rather than shortcomings of the methods themselves. Role descriptions function as an unintended archive of how work is understood at a given moment. They can be analyzed productively, but only with attention to the limitations of the source material. A useful direction for future research may be less about more advanced methods and more about understanding what gets written down, by whom, and for what purpose.

Introduction: This thesis studies lithium-ion battery degradation under time-varying operating conditions. Accurate lifetime prediction is critical for electric vehicles, as degradation directly affects performance, reliability, and cost. Standard models typically assume smooth, purely irreversible capacity fade governed by power-law dynamics. However, experimental data often show deviations from this behavior, suggesting that additional mechanisms—such as reversible lithium redistribution—may influence the observed capacity loss.

Research Question: The main question is: Does modeling the anode overhang effect (reversible lithium redistribution) improve degradation modeling and prediction? Additionally, how does it affect the estimated power-law exponent and predictive uncertainty?

Method: A Bayesian degradation model is developed that combines a calendar aging component with an anode overhang term capturing reversible lithium redistribution. The model is formulated in a dynamic form to handle time-varying temperature and State of Charge conditions. Parameters are estimated using Hamiltonian Monte Carlo in Stan. The approach is evaluated on both literature and internal datasets using RMSE and leave-one-out cross-validation (LOO).

Results: Including overhang improves fit and predictive performance, especially when conditions differ from initial storage. It captures transient capacity recovery, reduces uncertainty, and shifts the estimated power-law exponent toward physically consistent values (∼0.5).

Discussion: The results show that including reversible lithium redistribution improves model fit and predictive performance, particularly when operating conditions differ from initial storage. The overhang component explains transient effects and reduces unexplained variability, which also affects the estimated degradation dynamics. However, the study is limited to calendar aging and a simplified representation of the overhang mechanism. Future work should extend the model and validate it under real operating conditions.

Video game translation is a relatively new field of study with the popularisation of video games. This thesis analyses how Japanese video games are translated and what happens to the content when translating from one language to another. To achieve this, the study builds upon previous research by Santos and Humblé regarding the preservation of honorific suffixes by focusing on the translation of Yakuza 0. Japanese is a highly contextualised language with honorifics being of high importance. More specifically, this thesis focuses on the protagonist Kazuma Kiryu’s linguistic identity and the politeness strategies he uses and how these are kept or changed in the translation. By using Richard J. Watts’ theory of politic behaviour and Yoko Hasegawa’s theory on in-group and out-group, together with Venuti’s concept of domestication and foreignisation, a comparative textual analysis of selected Japanese dialogues is conducted. The results show that even if the game’s English translation successfully captures the main plot of the story, linguistic nuances are noticeably lost in the dialogue. Kiryu’s language use, which ranges from a formality shaped by internal hierarchies when addressing his in-group, to a more aggressive tone toward his out-group, is changed in the English translation. The linguistic dynamics that make up politic behaviour in the Japanese original tends to get domesticated in the English target text. By adapting the dialogues to English-speaking norms, the specific hierarchical structures are smoothed out. Consequently, the study shows that a domesticating translation strategy changes parts of the complex group relations and cultural nuances that shape the character’s yakuza identity.

This thesis studies the use of Japanese mascot characters, also known as yuru-kyara, in place marketing in contemporary Japan. The meaning of the term yuru-kyara have broadened since the introduction of these characters and this has necessitated the need to define what kind of a character a yuru-kyara represents in contemporary Japan. In this thesis the use of these characters is studied for the purposes of providing a part to the discourse of what a yuru-kyara is and how they are used. At the start of this thesis the theme is introduced and terms specific to the field of this study are defined. After that, previous research regarding yuru-kyara is presented from three different perspectives: Character theory, softness and place marketing. These perspectives represent different ways of studying yuru-kyara and also make up the theoretical framework of this thesis. In this thesis case studies on two yuru-kyara, Gunma-chan and Chikuwan, is conducted, in which these characters are studied based on their background, design and use. These aspects of the characters are then compared and discussed with the help of the thesis theoretical framework to answer the thesis research questions. The result of this study shows a trend in contemporary yuru-kyara where they are “humanised” through the application of personality and narrative to these characters. This trend is linked to an effort by the users of yuru-kyara to motivate interaction between yuru-kyara and consumers in order to promote the place that the character represents. The results of the case study also shows that popular yuru-kyara in contemporary Japan first and foremost is a part of the marketing strategy for place marketing, and that the bigger and more popular these characters become, the more controlled and regulated their use is. Furthermore, the results of this study shows that the motivation behind the use of yuru-kyara can be linked to the effectiveness with which these characters can be linked to, and market, a place.

