Model Of Language Research Articles

Incoherent coherence? Using systemic functional linguistics to improve oral language assessment literacy

AbstractThis paper discusses the challenges of defining coherence in the context of oral language assessment literacy and proposes that better understanding of the construct can be achieved through a systemic-functional linguistic lens. Coherence is taken to be a foundational quality of written and spoken discourse and is a standard feature in the assessment rubrics of most large-scale English-language oral proficiency tests such as IELTS, TOEFL, and Cambridge. However, the precise meaning of coherence, and how it can be identified in language, is difficult to grasp. This has implications for the makers and raters of tests as well as the candidates who take them. The literature suggests coherence is the most troublesome category to rate reliably and validly. The systemic-functional linguistic (SFL) model of language provides a rich description of coherence, but has been under-utilised by the language testing industry to date. In this paper we examine three aspects of coherence: (1) how it is currently conceptualised in oral language proficiency testing; (2) how it is conceptualised in linguistics in general and SFL in particular; and (3) how insights from SFL could inform the evaluation of coherence in oral language proficiency examinations.

The Politics of Memory: NLP Models as Liberating Artifacts

The main argument of this article is that natural language processing (NLP) projects of Indigenous African languages can be understood as an effort to restore and preserve Africa’s memory in the digital. Language has always occupied a particular place in the African decolonial imaginary. Several African states embarked on processes of Africanisation of postcolonial institutions in their respective national contexts. One of the main features of this project in many African countries was the restoration of African languages in education and public service. Taking a social constructivist approach rooted in science and technology studies and looking through the analytical lens of decoloniality in Africa, I conceptualise NLP models of Indigenous African languages as liberating artifacts. I mobilise decolonial theory to look at language as a technology that can be used to dis-member and re-member the African subjects of their memory of being as people. The coloniser erased the memory of the African subjugated by cutting off their linguistic connections and instead planted the European memory. From this perspective, decolonisation can be understood as a process of re-memberment that can be achieved through the return of the languages of those whose humanity has been denied or questioned. In this sense, NLP projects offer instances of the digital that represent spaces of restoration and negotiated digital futurities. I contend that these digital spaces in Africa are social spaces of contestation about what common investments should be made, what collective futures are desirable and what are the perceived material consequences of such shared imaginations.

The four strands: A simplified approach to SLA for CLIL

AbstractWhile it is widely accepted that language teachers require knowledge of second language acquisition (SLA) (Richards, 2010), what about content teachers, who are increasingly required to help develop their learners' academic language proficiency? How much do they need to know about SLA theories and how can they acquire such knowledge and expertise? This article proposes P. Nation's (2007) four strands framework as a straightforward and accessible model for content and language integrated learning (CLIL) teachers to address the challenge of balancing content and language instruction, a common concern within CLIL programs. The article begins by looking at two well‐known CLIL teacher competence frameworks. It then raises questions about the extent to which CLIL teachers need to master aspects of pedagogical content knowledge about SLA. The authors argue that while CLIL teachers do indeed need knowledge of how languages are learned and of SLA‐informed approaches to developing learners' second language (L2) proficiency, questions about just how much SLA knowledge CLIL teachers need and how they can be equipped with this knowledge remain unanswered. By delineating language learning opportunities into the four strands framework—meaning‐focused input, meaning‐focused output, language‐focused learning, and fluency development—the authors illustrate how this framework offers a practical guide for integrating language and content education effectively in CLIL and bilingual education contexts. The article ends with a presentation of three principles to help CLIL teachers achieve both the content and language teaching objectives of CLIL.

