Title Promoter Affiliations Abstract "On the role of eye gaze in Flemish Sign Language: A multifocal eye-tracking study on the phenomena of online turn processing and depicting" "Myriam Vermeerbergen" "Multimodality, Interaction and Discourse, Antwerp Campus, Multimodality, Interaction and Discourse, Leuven" "This dissertation provides a first systematic account into the role of eye gaze in Flemish Sign Language discourse. Using a multifocal eye-trakcking approach, I investigated 1) unaddressed participants' gaze behavior and 2) the role of eye gaze in the coordination of depictions.  In the first study, I used eye gaze as an empirical measure of anticipation processes. I have thereby shown that the tight temporal coordination of interaction is also reflected in unaddressed participants in question-response sequences. The second part of the dissertation involves an analysis on the construal of depictions and the role of eye gaze therein. I explored depicting as a method of communication and accounted for the variety of semiotic resources signers draw on when creating depictions rather than only looking at prototypical forms of depictions, i.e., classifier constructions and enactment. Continuing on that line, I highlight that signers show similar gaze behavior across different types of depictive tokens and thus argue that eye gaze is not related to particular sign types, i.e., particular forms, but rather to the method of depicting. Moreover, I also shed light on a dynamic interplay between different functions of eye gaze in turn management. In this way, this dissertation underlines the idea that signed languages are heterogeneous systems in which signers convey meaning by a range of different semiotic resources that differ in degree of conventionalization. Furthermore, I show that signers' gaze behavior differs according to the action(s) they are performing and I highlight striking similarities with findings reported on in studies on spoken languages. In this way, this dissertation is also a call to further explore the multimodal nature of human interaction and its fundamental characteristics in parallel corpora.      " "Interactive learning in medical imaging AI: exploiting domain knowledge provided by on-line user feedback" "Frederik Maes" "Processing Speech and Images (PSI)" "Deep learning for automated delineation and quantification in medical image analysis is complicated by the fact that expert annotated data are scarce, ground truth delineations are subject to observer variability and validation in the clinical setting is needed to get insight in the clinical requirements. Instead of striving for fully automated solutions, keeping the expert clinical user in the loop allows to interactively collect high-quality feedback on the performance of the AI-tools and the user expectations that helps to improve these tools and that offers new possibilities for including rich domain-specific knowledge in deep learning approaches. By collecting user feedback on-line, novel information becomes available to improve the quality of subsequent predictions (e.g. learning from mistakes) or to provide the user an estimate of the reliability of the predictions (e.g. related to local image ambiguity or global image quality). In this project, the concept of interactive learning will be investigated and explored in various real-world medical image segmentation applications in close collaboration with clinical users to develop proof-of-concepts and integrate these in the clinical workflow." "The analysis of on-line algorithms in problems with a network structure" "Ann Nowe" "Informatics and Applied Informatics" "Online distributed solvers will be investigated in this project, and specifically in the context of many-core systems. These systems address the scalability problems present in current multi-core systems, and are the future of computation, in terms of hardware." "Moving language barriers. A mixed-methods study of the implementation of school-based language policies in primary schools" "Kris Van den Branden" "Language, Education and Society, Leuven" "All schools in Flanders are mandated to implement a school-based language policy in order to enhance students’ proficiency in Dutch, the language of instruction (Corson, 1990; Mullis et al., 2016). To date, little empirical evidence is available on how schools implement their SLP, and whether it enhances language teaching and learning. This dissertation investigated whether schools manage to implement a policy that effectively promotes student language development. SLP, good for the strong, strong for the weak?The first two studies explored the effectiveness of SLP implementation. The first study that we conducted was a quantitative multilevel study in which indicators of SLP implementation in 28 primary schools were related to the reading outcomes of around 3000 students in first, third and sixth grade. Results indicate that some SLP indicators are significantly positively related to students’ technical reading performances. Lower-grade pupils performed better on tests measuring reading decoding skills in schools with more reflective capacity on language instruction than pupils in schools with less reflective capacity on language instruction. SLP seems less present. By contrast, no significant relationships between other indicators of SLP and decoding skills, and between SLP and student reading comprehension skills were found. In addition, SLP was not found to be differentially effective for pupils with a different language background or low SES. In sum, this study does not provide strong evidence that SLP implementation enhances all students’ Dutch language proficiency. We hypothesize that either SLP is not effective as a measure to promote more complex language skills, that it is not implemented effectively within schools (Fixsen et al., 2005), or that it was not measured adequately in this study.In order to