Search Results (56)

This paper addresses text simplification task for Kazakh, a morphologically rich, low-resource language, by introducing KazSim, an instruction-tuned model built on multilingual large language models (LLMs). First, we develop a heuristic pipeline to identify complex Kazakh sentences, manually validating its performance on 400 examples and comparing it against a purely LLM-based selection method; we then use this pipeline to assemble a parallel corpus of 8709 complex–simple pairs via LLM augmentation. For the simplification task, we benchmark KazSim against standard Seq2Seq systems, domain-adapted Kazakh LLMs, and zero-shot instruction-following models. On an automatically constructed test set, KazSim (Llama-3.3-70B) achieves BLEU 33.50, SARI 56.38, and F1 87.56 with a length ratio of 0.98, outperforming all baselines. We also explore prompt language (English vs. Kazakh) and conduct human evaluation with three native speakers: KazSim scores 4.08 for fluency, 4.09 for meaning preservation, and 4.42 for simplicity—significantly above GPT-4o-mini. Error analysis shows that remaining failures cluster into tone change, tense change, and semantic drift, reflecting Kazakh’s agglutinative morphology and flexible syntax. Full article

Background/Objectives: Neuropsychological testing is key in defining cognitive profiles at early stages of dementia. More importantly, the detection of subtle cognitive changes, such as subjective cognitive complaints (SCCs), an understudied phenomenon, is critical for early detection and preventive interventions. Methods: This systematic review analyzes the empirical data on the cognitive domains and neuropsychological tests used in studies addressing SCC in the last 15 years (2009–2024). Results: A selection of 15 papers with exploratory, cross-sectional, and prospective scope in this field was obtained from PubMed and Embase databases. They used screening tests (17%) and a broad spectrum of neurocognitive domains. Yet, we identified three main targeted cognitive domains: executive functions (28%), language (17%), and memory (17%). Myriad assessment tools were also applied, but the most commonly used was a set of eight tests: Mini-mental Scale Examination (MMSE), Trail Making Test (A-B), Stroop test, Digit span test (DST), Semantic and Phonological fluency test, Rey Auditory Verbal Learning Test (RAVLT), Weschler Memory Scale (WMS), and Boston Naming Test (BNT). New approaches involved including the Geriatric Depression Scale (GDS) and self/informant reports. Conclusions: Despite scarce agreement in the assessment protocols, the identification of early neurocognitive symptoms to objectivate the SCC phenomenon envisions a broad field of research. Full article

L2 reading fluency and accuracy are crucial aspects of second language (L2) acquisition, as they directly influence cognitive processing and overall reading comprehension. Developing these skills enables learners to engage with texts efficiently and build confidence in their reading abilities. This study uses reading software to examine the relationship between reading fluency and accuracy in the L2 learning context. Two primary research questions were addressed: the correlation between reading fluency and accuracy and potential gender-based differences in reading performance metrics, including reading time, comprehension, and developmental progress. Pearson’s correlation analysis revealed a strong, positive relationship between fluency and accuracy, indicating that improvements in one skill are associated with enhancements in the other. Additionally, fluency and accuracy were positively correlated with reading comprehension and developmental measures, underscoring the interconnected nature of these skills. Gender comparisons, analyzed through independent t-tests, indicated no significant differences in reading software engagement or performance. This suggests the software provides equitable support across genders, enabling consistent skill development in fluency, accuracy, and comprehension. The findings highlight the critical role of technology in facilitating balanced reading instruction and offer insights into its potential to address diverse learner needs effectively. Full article