This thesis examines how video game music constructs representations of Japanese and Nordic cultural identity, with particular attention to differences between internally and externally produced games. Through a comparative analysis of Ghost of Yōtei, Sekiro: Shadows Die Twice, Assassin’s Creed Valhalla, and Valheim, the study applies Thomas B. Yee’s framework of cultural width and depth through the analytical parameters of instrumentation and timbre, mode and pitch organization, harmony and texture, and rhythm and tempo. The findings suggest that cultural identity in game music is primarily communicated through recognizable sonic signifiers rather than strict historical authenticity. Internally produced games demonstrate greater structural integration of culturally specific musical characteristics, while externally produced games rely more heavily on surface-level stylistic markers embedded within familiar cinematic scoring conventions. The study concludes that video game music functions as an active site of cultural representation, shaped less by objective authenticity than by player expectation and aesthetic mediation.

The COVID-19 pandemic affected everyday life and mental health across populations.Older adults were particularly vulnerable due to both the health risks of the virus andthe social restrictions introduced during the pandemic, raising concerns about increasedsocial isolation and deteriorating mental health. The aim of this study is to examine thechange in depressive symptoms from before to during the COVID-19 pandemic and toinvestigate whether partnership functioned as a protective factor against worseningmental health. In addition, the study explores whether these associations varied acrossfour European countries with differing pandemic response strategies and levels ofstringency.This study uses panel data from the Survey of Health Ageing and Retirement in Europe(SHARE), Waves 8 and 9, to compare depression levels among older adults before andduring the COVID-19 pandemic in four European countries: France, Germany, Italy,and Sweden. The countries were selected based on differences in pandemic stringencylevels (CHI). Depression is measured using the EURO-D scale, and fixed effectsregression models are used to analyze within-individual changes in depressivesymptoms over time. Supplementary OLS models are estimated as robustness checks.The results show that depressive symptoms increased during the pandemic across thestudy population, although the estimated effects were generally modest. Womenreported higher overall levels of depression than men, although no statisticallysignificant gender difference was observed over the study period. Contrary to the initialhypothesis, living with a partner was not associated with a smaller within-individualincrease in depressive symptoms, but partnered individuals did still report lower levelsof depressive symptoms. Country-specific analyses suggested some variation across national contexts, with Italy displaying a distinct patter, although several of the country-specific estimates were not statistically significant.

The rapid adoption of embedded and Internet of Things (IoT) devices has dramatically expanded the cyberattack surface, while their resource constraints, protocol diversity, and architectural heterogeneity make security assessment increasingly difficult. Penetration testing remains the primary validation methodology for these devices, but it is largely manual or constrained by rigid automation. Recent advances in Large Language Models (LLMs) show promise for autonomous penetration testing, yet existing frameworks predominantly target traditional IT environments and do not support heterogeneous embedded systems.

Therefore, the central problem addressed in this thesis is the lack of a validated framework for integrating LLMs into the full embedded penetration testing process, and the research question guiding this study is: How can an LLM-driven embedded penetration testing framework be designed to effectively automate the security assessment of embedded systems?

Following the Design Science Research (DSR) methodology, this thesis designs, implements, and evaluates an LLM-driven framework that automates embedded penetration testing across heterogeneous interfaces. The artifact combines a hub-and-spoke multi-agent architecture with a centralized Orchestrator that maintains a dynamic Penetration Task Graph (PTG), specialist agents operating in bounded ReAct loops, and a novel pluggable Interface Layer that abstracts physical and logical transport channels (Ethernet, Serial, USB, CAN, Debug) into a uniform containerized execution environment. Firmware images are additionally supported as an alternative input, and a Human-in-the-Loop (HITL) mechanism gates state-modifying actions.

The artifact was demonstrated against five emulated targets and a firmware image, and evaluated against a real embedded device using manual penetration testing as a baseline. Results show that the framework reproduces meaningful outcomes, autonomously chains multi-step attacks from reconnaissance to post-exploitation, and substantially reduces engagement time even though coverage remains lower, confirming that intelligent multi-agent orchestration combined with interface abstraction can effectively automate embedded penetration testing.

This study aims to explore how visualizations of movement patterns through Registration, Documentation and Visualization (RDV) can support reflection and professional development among student teachers during their teaching practice placement (VFU). The study was conducted as an exploratory qualitative case study at Stockholm Science and Innovation School (SSIS) and is based on semistructured interviews with VFU-students, teachers and VFU-teachers. Observations of teaching situations of two participants were also conducted to provide contextual understanding of how the technology functions in practice, though these observations were not included in the main thematic analysis.

The findings were analyzed using thematic analysis and show that RDV is primarily perceived as a tool for reflection and mentoring sessions rather than for evaluating teaching quality. Participants described visualisations as a concrete basis for discussion during mentoring sessions, helping student teachers identify movement patterns and teaching practices that may otherwise be difficult to notice or recall. At the same time, the usefulness of the technology was found to depend heavily on the teaching context, mentoring relationships, and how the technology is introduced and implemented. The study also highlights tensions between professional support and feelings of surveillance, as well as practical and organizational barriers to the use of RDV in teacher education.