further explore these findings, the second study looked closer into the configuration of school-based SLP. Qualitative data were collected in a subsample of six schools from the first study. Focus group interviews were conducted with the SLP policy making team and teachers, and were supplemented with an analysis of policy documents, class observations and focus groups with pupils (Miles & Huberman, 1994). Insights from EER on what constitutes effective general school policies were used to score the quality of the SLP of the 6 schools (Kyriakides et al., 2010; Reynolds et al., 2014). The results of this study confirm the hypotheses put forward in the first study. School teams take several measures to implement an SLP, but only one out of six schools managed to implement an SLP that can be identified as ‘effective’, in line with the available empirical evidence on educational effectiveness (Kyrkiakides et al., 2010; 2016). The SLPs that schools implement are likely to foster student foundational language skills such as reading decoding and vocabulary knowledge, but they probably do not address complex language skills such as reading comprehension or writing to the same degree. The perceived complexity of promoting more complex language skills provides one explanation of this outcome and is also reported in other studies examining innovation in language education (e.g., Graham, 2019). The school management teams, which function as the policy initiators in all schools, were found to struggle with teacher autonomy. Teachers were only partly involved in the policy making process. Teachers, in turn, mainly interpret SLP as something additional to their current practice, and not something that is to replace traditional teaching practices. A final explanation for these findings corroborates the conclusion of many studies on language education policy, which is that educators’ beliefs about language and multilingualism, too, complicate the implementation process (e.g., Galdames & Gaete, 2010; Harklau et al., 2018; Johnson & Johnson, 2015).Since the schools in this study also participated in study 1, data were available on student language performances. Our results indicate that schools that are ineffective at promoting student language development (based on sixth-graders’ reading comprehension performances) did not manage to design and implement a strategic SLP – their policies mainly exist on paper, or within the rooms of policy management teams. Yet, the opposite might not be true: we could not find a one-on-one relationship between effective SLP implementation and effective student language development. Other practices and individual effective teachers seem to play a substantial role, in line with previous studies indicating that teachers are the final arbiters of policy (Hornberger & Johson, 2007; Menken & Garcia, 2010). It is important to note, however, that we cannot infer causal relationships from our data. It may also be the case that the effect of SLP on student achievement is reciprocal, with higher performances leading to a more strategic use of school autonomy (De Grauwe, 2005; Muijs et al., 2004).Overall, the first two studies confirm that increased school autonomy does not necessarily lead to school improvement, which has also been reported in other studies (e.g., Hanushek et al., 2013; Watson & Fullan, 2000). We hypothesized that SLP may require too much of a change from most schools, as traditional language teaching (with a main focus on basic skills and frontal teaching) and isolated teaching are still widely prevalent in most schools (AHOVOKS, 2019; De Smedt et al., 2016; Graham, 2019). Our findings corroborate the conclusion of May (2007, p. 402) about the implementation of whole-school literacy policies in secondary schools in New Zealand that ‘alongside knowledge of effective language and literacy instruction, schools need knowledge of sustainable change management and leadership, and of the school organization and culture’. Furthermore, our results confirm the findings of previous research conducted in multilingual school settings, which indicate that language education policies are continuously interpreted and negotiated at all levels (Shohamy, 2006; Menken & Garcia, 2010). Regardless of school language policy, individual teachers also look for ways to continuously adjust their practice to their students’ linguistic needs. In other words, effectively implemented strategic SLPs may help to improve education, but schools without a strategic SLP are not necessarily  ineffective. However, this  may become a problem when effective individual teachers leave the school (Fullan, 2015; Hargreaves & Fullan, 2012).  Conditions affecting SLP implementation: every school an SLP?For a language education policy issued by governments to lead to system-wide improvements in  language education, all schools need to be able to successfully implement the policy (Teddlie & Reynolds, 2000). However, a vast amount of studies, both in the language education policy literature and the general education policy literature indicates that policy implementation is often uneven across different schools (Honig, 2006). For instance,  the Flemish Inspectorate (2015) noticed that schools with a more socio-ethnically diverse student population more often develop an SLP than schools that are less diverse.The third study examined what role the school context plays in SLP implementation. The study adopted a mixed-method design by combining quantitative survey data collected in the 28 schools from study 1 with qualitative data collected in a subsample of six schools (study 2). The results indicate that student socio-ethnic composition indeed is a major driver for school teams to implement an SLP, revealing a deficit perspective: school teams feel the ‘need’ to implement an SLP in order to compensate for their students’ low language and literacy skills. Apart from student composition, average