Introduction: Achieving long-term impacts from cognitive remediation (CR) interventions is a key goal in rehabilitative care. Integrating virtual reality (VR) with psychoeducational approaches within CR programs has shown promise in enhancing user engagement and addressing the complex needs of individuals with bipolar disorder (BD). A previous randomized controlled crossover feasibility trial demonstrated the viability of a fully immersive VR-CR intervention for BD, reporting low dropout rates, high acceptability, and significant cognitive improvements. This secondary analysis aimed to evaluate the stability of these outcomes over time. Methods: This paper presents a 6- to 12-month follow-up of the initial trial. Secondary cognitive outcomes were assessed, including visuospatial abilities, memory, attention, verbal fluency, and executive function, using validated assessment tools. Statistical analyses were conducted using Friedman’s test. Results: A total of 36 participants completed the 6- to 12-month follow-up. Overall, cognitive functions showed a trend toward stability or improvement over time, except for visuospatial and executive functions, which demonstrated inconsistent trajectories. Significant improvements were observed in language (p = 0.02). Conclusion: This study highlights the overall stability of cognitive functions 12 months after a fully immersive VR-CR program for individuals with BD. To sustain long-term clinical benefits, an integrated approach, such as incorporating psychoeducational strategies within cognitive remediation interventions, may be essential. Further follow-up studies with control groups and larger sample sizes are needed to validate these findings. Full article

Divergent thinking tests are popular instruments to measure a person’s creativity. They often involve scoring fluency, which refers to the count of ideas generated in response to a prompt. The two-parameter Poisson counts model (2PPCM), a generalization of the Rasch Poisson counts model (RPCM) that includes discrimination parameters, has been proposed as a useful approach to analyze fluency scores in creativity tasks, but its estimation was presented in the context of generalized structural equation modeling (GSEM) commercial software (e.g., Mplus). Here, we show how the 2PPCM (and RPCM) can be estimated in a Bayesian multilevel regression framework and interpreted using the R package brms, which provides an interface for the Stan programming language. We illustrate this using an example dataset, which contains fluency scores for three tasks and 202 participants. We discuss model specification, estimation, convergence, fit and comparisons. Furthermore, we provide instructions on plotting item response functions, comparing models, calculating overdispersion and reliability, as well as extracting factor scores. Full article

Speech disorders encompass a complex interplay of neuroanatomical, genetic, and environmental factors affecting individuals’ communication ability. This review synthesizes current insights into the neuroanatomy, genetic underpinnings, and environmental influences contributing to speech disorders. Neuroanatomical structures, such as Broca’s area, Wernicke’s area, the arcuate fasciculus, and basal ganglia, along with their connectivity, play critical roles in speech production, comprehension, and motor coordination. Advances in the understanding of intricate brain networks involved in language offer insights into typical speech development and the pathophysiology of speech disorders. Genetic studies have identified key genes involved in neural migration and synaptic connectivity, further elucidating the role of genetic mutations in speech disorders, such as stuttering and speech sound disorders. Beyond the biological mechanisms, this review explores the profound impact of psychological factors, including anxiety, depression, and neurodevelopmental conditions, on individuals with speech disorders. Psychosocial comorbidities often exacerbate speech disorders, complicating diagnosis and treatment and underscoring the need for a holistic approach to managing these conditions. Future directions point toward leveraging genetic testing, digital technologies, and personalized therapies, alongside addressing the psychosocial dimensions, to improve outcomes for individuals with speech disorders. This comprehensive overview aims to inform future research and therapeutic advancements, particularly in treating fluency disorders like stuttering. Full article