Introducing an Epistemological Paradigm and Its Ontological Origin through the Metaphysical Deconstruction of Language as the Model of Expression

The model of language can relatively concretely reveal the mechanism of the epistemology, which is extended by the ontology that takes “person” as the unit. It unveils an intuitable dimension to represent the epistemolog’s paradigm and limitations. Deriving from the inherent relation between epistemology and ontology, the retrospection of the epistemology’s ontological origin can be actualized by locating or grasping the epistemological first person’s ontological essence. Hence, the illusional and frail essence of the ontology behind this epistemology would be uncovered. “Person” is not a stable unit in the ontological world, since the application of this unit stimulates the development of the operating style of the epistemology which relies on the supplement of paradox. After this study that intends to get the epistemological and ontological truth of the current using nominative, the new era of one or more unknown but solid epistemology and its ontology will be gradually recognized. Furthermore, the circumstance of past and current philosophical study is going to be changed.

Exploring Meditron-7b performance on a small-scale purpose made dataset in human nutrition queries

Abstract The interest in Large Languages Models (LLMs) has been rapidly evolving over the last years, boosting concerns about their potential to improve the efficiency and effectiveness of clinical, educational and research work in medicine. In this study, we aimed to test the performance of MEDITRON, a recently released suite of open-source LLMs with 7B parameters adapted to the medical domain, on a small sample of human nutrition queries. To facilitate our assessment, we curated a specialized dataset comprising a diverse range of human nutrition-related queries, which have been manually collected from various medical databases and nutrition exams. The dataset was adapted for the scope of our analysis to encompass variations in language complexity, query types, and semantic nuances commonly encountered in real-world settings. Additionally, to ensure a standardized and clinically relevant context for our evaluations, we engineered a specialized prompt designed to mimic interactions with a highly esteemed physician specializing in nutrition, food science, and diet-related disorders. The prompt guided the generation of nutrition-related queries in a structured format, enabling MEDITRON to provide responses consistent with the latest advancements in medical research. Our preliminary findings revealed promising capabilities of MEDITRON in understanding medical language, providing contextually appropriate responses to several human nutrition related questions. Through this study, we add valuable insights to the ongoing discussion around the deployment of LLMs in public health, highlighting their potential to improve access to essential human nutrition literacy. Key messages • Design of LLMs tools holds the potential to improve healthcare in clinical, research and education applications. • Further evidence is needed to foster their actual integration in real world settings.

Идиостиль автора: вопросы параметризации

The paper presents a systematization of the parameters identified for the study of the author’s individual style, which are found in scientific works of recent years, with the aim of creating a general research model. The work presents the results of analysing dissertations from 1996 to 2024, in which authors address the analysis of the author’s individual style in terms of material, object, or subject of study. As a basis for further systematization, features such as research approach, material, understanding of individual style, and, finally, the analyzed parameters were adopted. This work is the first one to present a systematization of parameters included in the system of the author’s individual style, based on dissertation research conducted within different fields, using different approaches and diverse material, with the aim of creating a general model, which accounts for the scientific novelty. The research findings are as follows: a system of parameters has been identified and subjected to authorial classification, consisting of six levels, which includes elements of a level model of language, a pragmatic component (functionality), and the author’s worldview.

Efficient labeling of french mammogram reports with MammoBERT

Recent advances in deep learning and natural language processing (NLP) have broadened opportunities for automatic text processing in the medical field. However, the development of models for low-resource languages like French is challenged by limited datasets, often due to legal restrictions. Large-scale training of medical imaging models often requires extracting labels from radiology text reports. Current methods for report labeling primarily rely on sophisticated feature engineering based on medical domain knowledge or manual annotations by radiologists. These methods can be labor-intensive. In this work, we introduce a BERT-based approach for the efficient labeling of French mammogram image reports. Our method leverages both the expansive scale of existing rule-based systems and the precision of radiologist annotations. Our experimental results showcase the superiority of the proposed approach. It was initially fine-tuned on a limited dataset of radiologist annotations. Then, it underwent training on annotations generated by a rule-based labeler. Our findings reveal that our final model, MammoBERT, significantly outperforms the rule-based labeler while simultaneously reducing the necessity for radiologist annotations during training. This research not only advances the state of the art in medical image report labeling but also offers an efficient and effective solution for large-scale medical imaging model development.

Distributional hypothesis as isomorphism between word-word co-occurrence and analogical parallelograms.