teacher experience and school size seemed to influence the implementation of an SLP. Of course, not all differences between schools could be explained by the school contextual factors included in this study; school cultural aspects such as teacher informal relationships and school leadership are likely to be  important predictors as well (März et al., 2018; Valckx et al., 2020).These findings nevertheless resonate with the conclusions of several other qualitative studies that investigate the implementation of innovative teaching practices and evidence-based programs, and that indicate that diversity in implementation is the norm rather than the exception (e.g., Ball et al., 2012; Johnson et al., 2018; Menken & Solarza, 2014; Liddicoat et al., 2018; Ruys et al., 2014; Samuda, Van den Branden & Bygate, 2019). This implies that some schools experience a stronger need to implement the policy (Viennet & Pont, 2017), and that for some schools, it is easier to implement an SLP than for others (cf. Van der Wildt et al., 2015). This also implies that different strategies and tailored (external) school support may be needed in order to improve language instruction in different school contexts (Honig, 2006; Hopkins, 2001). It seems  to be an illusion that merely giving schools the autonomy to tailor the policy to their local needs will be sufficient. Stakeholder experiences with SLP: what do they think is best?As SLP implementation, much like other policy implementations, is uneven across different contexts, study 4 examined what stakeholders involved in the implementation of SLP believe is necessary for SLP to contribute to student language achievement. Previous studies emphasize that language policy implementation is a multi-layered process that is constantly interpreted and appropriated as it travels through different layers of the education system, because different stakeholders bring their own beliefs and experiences into the process (Menken & Garcia, 2010; Ricento & Hornberger, 1996). The studies conducted so far have primarily examined teacher interpretations and implementation of language education policy, since teachers are the final arbiters of language policy, and mainly adopt a qualitative approach (Harklau et al., 2018; Johnson et al., 2018; Menken & Garcia, 2010; Varghese, 2008). There are, however, many more stakeholders involved in the implementation process (Burns & Köster, 2016), and qualitative results cannot be easily generalized to other contexts. This study therefore included different groups of stakeholders at different layers of the policy process (Ricento & Hornberger, 1996), and adopted a mixed-method approach.A Q-study (Brown, 1980; Lo Bianco, 2015b) was conducted in order to unravel the implicit beliefs of stakeholders regarding SLP implementation. Four viewpoints were identified: there are those who are convinced that the solution to bridge the gap between SLP intention and execution lies in engaging teachers to collectively improve their practices;  those that believe that it is necessary for school teams to follow a detailed roadmap;  those who think it is key to provide these teams with extensive support in order to substantially shift teachers’ mental models of language and language learning; and, finally, a limited number of respondents who question the overall logic of the policy.The most striking result of  our analysis is that the framing and interpretation of the policy is substantially different for actors that have the task to either guide or evaluate schools on the one hand, and actors that  have to implement the policy in their everyday practices on the other hand. While interpretations of the policy are rather uniform within stakeholder groups (Honig, 2006), and clearly reflect their professional background, no full agreement exists across groups on how best to implement an SLP, and what its function in schools should be. These differences in interpretation may provide an additional explanation of why school-based language policies are often not well implemented (cf. also Edgerton & Desimone, 2019; Levin, 2008). This outcome shows that it is important to involve different stakeholders in future policymaking (Lo Bianco, 2015a), and to intensify their mutual dialogue, as they all shape policy implementation by bringing their own experiences and beliefs (Spillane et al., 2006)." "Language productivity at work" "Timothy Colleman, Miriam Taverniers, Renata Enghels, Johanna Barddal, Peter Lauwers, Robert Hartsuiker" "Department of Linguistics, Department of Experimental psychology" "Language patterns are more or less ‘productive’, depending on their lexical scope. An interdisciplinary approach is needed to compare attested productivity, in present-day language use as well as through history, to on-line and off-line language processing, and to measure the impact of personal variables. Only in this way can one arrive at a better understanding of what productivity is." "Investigating teachers' practices in using technology for Adult Language Learning (TECH4ALL)" "Center for Applied Linguistics" "The project aims at bringing together various experts from European countries in order to exchange ideas and practices in the area of Adult Language Learning (ALL) through new technologies. The current project seeks to take advantage of the ubiquity of the new technologies in order to support also the needs of migrant language learners. The project also intends to address issues of interculturalism by involving adult language learners of diverse cultural backgrounds in an effort to cater for their linguistic needs as migrant, mature students. Specifically, the project partners intend to identify successful practices in Adult Language Learning (ALL), and to discuss the strengths and weaknesses as well as the benefits and challenges of using various technological tools in the teaching and learning process. The sharing of ideas from