Sustainability is one of the most common terms in public discourse. But what is sustainability? While social practice theory has advanced understanding of routines and social action, fewer studies trace how sustainability is produced within material production systems and across scales. This study conceptualises sustainability as an emergent feature of how social-ecological relations are configured and transformed through practice. Combining practice theory, object biography and relational political economy, it follows wool and silk across four nodes of the Albanian animal fibre supply chain, from shepherding and sericulture to industrial manufacturing. The analysis shows how practices are continuously configured through relations between fibres, skills and meanings, and how fibres actively shape the practices that surround them. Sustainability emerges where practices remain closely aligned with ecological conditions and declines as these connections are broken through commodification, scalability and alienation. The push and pull between practices and materials reveals how human action and ecological conditions continuously remake each other rather than operating as distinct domains. The study contributes to the sociology of sustainability by grounding an abstract concept in lived material realities and offering a transferrable analytical approach for tracing how social-ecological relations are maintained and reshaped across contemporary production systems. 

Forests in Swedish Sápmi are socio-ecological and political landscapes, co-produced through historically situated relationships between humans, non-humans, and governance regimes, structured by unequal power relations in which Western extractivist and conservationist knowledge systems are privileged over Indigenous ones.Focusing on the Sami of Jokkmokk, this thesis examines forests as living landscapes that sustain livelihoods, memory, and spiritual relations, while functioning as political and epistemic arenas. Sami relationships with forests are grounded in Indigenous knowledge systems based on reciprocity and relational ontologies, underpinning practices such as reindeer herding as well as broader cultural, ethical, and cosmological frameworks. Industrial forestry and extractive industries increasingly impose legible, manageable landscapes, marginalising Sami epistemologies and exerting cumulative pressure on Sami territories. This threatens Sami livelihoods, lands, cosmological relations, and the cultural and archaeological traces embedded in forests such as culturally modified trees, sacred sites, place names, and oral histories.Drawing on a multi-layered theoretical framework, this thesis foregrounds forests as co-produced and contested spaces to demonstrate how forest transformation and destruction are inherently political acts rooted in epistemic injustice. Addressing these conflicts requires approaches that recognise forests as relational landscapes rooted in history, and that take Indigenous knowledge systems seriously as vital to just and sustainable forest futures.

Introduction: Young adults’ employability remains a significant challenge in Sri Lanka, where many young people struggle to secure employment aligned with their qualifications. As digital learning opportunities expand, Massive Open Online Courses (MOOCs) have emerged as accessible platforms for developing relevant skills. However, limited empirical evidence exists on how MOOC participation relates to employability within the Sri Lankan context, particularly from the learner’s perspective.

Drawing on self-determination theory and human capital theory, the study examines the roles of self-determination theory constructs and perceived skill development in explaining the relationship between MOOC participation and perceived employability.

Research Question: How does participation in Massive Open Online Courses (MOOCs) influence perceived employability among Sri Lankan young adults?

Method: The research adopts a quantitative survey-based design using a structured online questionnaire distributed among Sri Lankan young adults aged 18-35 who had completed at least one MOOC. Participants were recruited through convenience sampling using social media platforms, professional networks, online learning communities, and personal contacts. A total of 251 responses were collected, of which 238 valid responses were retained for analysis after removing incomplete responses. Data were analyzed using correlation, regression and mediation analysis techniques. Employability and skill development were measured based on the learner’s self-reported perceptions rather than employer evaluations or objective employment outcomes.

Results: The findings revealed a positive association between MOOCs participation and perceived autonomy, perceived competence, perceived relatedness, perceived skill development, and perceived employability among Sri Lankan young adults. The results further indicated that self-determination theory constructs partially mediate the relationship between MOOCs participation and perceived skill development, while perceived skill development mediates the relationship between self-determination theory constructs and perceived employability.

Conclusion: Overall, the findings suggest that MOOCs participation is positively associated with perceived employability not only through knowledge acquisition but also by supporting learner motivation, engagement and perceived skill development. The study further highlights the potential of MOOCs as flexible learning platforms that support perceived skill development and perceived employability while encouraging workforce readiness and continuous learning among young adults in Sri Lanka.

The study also contributes to the limited empirical literature on MOOCs and employability in Sri Lanka by demonstrating the roles of self-determination theory constructs and perceived skill development in explaining perceived employability outcomes.

Predicting startup success with machine learning is complicated by high-dimensional, noisy features, class imbalance, and data leakage. This thesis uncovers widespread data leakage issues in the literature. It examines whether previous results on startup success prediction replicate under a leakage-aware, temporal evaluation scheme, and whether LLM-derived textual features and macroeconomic indicators add predictive value when offered through randomized search.