Most of the modern natural language processing (NLP) techniques are based on the vector space models of language, in which each word is represented by a vector in a high dimensional space. One of the earliest successes was demonstrated by the four-term analogical reasoning task: what is to C as B is to A? The trained word vectors form "parallelograms" representing the quadruple of words in analogy. This discovery in NLP offers us insight into our understanding of human semantic representation of words via analogical reasoning. Despite successful applications of the large-scale language models, it has not been fully understood why such parallelograms emerge by learning through natural language data. As the vector space model is not optimized to form parallelograms, the key structure related to geometric shapes of word vectors is expected to be in the data, rather than the models. In the present article, we test our hypothesis that such parallelogram arrangement of word vectors readily exists in the co-occurrence statistics of language. Our approach focuses more on the data itself, and it is different from the existing theoretical approach trying to find the mechanism of parallelogram formation in the algorithms and/or vector arithmetic operations on word vectors. First, our analysis suggested that analogical reasoning is possible by decomposition of the bigram co-occurrence matrix. Second, we demonstrated the formation of a parallelepiped, a more structured geometric object than a parallelogram, by creating a small artificial corpus and its word vectors. With these results, we propose a refined form of distributional hypothesis pointing out an isomorphism between a sort of symmetry or exchangeability and word co-occurrence statistics.

Context-Aware Embedding Techniques for Addressing Meaning Conflation Deficiency in Morphologically Rich Languages Word Embedding: A Systematic Review and Meta Analysis

This systematic literature review aims to evaluate and synthesize the effectiveness of various embedding techniques—word embeddings, contextual word embeddings, and context-aware embeddings—in addressing Meaning Conflation Deficiency (MCD). Using the PRISMA framework, this study assesses the current state of research and provides insights into the impact of these techniques on resolving meaning conflation issues. After a thorough literature search, 403 articles on the subject were found. A thorough screening and selection process resulted in the inclusion of 25 studies in the meta-analysis. The evaluation adhered to the PRISMA principles, guaranteeing a methodical and lucid process. To estimate effect sizes and evaluate heterogeneity and publication bias among the chosen papers, meta-analytic approaches were utilized such as the tau-squared (τ2) which represents a statistical parameter used in random-effects, H-squared (H2) is a statistic used to measure heterogeneity, and I-squared (I2) quantify the degree of heterogeneity. The meta-analysis demonstrated a high degree of variation in effect sizes among the studies, with a τ2 value of 8.8724. The significant degree of heterogeneity was further emphasized by the H2 score of 8.10 and the I2 value of 87.65%. A trim and fill analysis with a beta value of 5.95, a standard error of 4.767, a Z-value (or Z-score) of 1.25 which is a statistical term used to express the number of standard deviations a data point deviates from the established mean, and a p-value (probability value) of 0.2 was performed to account for publication bias which is one statistical tool that can be used to assess the importance of hypothesis test results. The results point to a sizable impact size, but the estimates are highly unclear, as evidenced by the huge standard error and non-significant p-value. The review concludes that although contextually aware embeddings have promise in treating Meaning Conflation Deficiency, there is a great deal of variability and uncertainty in the available data. The varied findings among studies are highlighted by the large τ2, I2, and H2 values, and the trim and fill analysis show that changes in publication bias do not alter the impact size’s non-significance. To generate more trustworthy insights, future research should concentrate on enhancing methodological consistency, investigating other embedding strategies, and extending analysis across various languages and contexts. Even though the results demonstrate a significant impact size in addressing MCD through sophisticated word embedding techniques, like context-aware embeddings, there is still a great deal of variability and uncertainty because of various factors, including the different languages studied, the sizes of the corpuses, and the embedding techniques used. These differences show how future research methods must be standardized to guarantee that study results can be compared to one another. The results emphasize how crucial it is to extend the linguistic scope to more morphologically rich and low-resource languages, where MCD is especially difficult. The creation of language-specific models for low-resource languages is one way to increase performance and consistency across Natural Language Processing (NLP) applications in a practical sense. By taking these actions, we can advance our understanding of MCD more thoroughly, which will ultimately improve the performance of NLP systems in a variety of language circumstances.