different contexts will result in the compilation of an online collection of best practices in ALL which will also include sample material for adult foreign language learning. The partners will also work collaboratively to evaluate the material implementation in different contexts aiming at improving practices and materials in foreign language teaching to adults of diverse origin, as well as to mature students at large. Furthermore, partners will develop an on-line guide describing how the compiled material can be. Finally, the project partners intend to create an online community which will allow learners to exchange ideas and experiences in foreign/second language learning. Overall, the project intends to bring together language teachers, trainers, ICT experts and adult learners from different countries, thus promoting multiculturalism and enhancing communication and cooperation among European citizens who share similar interests." "SRP-Groeifinanciering: Multilingual language control" "Mathieu Declerck" "Linguistics and Literary Studies" "When multilinguals use language, all known languages are activated to some degree and compete with each other, leading to cross-language interference which hinders fluent and efficient language use. Language control is the process used to minimize this cross-language interference and to allow multilinguals to select words and grammatical structures in the appropriate or intended language. This uniquely multilingual process has mainly been examined in contexts where multilinguals use two or more languages, even though multilinguals are often in a context where solely one language is used. Since the latter linguistic context is assumed to result in quantitative and qualitative differences in language control relative to the former linguistic context, in the current proposal we will focus on single language contexts to further our knowledge of language control and the influence it has on multilinguals. A first line of research focuses on the nature and influence of language control when multilinguals only use one language within a context. While we know that cross- language interference occurs in such a context, and thus that language control is necessary, very few studies have empirically investigated this possibility. A second line of research focuses on the link between language control and second language acquisition since there is very little empirical evidence into this connection. Because second language acquisition usually relies on a single language context, we will focus on this linguistic context. A final line of research focuses on the influence of language control on multilinguals’ claimed cognitive advantage in the domain of executive functions. Whereas it has often been assumed that language control is the main source of improved executive functions, few studies have investigated this claim and little to no research has focused on the influence of language control in single language contexts. The three proposed lines of research will thus significantly deepen our understanding of how the minds of multilinguals - who according to some estimates represent up to 75% of the world’s population - operate during language use, and will yield important implications for real-world language use, learning and policy." "Motoric, cognitive and linguistic writing process characteristics in Alzheimer's disease: A task validity, cross-sectional and longitudinal study." "Mariëlle Leijten" Management "Research has shown that Alzheimer (AD) patients manifest deterioration in language production. This research project aims to develop complementary screening tasks that enable researchers and clinicians to characterize the longitudinal decline of on-line written language production in AD . These aims will be addressed in three consecutive studies: 1. a task-validity study, 2. a cross-sectional study, and 3. a longitudinal study.Since findings from previous research indicate the need for a screening task that assesses multiple components of language, we will combine writing tasks that focus on motor and cognitive and linguistic aspects. The more general screening tasks (i.c., describing two pictures depicting a situation) that focus on the cognitive and linguistic aspects need to be validated first. The quality of the elicitation, and consequently of the screening tool, will benefit from a careful validation of the tasks at hand. Based on the findings of the task validity study, we will design an experimental cross-sectional study involving patients with mild cognitive impairment (MCI), mild AD and a healthy control group. The observation method used, keystroke logging, is non-intrusive, time-saving, and hence cost-reducing, when compared to other diagnostic tools. Finally, in a longitudinal study, the cross-sectional study will be replicated twice (after 6 and 12 months) in order to document the (potential) gradual cognitive decline." "Cognitive decline in Alzheimer's disease: motoric, cognitive and linguistic aspects of sentence production" "Mariëlle Leijten" "EduBROn, Management" "Research has shown that Alzheimer (AD) patients manifest deterioration in language production. The current (diagnostic) instruments for language assessment (in general, as well as in relation to dementia) largely focus on product measures, such as phrase length and number of errors. However, the processes leading to spoken and written text production provide insight into language ability as well. This research project aims to develop valid sentence production tasks that enable researchers and clinicians to adequately characterize the longitudinal decline of on-line written language production in AD. These aims will be addressed in three consecutive studies: 1. an exploratory study, 2. a cross-sectional study, and 3. a longitudinal study. Since findings from previous research indicate the need for a screening task that assesses multiple components