A conversational system is an artificial intelligence application designed to interact with users in natural language, providing accurate and contextually relevant responses. Building such systems for low-resource languages like Swahili presents significant challenges due to the limited availability of large-scale training datasets. This paper proposes a Retrieval-Augmented Generation-based system to address these challenges and improve the quality of Swahili conversational AI. The system leverages fine-tuning, where models are trained on available Swahili data, combined with external knowledge retrieval to enhance response accuracy and fluency. Four models—mT5, GPT-2, mBART, and GPT-Neo—were evaluated using metrics such as BLEU, METEOR, Query Performance, and inference time. Results show that Retrieval-Augmented Generation consistently outperforms fine-tuning alone, particularly in generating detailed and contextually appropriate responses. Among the tested models, mT5 with Retrieval-Augmented Generation demonstrated the best performance, achieving a BLEU score of 56.88%, a METEOR score of 72.72%, and a Query Performance score of 84.34%, while maintaining relevance and fluency. Although Retrieval-Augmented Generation introduces slightly longer response times, its ability to significantly improve response quality makes it an effective approach for Swahili conversational systems. This study highlights the potential of Retrieval-Augmented Generation to advance conversational AI for Swahili and other low-resource languages, with future work focusing on optimizing efficiency and exploring multilingual applications. Full article

The focus of the present paper is twofold; the first objective is to examine how children with dyslexia (henceforward DYS children) and typically developing children (henceforward TD children) performed in Greek (first language; L1) compared to English (second language; L2) in reading, phonological awareness (PA), rapid automatized naming (RAN), working memory (WM), and short-term memory (STM) tasks. Our second goal is to investigate DYS children’s performance compared to that of TD children in the L1 and L2 domains mentioned above. Thirty-two (DYS = 16; TD = 16) school-aged children (9;7–11;9 years old; M_age = 130.41), basic users of English (level ranging from A1 to A2), carried out a battery test in L1 and L2, respectively, including reading, PA, STM, and WM tasks. More specifically, the tasks were the following: word and nonword decoding, reading accuracy and reading fluency, word and nonword reading per minute, PA, RAN, nonword repetition, as well as forward, backward, and digit span sequencing. This is a work-in-progress study, and preliminary results reveal that DYS students exhibit important reading and memory deficits in both languages. The data analysis indicated that DYS children have particular difficulties and statistically significant differences in L1 and L2 compared to TD in all tasks. In conclusion, this is the first study, at least in Greek, which assesses both reading and memory skills of DYS children in L2. The results reveal deficits in both languages, and the overall findings contribute to theories on the transfer of difficulties of linguistic skills between L1 and L2, while memory scores also underline this co-occurrence. Future implications of this study include a combination of reading and cognitive activities in the teaching methods of English teachers to improve DYS children’s overall performance in learning English as L2. Full article

Today, Smart Assistants (SAs) are supported by significantly improved Natural Language Processing (NLP) and Natural Language Understanding (NLU) engines as well as AI-enabled decision support, enabling efficient information communication, easy appliance/device control, and seamless access to entertainment services, among others. In fact, an increasing number of modern households are being equipped with SAs, which promise to enhance user experience in the context of smart environments through verbal interaction. Currently, the market in SAs is dominated by products manufactured by technology giants that provide well designed off-the-shelf solutions. However, their simple setup and ease of use come with trade-offs, as these SAs abide by proprietary and/or closed-source architectures and offer limited functionality. Their enforced vendor lock-in does not provide (power) users with the ability to build custom conversational applications through their SAs. On the other hand, employing an open-source approach for building and deploying an SA (which comes with a significant overhead) necessitates expertise in multiple domains and fluency in the multimodal technologies used to build the envisioned applications. In this context, this paper proposes a methodology for developing and deploying conversational applications on the edge on top of an open-source software and hardware infrastructure via a multilayer architecture that simplifies low-level complexity and reduces learning overhead. The proposed approach facilitates the rapid development of applications by third-party developers, thereby enabling the establishment of a marketplace of customized applications aimed at the smart assisted living domain, among others. The supporting framework supports application developers, device owners, and ecosystem administrators in building, testing, uploading, and deploying applications, remotely controlling devices, and monitoring device performance. A demonstration of this methodology is presented and discussed focusing on health and assisted living applications for the elderly. Full article