Using Dealroom data on 10,000 European startups (from Sweden, Norway, Denmark, Finland, Germany, the Netherlands, and Belgium), we introduce per-instance success windows, rigorous leakage controls, and walk-forward evaluation, and conduct a comparison between walk-forward evaluation and random cross-validation. We train XGBoost classifiers on five binary success targets: achieving a funding round, exit, and upper-quantile growth in funding, employees, and valuation. Under walk-forward evaluation, AUROC ranges from 0.61 to 0.77 across targets, with precision 1.05–2.29× over baseline; even though we employ randomized search with highly regularizing hyperparameters and aggressive early stopping, overfitting remains substantial.

Funding round and exit targets, which are common in prior research, yield weak predictive performance, whereas upper-quantile targets for funding, employee, and valuation growth are promising. Embeddings of company descriptions account for 19–50% of SHAP feature importance, whereas LLM-generated scores account for only 2–8%, suggesting the reliability of zero-shot prompting in this domain is questionable. Macroeconomic features (10-year treasury bond yields) are negligible, likely because the ten-year data window covers too few distinct macroeconomic regimes for the model to learn a generalizable relationship under walk-forward evaluation. Headquarters in Germany, the founder's business background, and total work experience emerge as the most important individual predictors. Replacing walk-forward with random cross-validation inflates AUROC by up to 20 percentage points, suggesting that prior results are likely optimistic. These findings underscore the need for temporal evaluation, leakage controls, and multi-target assessment in research on predicting startup success.

The increasing reliance on connected products and digital systems has created new challenges for cybersecurity governance in the European Union. In response, the Cyber Resilience Act (CRA) introduces broad cybersecurity requirements for products with digital elements, requiring manufacturers to ensure security throughout the product lifecycle. This thesis examines how organizations in Sweden work with post-market cybersecurity practices in relation to these requirements, with a focus on vulnerability management, security updates, incident reporting and documentation.

A qualitative research approach was used, based on semi-structured interviews with security specialists and consultants experienced in cybersecurity, regulatory compliance and product security. The material was analyzed using thematic analysis.

The findings show that organizations with higher cybersecurity maturity more often have established processes for vulnerability management, risk prioritization, patching and incident reporting. Other organizations work in a more reactive and situation-dependent manner. The study also identifies that uncertainty regarding the Cyber Resilience Act’s requirements on substantial modification may create incentives to avoid larger security patches for older products.

The study concludes that current post-market cybersecurity practices partly align with the requirements of the Cyber Resilience Act, but that the degree of alignment varies depending on cybersecurity maturity, use of standards, industry sector and available resources. The main gaps concern proactive vulnerability management, third-party dependencies and systematic documentation. This thesis contributes empirical knowledge on how organizational factors shape post-market cybersecurity work in relation to the Cyber Resilience Act. It particularly highlights how uncertainty regarding legacy management and vulnerability reporting may create practical barriers.

States organize cryptographic governance in different ways, particularly when secure communication must support both national-security control and civilian participation. In Sweden, this area is known as signalskydd: a holistic protection of communications. Swedish signalskydd faces increasing pressure to adapt to emerging technologies, and civil-military interoperability while preserving state authority and classification-based control. This thesis compares national models to determine their implications for the future development of Swedish signalskydd. It then applies a qualitative multiple-case study strategy using a structured, focused comparison of Belgium, the Netherlands, and the United States. The analysis is based on three semi-structured expert interviews, public legal and policy documents, and scientific literature. The material is analyzed through thematic analysis and then synthesized through three analytical dimensions: cryptographic key management, commercial cryptography governance, and civil–military synergies. Three governance models were identified, a legally centralized and classification-driven model, a federated model supported by trusted intermediary institutions, and a layered model combining public standard-setting with controlled but opaque cryptographic governance. Across the cases, key management emerges as a life-cycle governance issue rather than a purely technical function, while civil–military cooperation depends on trust but not open sharing. The thesis concludes that Swedish signalskydd should preserve strong state authority and classification-based control while becoming more agile in relation to commercial cryptography, and emerging technologies to retain civil–military interoperability. Sweden should not copy one foreign model, but let future development combine clear responsibilities, usable secure communication, and accountability-preserving governance.

Svordomar är en central del av språket och används i olika syften, från att uttrycka känslor till att stärka sociala relationer. Trots detta har svärande hos tvåspråkiga ungdomar endast studerats i begränsad omfattning. Denna studie undersöker svordomsanvändningen hos polsk-svensktalande ungdomar i Sverige, med fokus på deras språkpreferenser samt attityder gentemot svordomar i olika sociala sammanhang. Genom en kombination av enkäter och intervjuer med ungdomar och deras föräldrar analyseras vilka språk som föredras vid svärande, samt hur svordomar uppfattas i familjära och vänskapliga relationer. 