Evaluation-Focused Multidimensional Score for Turkish Abstractive Text Summarization

Despite the inherent complexity of Abstractive Text Summarization, which is widely acknowledged as one of the most challenging tasks in the field of natural language processing, transformer-based models have emerged as an effective solution capable of delivering highly accurate and coherent summaries. In this study, the effectiveness of transformer-based text summarization models for Turkish language is investigated. For this purpose, we utilize BERTurk, mT5 and mBART as transformer-based encoder-decoder models. Each of the models was trained separately with MLSUM, TR-News, WikiLingua and Fırat_DS datasets. While obtaining experimental results, various optimizations were made in the summary functions of the models. Our study makes an important contribution to the limited Turkish text summarization literature by comparing the performance of different language models on existing Turkish datasets. We first evaluate ROUGE, BERTScore, FastText-based Cosine Similarity and Novelty Rate metrics separately for each model and dataset, then normalize and combine the scores we obtain to obtain a multidimensional score. We validate our innovative approach by comparing the summaries produced with the human evaluation results.

Towards SDG Digital Quality Learning: a Conceptual DLR based Model for Online Language Encoding-Skill Enhancement

Objectives: This present research has objectives to encourage digital quality learning as part of SDG quality education through proposing a DLR-based model on Zoom and WhatsApp as the main digital learning platforms for enhancing language encoding skills, i.e. the speaking and writing skills, within online learning activities. DLR refers to digital learning resources. Method: This qualitative research model is interpretatively descriptive and digitally creative. The research design consists of a system model which consists of Input, Throughput or Process, and Output, with Zoom and WhatsApp as the main platforms, and other digital platforms as digital learning resources to support the endeavor. The Zoom main platform is mainly used for observing the spoken activities, while the other WhatsApp main platform is used for observing the written interactions or communication. Results and Discussion: The novelty or the strengths of this presently proposed DLR-based model for enhancing the learners’ speaking and writing skills are on the comprehensive surveys to include the speaking and writing tests, on the learners’ online learning independence or their flexible ways or styles of online learning, on the equally sufficient opportunities for the learners’ speaking and writing activities, and on the inclusive and encouraging roles of the digitally-literate teachers. Research Implications: The presently proposed DLR-based model is aimed at enhancing the speaking and writing skills within online learning platforms for any language learners or research subjects, not necessarily high-school teachers as the previous model suggested. Originality/Value: This conceptual DLR-based model for learning language encoding skills is, therefore, promoting the world endeavor for digital quality learning, thus encouraging the quality education within the United Nations’ 17 Goals of Sustainable Development.

Optimizing hierarchical classifiers with parameter tuning and confidence scoring

Hierarchical classifiers play a crucial role in addressing complex classification tasks by breaking them down into smaller, more manageable sub-tasks. This paper continues a series of works, focused on the technical Ukrainian texts hierarchical classification, specifically the classification of repair works and spare parts used in automobile maintenance and servicing. We tackle the challenges posed by multilingual data inputs – specifically Ukrainian, Russian, and their hybrid – and the lack of standard data cleaning models for the Ukrainian language. We developed a novel classification algorithm, which employs TF-IDF victimization with unigrams and bigrams, keyword selection, and cosine similarity for classification. This paper describes a method for training and evaluating a hierarchical classification model using parameter tuning for each node in a tree structure. The training process involves initializing weights for tokens in the class tree nodes and input strings, followed by iterative parameter tuning to optimize classification accuracy. Initial weights are assigned based on predefined rules, and the iterative process adjusts these weights to achieve optimal performance. The paper also addresses the challenge of interpreting multiple confidence scores from the classification process, proposing a machine learning approach using Scikit-learn's GradientBoostingClassifier to calculate a unified confidence score. This score helps assess the classification reliability, particularly for unlabeled data, by transforming input values, generating polynomial parameters, and using logarithmic transformations and scaling. The classifier is fine-tuned using hyper parameter optimization techniques, and the final model provides a robust confidence score for classification tasks, enabling the verification and classification results optimization across large datasets. Our experimental results demonstrate significant improvements in classification performance. Overall classification accuracy nearly doubled after training, reaching 92.38 %. This research not only advances the theoretical framework of hierarchical classifiers but also provides practical solutions for processing large-scale, unlabeled datasets in the automotive industry. The developed methodology can enhance various applications, including automated customer support systems, predictive maintenance, and decision-making processes for stakeholders like insurance companies and service centers. Future work will extend this approach to more complex tasks, such as extracting and classifying information from extensive text sources like telephone call transcriptions.