Several studies have reported subtle differences in cognition between individuals with subjective cognitive decline (SCD) compared to those with normal cognition. This study aimed to (i) identify these differences using discrepancy scores (e.g., categorial–phonemic verbal fluency performance) derived from neuropsychological tests in three cognitive domains (memory: Wechsler’s Word List and Digits; executive functions: Stroop and verbal fluency; and language: BNT and ECCO_Senior) and (ii) determine which discrepancy scores are significant for classification. Seventy-five older adults were included: 32 who were labeled SCD+ (age 71.50 ± 5.29), meeting Jessen et al.’s criteria, and 43 in the normal cognition group (SCD−; age 69.81 ± 4.62). Both groups completed a protocol including screening and the specified neuropsychological tests. No differences were found between the groups in their age, education, episodic memory, global cognitive state, or mood. Significant differences between the groups were observed regarding the discrepancy scores derived from BNT (naming) and ECCO_Senior (sentence comprehension). These scores accurately classified participants (71.6%), with ECCO_Senior having a primary role. ROC curves indicated a poor-to-fair model quality or diagnostic accuracy (AUC__BNT = 0.690; AUC__ECCO = 0.722). In conclusion, discrepancy scores in the language domain are important for distinguishing between individuals with SCD and normal cognition, complementing previous findings in this domain. However, given their relatively poor diagnostic accuracy, they should be used with caution as part of a more detailed neuro-psychological assessment. Full article

This study examined the cognitive performance of older adults aged 60 and above using the Montreal Cognitive Assessment (MoCA) test and DigiMoCA, a digital tool for cognitive screening administered by means of a smart speaker, to investigate whether the additional variables utilised by DigiMoCA allow for the identification of significant differences between individuals with depressive symptoms and those with mild cognitive impairment, which are not detected using the original MoCA test. A total of 73 senior adults located in Northwestern Spain, 22 male and 51 female, participated in this study. Subjects were divided into four groups based on the presence of depressive symptoms and mild cognitive impairment, with the aim of analysing the results of each dimension of the MoCA and DigiMoCA tests and assessing the additional insights provided by the digital administration tool. The results indicate significant differences among groups. Individuals with depressive symptoms exhibited poorer performance in forward number span, attention, and clock drawing compared to healthy controls. Furthermore, individuals with depressive symptoms and mild cognitive impairment exhibited significantly worse memory and orientation compared to those with cognitive impairment alone. Correlations revealed that a greater severity of depressive symptoms was associated with poorer performance across cognitive domains, including visuospatial skills, attention, language, memory, and phonemic verbal fluency. This study also illustrated how the exploitation of additional variables systematically captured by digital instruments, such as completion times or response delays to individual interactions, may facilitate the early identification of cognitive and depressive conditions, providing initial evidence about the importance of integrating advanced digital tools in cognitive assessment to inspire the development of more effective, personalised interventions. Full article

In the domain of automated story generation, the intricacies of the Arabic language pose distinct challenges. This study introduces a novel methodology that moves away from conventional event-driven narrative frameworks, emphasizing the restructuring of narrative constructs through sophisticated language models. Utilizing mBERT, our approach begins by extracting key story entities. Subsequently, XLM-RoBERTa and a BERT-based linguistic evaluation model are employed to direct beam search algorithms in the replacement of these entities. Further refinement is achieved through Low-Rank Adaptation (LoRA), which fine-tunes the extensive 3 billion-parameter BLOOMZ model specifically for generating Arabic narratives. Our methodology underwent thorough testing and validation, involving individual assessments of each submodel. The ROCStories dataset provided the training ground for our story entity extractor and new entity generator, and was also used in the fine-tuning of the BLOOMZ model. Additionally, the Arabic ComVE dataset was employed to train our commonsense evaluation model. Our extensive analyses yield crucial insights into the efficacy of our approach. The story entity extractor demonstrated robust performance with an F-score of 96.62%. Our commonsense evaluator reported an accuracy of 84.3%, surpassing the previous best by 3.1%. The innovative beam search strategy effectively produced entities that were linguistically and semantically superior to those generated using baseline models. Further subjective evaluations affirm our methodology’s capability to generate high-quality Arabic stories characterized by linguistic fluency and logical coherence. Full article