of language, we will combine writing tasks that focus on motor (physical typing skills), cognitive (mind) and linguistic (language) aspects. Therefore, during an exploratory study we aim to develop a number of highly controlled sentence production tasks that assess a set of predefined cognitive and linguistic process aspects. Exploratory study In the exploratory study, we aim to initially develop 60 picture combination cards; groups of separate pictures that elicit a written sentence. During the experiment, participants are instructed to include the objects depicted when producing a sentence and to link them with a verb that fits the situation. The observation method used is keystroke logging via the in-house developed tool Inputlog, which is non-intrusive, time-saving, and hence cost-reducing, when compared to other tools or techniques. By using validated pictures we can easily predetermine the words participants will probably use when producing a sentence and hence monitor specific word characteristics that influence cognitive pauses. Recent explorations on related linguistic data proved the discriminatory power of determiners, nouns, verbs (and adjectives) for distinguishing between participant groups. Therefore, these word categories will be the main focus of the elicited sentences. Cross-sectional and longitudinal study Based on the findings of the exploratory study, the number of picture combination cards needed (via statistical techniques like Generalizability and Design study) will be reduced to an optimal minimum. In the experimental cross-sectional study patients with mild cognitive impairment (MCI), mild AD and a healthy control group will be involved. The participants will be asked to perform a brief copy task (motor and typing skills) and to produce sentences (cognitive and linguistic skills) based on randomly selected picture combination cards. Finally, in the longitudinal study, the cross-sectional study will be replicated after six and twelve months in order to document the (potential) cognitive decline. We will opt for mixed effects models in which the random effects of subjects, sentences and words are modeled. A secondary goal of this research project is to further automatize data analysis and report generation allowing clinicians to interpret the data instantly. Since we trigger predefined nouns, verbs and adjectives in our study, we should be able to further automatize the analysis and reporting via Inputlog." "Cognitive decline in Alzheimer's disease: motoric, cognitive and linguistic aspects of sentence production." "Mariëlle Leijten" "EduBROn, Management" "Research has shown that Alzheimer (AD) patients manifest deterioration in language production. The current (diagnostic) instruments for language assessment (in general, as well as in relation to dementia) largely focus on product measures, such as phrase length and number of errors. However, the processes leading to spoken and written text production provide insight into language ability as well. This research project aims to develop valid sentence production tasks that enable researchers and clinicians to adequately characterize the longitudinal decline of on-line written language production in AD. These aims will be addressed in three consecutive studies: 1. an exploratory study, 2. a cross-sectional study, and 3. a longitudinal study. Since findings from previous research indicate the need for a screening task that assesses multiple components of language, we will combine writing tasks that focus on motor (physical typing skills), cognitive (mind) and linguistic (language) aspects. Therefore, during an exploratory study we aim to develop a number of highly controlled sentence production tasks that assess a set of predefined cognitive and linguistic process aspects. Exploratory study In the exploratory study, we aim to initially develop 60 picture combination cards; groups of separate pictures that elicit a written sentence. During the experiment, participants are instructed to include the objects depicted when producing a sentence and to link them with a verb that fits the situation. The observation method used is keystroke logging via the in-house developed tool Inputlog, which is non-intrusive, time-saving, and hence cost-reducing, when compared to other tools or techniques. By using validated pictures we can easily predetermine the words participants will probably use when producing a sentence and hence monitor specific word characteristics that influence cognitive pauses. Recent explorations on related linguistic data proved the discriminatory power of determiners, nouns, verbs (and adjectives) for distinguishing between participant groups. Therefore, these word categories will be the main focus of the elicited sentences. Cross-sectional and longitudinal study Based on the findings of the exploratory study, the number of picture combination cards needed (via statistical techniques like Generalizability and Design study) will be reduced to an optimal minimum. In the experimental cross-sectional study patients with mild cognitive impairment (MCI), mild AD and a healthy control group will be involved. The participants will be asked to perform a brief copy task (motor and typing skills) and to produce sentences (cognitive and linguistic skills) based on randomly selected picture combination cards. Finally, in the longitudinal study, the cross-sectional study will be replicated after six and twelve months in order to document the (potential) cognitive decline. We will opt for mixed effects models in which the random effects of subjects, sentences and words are modeled. A secondary goal of this research project is to further automatize data analysis and report generation allowing clinicians to interpret the data instantly. Since we trigger predefined nouns, verbs and adjectives in our study, we should be able to further automatize the analysis and reporting via Inputlog."