ZNO-Eval: Benchmarking reasoning capabilities of large language models in Ukrainian

As the usage of large language models for problems outside of simple text understanding or generation increases, assessing their abilities and limitations becomes crucial. While significant progress has been made in this area over the last few years, most research has focused on benchmarking English, leaving other languages underexplored. This makes evaluating the reasoning and robustness level of language models in Ukrainian particularly challenging. The purpose of this work is to establish a comprehensive benchmark for the reasoning capabilities evaluation of large language models in the Ukrainian language. This paper presents the ZNO-Eval benchmark based on real exam tasks from Ukraine's standardized educational testing system: the External Independent Evaluation and the National Multi-subject Test. With single- answer options, multiple-choice, matching, and open-ended questions from diverse subjects, including Ukrainian language, mathematics, history, and geography, this dataset paves the way toward a thorough analysis of reasoning capabilities across different domains and complexities. Evaluation of several well-known language models, such as GPT-3.5-Turbo, GPT-4o, GPT-4-Turbo, Mistral Large, Claude 3 Opus, and Gemini-1.5 Pro on this benchmark demonstrated the superiority of GPT-4o in both common knowledge reasoning and intricate language tasks. At the same time, Gemini Pro and GPT-4 Turbo excelled in the arithmetic domain, leading in single-answer and open-ended math problems. While all models were close to max performance in text-only common knowledge tasks like history and geography, there still is a gap for Ukrainian language and math, thus highlighting the importance of developing specialized language benchmarks for more accurate assessments of model capabilities and limitations across different languages and contexts. This research introduced ZNO-Eval, an effective benchmark for evaluating reasoning capabilities, and thoroughly explored the abilities and limitations of modern solutions in the Ukrainian language. Future research should aim to expand the scope of ZNO-Eval to other modalities like images commonly used for exam problem description.

Proving Correctness of Parallel Implementations of Transition System Models

This article addresses the long-standing problem of program correctness for programs that describe systems of parallel executing processes. We propose a new method for proving correctness of parallel implementations of high-level models expressed as transition systems. The implementation language underlying the method is based on the concurrency model of actors and active objects. The method defines program correctness in terms of a simulation relation between the transition system that specifies the program semantics of the parallel program and the transition system that is described by the correctness specification. The simulation relation itself abstracts from the fine-grained interleaving of parallel processes by exploiting a global confluence property of the concurrency model of the implementation language considered in this article. As a proof of concept, we apply our method to the correctness of a parallel simulator of multicore memory systems.

Moral attitudes towards effort and efficiency: a comparison between American and Chinese history

In some cultures, merely exerting effort is considered virtuous, even when the effort is inefficient. Our study examines how this moral attitude towards effort (relative to efficiency) has evolved historically across two distinct sociopolitical and linguistic contexts: the People’s Republic of China and the United States, using natural language processing techniques. Specifically, two formal political corpora were used—the People’s Daily (1950–2021) and the Congressional speeches for the U.S. (1873–2011). We developed dictionaries for each concept based on pre-trained word embedding models in both languages. Moral attitudes towards effort and efficiency were calculated on a year-by-year basis as the cosine similarity between the dictionaries of these concepts and an existing dictionary of morality. We benchmarked the fluctuations of moral attitude towards inefficient effort against critical historical events in the two countries. Additional time series analysis and Granger tests revealed the association and potential directionality between the evolution of moral attitude towards inefficient effort and critical socio-cultural variables such as collectivism and cultural looseness. Our research sheds light on the historical and socio-cultural roots of moralization of effort and has implications for historical psychology research on moral attitudes.