Resultaten visar att ungdomarna svär både på svenska och polska, men att deras språkval påverkas av sammanhang och mottagare. Svenska används oftare i informella situationer, särskilt bland vänner, medan polska svordomar upplevs som starkare och används mer selektivt, särskilt i föräldrakontext. Studien belyser även föräldrarnas generellt negativa inställning till svordomar och hur detta påverkar ungdomarnas språkbruk. Genom att bidra till forskningen om tvåspråkighet och känslomässigt språkbruk ger denna studie en djupare förståelse för hur språkliga normer och kulturella bakgrunder påverkar svärande hos flerspråkiga ungdomar.

This study examines how managers within the Swedish social service interpret and understand the new social service law which came into effect 1 July 2025. The reform aims to create a social service that is more preventative, easily accessible and knowledge based. But this reform was enforced without detailed instructions concerning its practical execution. The study depicts how low government governance causes leadership freedom, which consequently leads to differences in working methods amidst municipalities. The results showcase a balancing act between the social service’s certainty in fulfilling their responsibility as an authority and adjustments based on local needs.

The rapid influx of data-driven models into the industrial sector has been facilitated by the proliferation of sensor technol-ogy, enabling the collection of vast quantities of data. However, leveraging these models for failure detection and progno-sis poses significant challenges, including issues like missing values and class imbalances. Moreover, the cost sensitivityassociated with industrial operations further complicates the application of conventional models in this context. This paperintroduces a novel cost-sensitive transformer model developed as part of a systematic workflow, which also integratesa hybrid resampler and a regression-based imputer. After subjecting our approach to rigorous testing using the air pres-sure system failure dataset from Scania trucks and the semiconductor manufacturing dataset, we observed a substantialenhancement in performance compared to state-of-the-art methods. Moreover, we conduct an ablation study to analyzethe contributions of different components in our proposed method. Our findings highlight the potential of our method inaddressing the unique challenges of failure prediction in industrial settings, thereby contributing to enhanced reliabilityand efficiency in industrial operations.

Introduction: Working dogs have been used in areas of societal importance, for example in the armed forces where they assist with patrolling, searching and guarding. At the same time, the research field of more-than human design and Animal-Computer interaction has emerged with a focus on including animals as active participants in the design process. However, previous research still believes that animals are not seen as their own individuals, and lack control over how they use technology as a tool. This study takes a closer look at Dog-Computer interaction, to investigate how technology can be designed to support the dog’s own needs and desires.

Research question: To investigate how technology can support dogs’ control and interaction with technology, the study aims to answer the research question: “How can a training robot for working dogs be designed to complement humans in search training in a home environment?”

Methodology: The study was conducted using a Research through Design strategy where dogs and dog handlers were involved through observations, semi-structured interviews and preference tests. A total of eight dog handlers participated together with their dogs, all of whom are in active service within Hemvärnet, a part of Försvarsmakten. The collected data was analysed through affinity diagrams to identify themes and user needs that formed the basis for the development of several prototypes and design concepts. The dogs were allowed to explore different elements of the intended training robot; movement, sound and physical interaction with handles. Based on the observation data and early concepts, a final design concept was developed.

Results: The results show that dogs reacted either positively or neutrally to a prototype that moves, in the form of a radio-controlled car. The dogs also responded more strongly to commands from the dog handler’s recorded voice than a text-to-speech voice. Preference tests showed that the dogs preferred certain types of handles and materials for physical interaction. Based on these observations, a final design concept of a training robot was created, with the aim of allowing the dog to train search at home without the need for the dog handler to be present.

Discussion: The results of the study show that dogs’ behavior and preferences can be used as an active part of the design process and contribute to the development of technology adapted to their needs. The results also indicate that the dogs’ choices are guided by their motivation, although it is unclear whether the motivation is inherent or learned through experience. Further research could have explored a wider diversity of dogs and how this could have affected the results, as well as further developing the training robot to include several critical elements for the dog’s training.

This thesis examines how formal information and cybersecurity requirements are operationalised in practice in Swedish small and medium-sized enterprises. It focuses on governance-practice gaps between formal cybersecurity governance, including policies, standards and regulatory requirements, and everyday organisational practices.

More specifically, the thesis addresses four research questions concerning: (1) how cybersecurity governance structures and regulatory requirements are translated into operational practices; (2) how information classification is applied in SMEs; (3) what governance-practice gaps emerge between formal policies and everyday practice; and (4) how a structured assessment model can support the identification and analysis of such gaps.

The study adopts a qualitative multiple-case design involving four Swedish organisations. It combines semi-structured interviews, document analysis and employee survey responses. The analysis is guided by IT governance theory and organisational perspectives on security implementation. The assessment model served as an analytical prototype for comparing evidence across cases, rather than as a standalone compliance tool.