Our study examined the complex relationships among reading performance (decoding, comprehension) and language, visuo-spatial, and attentional control abilities in 115 Italian-speaking children. Latent profile analysis was used to identify distinct clusters of participants showcasing quantitative differences in decoding skills, including word, pseudo-word, text reading speed and accuracy. Then, we used this classification to investigate group differences in a variety of linguistic, working memory, and visuo-spatial tasks, as well as in reading comprehension skills, by means of multivariate and univariate tests. Our results reveal significant links between reading proficiency and several key factors: language skills, visuo-spatial abilities, and attentional control. These findings illuminate the nuanced impact of domain-general processes that govern a series of linguistic and visuo-perceptive subcomponents during reading tasks. Additionally, using dominance analysis, predictors of written text comprehension were identified. Our findings suggest that effective reading comprehension relies on a synergistic interplay of adequate reading speed, attentional control, working memory, and verbal fluency, accounting for 23% of the explained variance. This study highlights the multifaceted nature of reading proficiency and suggests that a broader perspective is necessary to fully understand reading development and support its improvement. Full article

Research on the influence of chronotype and time of day (TOD) on cognitive performance, especially in children, is limited. We explored potential interactive effects, hypothesizing that performance differs when comparing preferred vs. non-preferred TOD. In total, 76 morning-type (MT = 37) or evening-type (ET = 39) children from the third and fourth grades (48.7% girls; M age = 8.05; SD age = 0.51), identified through the Children Chronotype Questionnaire, completed two 30-min neuropsychological assessment sessions via videoconference on the first (9:00) or last hour (16:00) of the school day. The protocol included neuropsychological tests targeting memory, language, and attention/executive domains. The results revealed an interactive effect of medium size between chronotype and TOD on a Rapid Alternating Stimulus (Naming) Task. MT and ET performed faster in asynchrony conditions (morning for ET; afternoon for MT). Additionally, ET outperformed MT in a Backward Digit Span Task, irrespective of TOD. TOD also influenced performance on an Alternating Verbal Fluency Task, with both MT and ET children performing better in the morning. These results underscore the importance of chronotype and TOD in children’s cognitive performance, particularly in working memory and verbal fluency. Children assessed during non-preferred TOD exhibited better performance on some cognitive tasks, challenging the assumption that optimal times always yield superior results. Full article

One important internal factor influencing reading comprehension is a child’s executive functioning. The primary objective of this paper is to evaluate the effectiveness of a cognitive stimulation program, ExeFun-READ (a program focusing on executive functioning stimulation via the L1 (Slovak language) curriculum with a specific focus on enhancing reading comprehension). The program is explicitly based on the assumed bi-directional relationship between executive functioning and language abilities related to reading comprehension. The program is domain-specific; the curriculum of L1 (Slovak) is a curricular area in which cognitive mediation occurs. The study will investigate whether the domain-specific ExeFun-READ intervention had a positive effect on children’s executive functioning and language subcomponents of reading comprehension. Keeping in mind ecological validity, ExeFun-READ was designed primarily for educational purposes, specifically professional tutoring for low-performing students. The intervention consists of 30 units; each unit lasts for 45–60 min. A stimulation unit approximates a teaching unit rather than a clinical experimental intervention. In total, 151 low-performing students attending grade four from seven elementary schools took part in the project. The study employs a pre-test–training–post-test design with three conditions: experimental, active control, and passive control. In the current study, the intervention led to improved language abilities related to reading comprehension. Significant improvements were found in vocabulary (semantic knowledge), completion of sentences (syntactic knowledge), and classification of terms (verbal fluency and inferencing) in the group of children that received the ExeFun-READ intervention. In terms of executive functioning, the improvement only extended to switching fluency. Full article