Measuring successful conversations in couples with and without aphasia: A scoping review.

Aphasia impacts communication and relationships. While counselling is increasingly recognised as a component of the speech-language therapy role, the success of conversation partner training is typically measured in terms of communication alone. This scoping review aimed to describe how successful conversation is currently measured with couples-with and without aphasia, to inform the development of an ecologically valid measure for couples with aphasia. The scoping review was reported in alignment with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Extensions for Scoping Reviews (PRISMA-SCR). MEDLINE, EMBASE, CINAHL and PsycINFO databases were searched for conversation treatment studies for couples with and without aphasia. Patient-reported outcome measures (PROMs) were extracted from included studies and screened against the three-tier model of situated language to shortlist those that measure everyday communication. Items within shortlisted PROMs were further described using the refined International Classification of Functioning, Disability and Health linking rules. Following screening and full-text review, 46 studies were included, consisting of 24 studies conducted with couples with aphasia and 22 studies conducted with couples without aphasia. For couples with aphasia, 13 PROMs were identified that measured everyday communication. Of these, 23% were dyadic (i.e., measured from the perspectives of both members of the couple); however, they usually only appraised the communication behaviours of the person with aphasia. For couples without aphasia, eight PROMs were identified that measured everyday communication; all were dyadic and measured both attitudes and communication behaviours of both partners. Conversation relies on the interaction of two people, and success in conversation is best rated by those having the conversation. The use of PROMs is recommended as part of person and relationship-centred practice; however, there is currently no validated PROM for conversation in aphasia that considers the perspectives and behaviours of both the person with aphasia and their communication partner. The PROM items identified in this study will form the basis of future research to develop a PROM for couples' conversations in aphasia. What is already known on the subject Communication partner training (CPT) for couples, where one person has aphasia, can improve conversation skills and enhance relationships. Clinician-rated measures, such as conversation analysis, have been used to measure the outcomes of CPT. However, there is a lack of person-centred or self-rated outcome measurement instruments. that is, patient-reported outcome measures (PROMs) for couples' conversations in aphasia. What this paper adds to existing knowledge We have identified outcome measurement instruments used in conversation treatment studies for couples with and without aphasia. We found that most PROMs used in aphasia treatment studies were not dyadic, that is, they did not include the self-report of both communication partners. In contrast, the PROMs used with couples without aphasia were dyadic and contained items that measured a more complex range of both partners' emotions, behaviours and attitudes. What are the potential or actual clinical implications of this work? This study provides insights into the content and characteristics of PROMs for couples' conversation therapy and may assist clinicians in selecting outcome measurement instruments in their practice.

Language and the Cerebellum: Structural Connectivity to the Eloquent Brain.

Neurobiological models of receptive language have focused on the left-hemisphere perisylvian cortex with the assumption that the cerebellum supports peri-linguistic cognitive processes such as verbal working memory. The goal of this study was to identify language-sensitive regions of the cerebellum then map the structural connectivity profile of these regions. Functional imaging data and diffusion-weighted imaging data from the Human Connectome Project (HCP) were analyzed. We found that (a) working memory, motor activity, and language comprehension activated partially overlapping but mostly unique subregions of the cerebellum; (b) the linguistic portion of the cerebello-thalamo-cortical circuit was more extensive than the linguistic portion of the cortico-ponto-cerebellar tract; (c) there was a frontal-lobe bias in the connectivity from the cerebellum to the cerebrum; (d) there was some degree of specificity; and (e) for some cerebellar tracts, individual differences in picture identification ability covaried with fractional anisotropy metrics. These findings yield insights into the structural connectivity of the cerebellum as relates to the uniquely human process of language comprehension.