The findings show that cybersecurity requirements are translated pragmatically and incrementally rather than implemented as fixed compliance packages. Governance-practice gaps were identified in all cases, but appeared in different forms, including limited policy familiarity, weak follow-up, informal workarounds, unclear routines and inconsistent employee adoption. Information classification emerged as a concrete example of this broader implementation challenge.

The study shows that cybersecurity governance in SMEs depends not only on formal policies and responsibilities, but also on whether these are understood, maintained and embedded in daily work. The assessment model supported systematic comparison across cases, but its outputs remained supplementary and required qualitative interpretation.

Introduction:

Harm reduction has gained international policy traction and increasing recognition as essential for health equity, while punitive drug control approaches have re-emerged. This study examines Sweden, known for its restrictive drug policy, to explain why harm reduction measures have gained greater acceptance alongside increasingly severe penal responses to drug offences. Furthermore, it explores the role of stakeholders in shaping these policy processes.

Methods:

We analyzed three data sets 2015–2025 (media texts, key informant interviews, stakeholder comments to two commissions of inquiry) by using the Multiple Streams Framework.

Results:

Two main problem constructions were identified: drug-related mortality and organized criminality controlling illicit drug markets. Both were related to the opening of two policy windows; however, only one led to policy change (stricter penalties), while the other did not, despite broad stakeholder support (evaluating drug use criminalization). The results highlight the role of specific stakeholders as policy entrepreneurs and indicate growing acceptance of harm reduction, although measures perceived as challenging the control-oriented track remain controversial. The results support previous findings of Sweden adopting a dual-track policy structure. However, the control-oriented policy track continues to dominate drug policy formation.

Conclusion:

This study suggests that harm reduction is likely to remain subordinate to control-oriented approaches, when the policy tracks remain institutionally separate, and when control institutions, conventions, and key stakeholders hold greater authority than human rights frameworks and may act as status quo entrepreneurs. Furthermore, focusing on control measures due to criminal networks and drug trafficking may hinder harm reduction development.

Syftet med vår studie är att undersöka hur rastverksamheten i skolan uppfattas av både elever ochpedagoger, samt att belysa de likheter och skillnader som finns i deras olika perspektiv. Studienbygger på en kvalitativ metod och genomfördes på två skolor genom semistrukturerade intervjuer medpedagoger och samtalspromenader med elever. Det empiriska materialet analyserades genom tematiskanalys.Resultatet av studien visar att rasten framstår som en komplex del av skolans verksamhet därpedagogiska ambitioner, sociala relationer och organisatoriska förutsättningar samspelar. Pedagogernabeskriver rasten som en viktig plats för elever som fokuserar på att kunna erbjuda elever enhändelserik upplevelse med aktiviteter för att skapa trygghet och relationsskapande situationer.Samtidigt framkommer att organisatoriska faktorer, såsom bemanning, ansvarsfördelning och olikastrukturella förändringar, som påverkar vad som faktiskt blir möjligt att genomföra i den praktiskavardagen.Eleverna lyfter framför allt fram rastens betydelse som en social mötesplats där de får möjlighet attumgås med andra. Studien visar också att det kan uppstå spänningar mellan pedagogiska ambitioneroch organisatoriska villkor, vilket i sin tur påverkar hur elevinflytande och delaktighet kan ta form irastverksamheten.

This thesis presents a comparative analysis of Swedish and Polish idiomatic expressions related to emotions. The aim of the study was to examine and characterize metaphorical patterns revealed in selected Swedish and Polish idioms. The starting point for the analysis of the selected material was the concept of conceptual metaphors developed by George Lakoff and Mark Johnson. The research material was collected from Swedish and Polish phraseological dictionaries. The analysis is based on a model inspired by the research of Zoltán Kövecses, focusing on source domains present in English idiomatic expressions. The thesis addresses research questions concerning the occurrence of conceptual metaphors in selected Swedish and Polish idioms describing anger, sadness, joy, fear, and surprise, as well as the differences and similarities in the metaphorical conceptualization of emotions in both languages based on the selected material. The results primarily indicate similarities in the conceptualization of emotions in both languages, as well as the presence of similar source domains in the selected idioms from both languages. Additionally, a dominance of so-called embodied metaphors in idiomatic expressions was observed, along with certain differences in the frequency of occurrence of source domains within particular emotions.