Blame-Correct Support for Receiver Properties in Recursively-Structured Actor Contracts

Actor languages model concurrency as processes that communicate through asynchronous message sends.Unfortunately, as the complexity of these systems increases, it becomes more difficult to compose and integrate their components. This is because of assumptions made by components about their communication partners which may not be upheld when they remain implicit. In this paper, we bring design-by-contract programming to actor programs through a contract system that enables expressing constraints on receiver-related properties.Expressing properties about the expected receiver of a message, and about this receiver's communication behavior, requires two novel types of contracts.Through their recursive structure, these contracts can govern entire communication chains. We implement the contract system for an actor extension of Scheme, describe it formally, and show how to assign blame in case of a contract violation.Finally, we prove our contract system and its blame assignment correct by formulating and proving a blame correctness theorem.

Building a Speech Dataset and Recognition Model for the Minority Tu Language

Speech recognition technology has many applications in our daily life. However, for many low-resource languages without written forms, acquiring sufficient training data remains a significant challenge for building accurate ASR models. The Tu language, spoken by an ethnic minority group in Qinghai Province in China, is one such example. Due to the lack of written records and the great diversity in regional pronunciations, there has been little previous research on Tu-language speech recognition. This work seeks to address this research gap by creating the first speech dataset for the Tu language spoken in Huzhu County, Qinghai. We first formulated the relevant pronunciation rules for the Tu language based on linguistic analysis. Then, we constructed a new speech corpus, named HZ-TuDs, through targeted data collection and annotation. Based on the HZ-TuDs dataset, we designed several baseline sequence-to-sequence deep neural models for end-to-end Tu-language speech recognition. Additionally, we proposed a novel SA-conformer model, which combines convolutional and channel attention modules to better extract speech features. Experiments showed that our proposed SA-conformer model can significantly reduce the character error rate from 23% to 12%, effectively improving the accuracy of Tu language recognition compared to previous approaches. This demonstrates the effectiveness of our dataset construction and model design efforts in advancing speech recognition technology for this low-resource minority language.

Prior Context and Individual Alpha Frequency Influence Predictive Processing during Language Comprehension.

The extent to which the brain predicts upcoming information during language processing remains controversial. To shed light on this debate, the present study reanalyzed Nieuwland and colleagues' (2018) [Nieuwland, M. S., Politzer-Ahles, S., Heyselaar, E., Segaert, K., Darley, E., Kazanina, N., et al. Large-scale replication study reveals a limit on probabilistic prediction in language comprehension. eLife, 7, e33468, 2018] replication of DeLong and colleagues (2015) [DeLong, K. A., Urbach, T. P., & Kutas, M. Probabilistic word pre-activation during language comprehension inferred from electrical brain activity. Nature Neuroscience, 8, 1117-1121, 2005]. Participants (n = 356) viewed sentences containing articles and nouns of varying predictability, while their EEG was recorded. We measured ERPs preceding the critical words (namely, the semantic prediction potential), in conjunction with postword N400 patterns and individual neural metrics. ERP activity was compared with two measures of word predictability: cloze probability and lexical surprisal. In contrast to prior literature, semantic prediction potential amplitudes did not increase as cloze probability increased, suggesting that the component may not reflect prediction during natural language processing. Initial N400 results at the article provided evidence against phonological prediction in language, in line with Nieuwland and colleagues' findings. Strikingly, however, when the surprisal of the prior words in the sentence was included in the analysis, increases in article surprisal were associated with increased N400 amplitudes, consistent with prediction accounts. This relationship between surprisal and N400 amplitude was not observed when the surprisal of the two prior words was low, suggesting that expectation violations at the article may be overlooked under highly predictable conditions. Individual alpha frequency also modulated the relationship between article surprisal and the N400, emphasizing the importance of individual neural factors for prediction. The present study extends upon existing neurocognitive models of language and prediction more generally, by illuminating the flexible and subject-specific nature of predictive processing.