 The aim of this study was to analyze Sida’s loan guarantees in Latin America to assess whether they are shaped by global “best practices” or a “best-fit” model where the projects are adapted to the local conditions and context. Financial inclusion is widely promoted as a development tool to stimulate economic growth and reduce poverty, with loan guarantees serving as a risk-sharing mechanism between financial institutions and donors when lending to micro, small and medium-sized enterprises (MSMEs). However, the extent to which the design of these Sida projects is adapted to the specific local contexts and conditions as opposed to are designed based on set best practices, remains unexplored. Using a case study approach and a directed qualitative content analysis, this research applies a “best practices” and a “best-fit” model as the theoretical framework to evaluate the design and implementation of three Sida guarantees for MSME-projects. More specifically the theories are operationalized in the form of two sets of indicators that the material is checked against which is Sida’s project documentation. The study examines six different areas: financial design and institutional structure; client targeting and scale; risk assessment and client protection; structural barriers and nonfinancial services; and donor role. The findings show that the design of Sida’s guarantees for MSME-projects in Latin America is both aligned with and deviate from both the “best practices” and the “best-fit” models.

Defense Procurement in Brazil: A Case Study of the FX-2 Program. The thesis examines Brazil’s defense procurement system through a case study of the FX-2 Program, focusing on its contribution to technological autonomy and defense industrial development. Situated within a broader context of shifting global power relations and the critical role of sovereignty, the study investigates how defense acquisition processes, namely offset agreements, operate as instruments of state-led industrial and technological development. The FX-2 Program is selected as a case study due to its extensive co-production framework and the strategic partnership between Brazil and Sweden. The analysis traces the evolution of the program across its implementation trajectory, beginning with the signing of the agreement in 2014.

This thesis comprises a substantial literary translation from German into Swedish, accompanied by a translation commentary. The source text is an extract from Christiane F. – Mein Zweites Leben (2013), an autobiography by Christiane Felscherinow, ghostwritten by journalist Sonja Vukovic. The translation aims to introduce a Swedish readership to the story of Felscherinow’s life after her global recognition as the main character in the 1978 book and 1981 film Wir Kinder vom Bahnhof Zoo. The thesis outlines the book and its protagonist, presents theories from translation studies, analyses the source text and comments the translation process. Guiding principles are acceptability‑oriented translation and a non-manifest voice of the translator. A recurring challenge during translation was to preserve the character of the original while creating a credible target language with an invisible translator. Additional issues include style, slang expressions, sentence structure, medical terminology and cultural references. The thesis concludes with a discussion on the complexity of translation and suggestions for further research.

This ethnomusicological research explores the establishment of salsa as a music genre and cultural phenomenon in Sweden through qualitative interviews with four musicians active in the Stockholm area. 

Previous studies have primarily focused on the genre’s formation in the United States and Latin America, particularly during the Golden Age of Salsa in New York in the 1970s. In contrast, the development of salsa in Sweden has remained unexplored. 

Through oral histories of the musicians who stood at the origins of the salsa scene the study has succeeded in reconstructing the early history of Swedish salsa music. Special attention is paid to Annorlunda Music School (Sollentuna Jazz Workshop), which celebrates its 50th anniversary this year. As Sweden's first institution dedicated to Latin American music, it has played a crucial role in educating and connecting generations of musicians. 

The interview material, examined alongside previous research, reveals significant differences between the emergence of salsa in the United States and its development in Sweden. While both scenes were shaped by transnational cultural exchanges, their formation was influenced by distinct social factors, including gender, socioeconomic status, ethnicity, and migration patterns. 

Moreover, the study demonstrates how salsa became a transnational cultural phenomenon that, despite being “conceived, nurtured and developed” in the Latin American El Barrio in New York, spread far beyond its geographical origins and united musicians across national, ethnic, and social boundaries. 

This article discusses the documentary Jesus Christ Airlines (2001) which focuses on Joint Church Aid, a collaboration platform that transported humanitarian aid to Biafra through an airlift 1968–1970. The documentary is examined as a historical film that seeks to inspire collective humanitarian action in the present. The article addresses the lack of historical analysis in a sector which remains focused on the present and the lingering negligence of the masculine norms that are upheld by humanitarian work. It argues that the documentary replicates the logic of emergency thinking: the immediacy of humanitarian action leaves no room for reflection, neither in the 1960s nor in 2001. Instead, it provides men from the Global North both an escape from, and a means to live up to masculine norms, provided that they are willing to take risks and provided that the humanitarian other remains a perpetual victim.

We evaluate a temporary public sector employment program targeted at individuals with a weak labor market attachment. To account for dynamic selection into the program, we apply the dynamic inverse probability weighting approach. We show that the program is successful in terms of increasing employment and reducing social assistance. The positive employment effects are driven by individuals placed at regular workplaces. For participants with temporary employments at a non-regular workplace, we instead find negative employment effects. The decrease in social assistance is partially countered by an increase in the share receiving unemployment insurance benefits, indicating that municipalities are able to shift costs from the local to the central budget. We further find that cost-effectiveness, as measured by the marginal value of public funds, varies depending on which level of government is considered.

The purpose of this study is to analyze the genericity of Sentimental Education (1869) by Gustave Flaubert, through money as a literary theme. Bernard Gendrel’s definition of the novel of manners is used to examine how money is represented through the three dimensions identified in his theory: the novelistic, psychological, and social dimensions. The results of the study show that all three types of justifications are present in the representation of money in Sentimental Education. The analysis indicates a strong presence of the social dimension, which makes the literary representation of money in the novel largely consistent with its representation in the novel of manners. However, the novelistic dimension also remains relatively important, which is not typical of this very genre. The distinction between the social and psychological justifications proves less clear than the distinction between these two dimensions and the novelistic justification. Gendrel’s definition of the novel of manners also appears limited by the absence of clear criteria for determining what constitutes a dominant literary justification. 

This study analyzes the opportunities and challenges of reading comprehension in a foreign language (L2), with particular attention to the teaching of Spanish as a foreign language (SFL) in the context of the digital era and generative artificial intelligence. A qualitative case study approach is adopted, based on data collection through semi-structured interviews, questionnaires, and the observation of a multimodal digital reading activity with upper secondary school students in Sweden. Data analysis followed a thematic categorization procedure integrating dimensions related to reading habits, cognitive processes (attention, working memory, inferences and metacognition), the use of digital tools, and multimodality, based on a structured instructional observation scheme.

This research is grounded in the perspective of multiliteracies, which understand reading as a complex practice involving multiple semiotic modes, including text, image, audio, and interactive digital resources. It also considers the role of generative artificial intelligence as part of the ecosystem of tools that shape contemporary reading practices in digital educational environments. From this perspective, the study explores how reading comprehension processes are mediated by the interaction between the reader, technological resources, and the multimodal environment.

Este trabajo de investigación presenta un análisis ecocrítico y ecosocial de la novela El hombre del cartel de María José Ferrada. El objetivo central es investigar cómo la representación del territorio y el espacio urbano refleja la exclusión social, la precariedad y la crisis ecológica contemporánea. La investigación se sustenta teóricamente en el concepto de Capitaloceno, la ecocrítica urbana y las teorías sobre marginalidad y expulsión.

Mediante una metodología cualitativa basada en una lectura detallada y descriptiva, se examina el entorno narrativo no como un fondo pasivo, sino como un agente que articula dinámicas de poder y desigualdad. El análisis demuestra que la obra evidencia la lógica de acumulación capitalista, que transforma el territorio en un recurso explotable y produce sujetos “desechables”. Asimismo, se explora la existencia de una ética del cuidado en espacios marginales como forma de resistencia. Se concluye que la novela vincula inseparable la degradación ambiental con la vulnerabilidad social, proponiendo una ecología de la percepción que cuestiona los modelos de desarrollo deshumanizantes y plantea nuevas formas de habitar un mundo en crisis.

Este estudio analiza cómo estudiantes de español como lengua extranjera (ELE) en Suecia interpretan representaciones de la familia latina en videos humorísticos. También examina qué tipo de reflexión intercultural aparece después de trabajar con estos videos en clase. El estudio se basa en el enfoque de Intercultural Language Teaching (ILT), que entiende el aprendizaje intercultural como un proceso de observar, comparar, interpretar y posicionarse ante distintas perspectivas culturales (Liddicoat, 2004). La intervención se realizó con estudiantes de ELE del nivel 1 en el curso Lenguas extranjeras - continuación en Suecia y los datos se recogieron mediante cuestionarios y hojas de trabajo. El análisis combinó análisis temático y estadísticas descriptivas. Los resultados muestran que los estudiantes ya poseían ideas previas sobre la familia latina, asociadas principalmente con elementos visibles como la música, el baile y la comida. Durante la intervención, algunas respuestas reprodujeron estereotipos, mientras que otras evidenciaron formas iniciales de interpretación, comparación y reflexión crítica. El estudio sugiere que los materiales audiovisuales humorísticos, cuando se acompañan de tareas guiadas, pueden favorecer procesos iniciales de reflexión intercultural en el aula de ELE 

This study aims to contribute knowledge about how preschool teachers experience and understand responsibility and professional roles in preschool practice, as well as how the relationship between formal requirements and the practical organization of responsibility is expressed in everyday work. The study is based on a qualitative approach, and empirical data were collected through an anonymous asynchronous online focus group with licensed preschool teachers. The analysis was conducted using qualitative content analysis and interpreted through a neo-institutional theoretical perspective. 

The results show that responsibility and professional roles are shaped in a tension between formal governing documents and established organizational practices. Although preschool teachers describe professional responsibility as clearly formulated in policy documents, responsibility is often negotiated and interpreted in relation to collegial norms, relationships and organizational conditions. The study also shows that a flat organizational culture influences how responsibility, leadership and professional knowledge are understood and practiced. Furthermore, the results indicate that unclear responsibility and limited opportunities for professional dialogue contribute to the individualization of responsibility. The study highlights how institutional norms and organizational structures affect the possibilities for professional responsibility to gain legitimacy and practical impact in preschool practice.