Universal Language Research Articles

ULMFiT: Universal Language Model Fine-Tuning for Text Classification

While inductive transfer learning has revolutionized computer vision, current approaches to natural language processing still need training from the ground up and task-specific adjustments. As a powerful transfer learning approach applicable to any NLP activity, we provide Universal Language Model Fine-tuning (ULMFiT) and outline essential strategies for language model fine-tuning. With an error reduction of 18–24% on most datasets, our technique considerably surpasses the state-of-the-art on six text categorization tasks. Additionally, it achieves the same level of performance as training on 100 times more data with only 100 annotated examples. We have made our pretrained models and code publicly available.

Navigating complexity: an evaluation framework for assessing co-creation in public health

Abstract Background Co-creation aims to democratize research and bridge the gap between research and practice. Key characteristics like empowerment and meaningful participation shape sustainable outcomes. However, existing frameworks like Arnstein’s Ladder of Citizen Participation broadly overlook individual method contributions. This study aims to address this gap by creating and validating a framework for intrinsic evaluation, guiding stakeholders and researchers in genuine co-creation of public health interventions. Methods We systematically screened and extracted frameworks aligned with five key co-creation characteristics to identify and evaluate methods. Through iterative testing, we identified key features for distinguishing methods and integrated them into a novel framework. Co-creation experts validated and refined this framework using the closed card sort method, qualitative assessment, and group discussions. Results The Co-Creation Rainbow framework includes key features from 20 existing frameworks, with an Individual-to-Collective Continuum, five sections (Informing, Understanding, Stimulating, Collaborating, and Collective Decision-making), and three overarching themes (Engage, Participate, and Empower). Analysis showed successful categorization of 436 methods and revealed a significant disparity in co-creation practices between researchers and practitioners. Conclusions The Co-Creation Rainbow framework provides a structured approach to evaluating and reporting co-creation processes. Acknowledging co-creation’s diversity, the framework aids in tailoring approaches for meaningful participation. By offering a unified language for diverse co-creation methods, the framework also facilitates transparent evaluation, enhancing the reliability and reproducibility of co-created public health interventions. Key messages • Employing a universal language across diverse co-creation methods is pivotal in fostering transparent evaluation, thereby bolstering the reliability of co-created public health interventions. • Effective collaboration of key stakeholders necessitates the use of suitable methods. By integrating diverse perspectives, public health interventions gain relevance, efficacy, and sustainability.

Progress Toward Estimating the Minimal Clinically Important Difference of Intelligibility: A Crowdsourced Perceptual Experiment.

The purpose of the current study was to estimate the minimal clinically important difference (MCID) of sentence intelligibility in control speakers and in speakers with dysarthria due to multiple sclerosis (MS) and Parkinson's disease (PD). Sixteen control speakers, 16 speakers with MS, and 16 speakers with PD were audio-recorded reading aloud sentences in habitual, clear, fast, loud, and slow speaking conditions. Two hundred forty nonexpert crowdsourced listeners heard paired conditions of the same sentence content from a speaker and indicated if one condition was more understandable than another. Listeners then used the Global Ratings of Change (GROC) Scale to indicate how much more understandable that condition was than the other. Listener ratings were compared with objective intelligibility scores obtained previously via orthographic transcriptions from nonexpert listeners. Receiver operating characteristic (ROC) curves and average magnitude of intelligibility difference per level of the GROC Scale were evaluated to determine the sensitivity, specificity, and accuracy of potential cutoff scores in intelligibility for establishing thresholds of important change. MCIDs derived from the ROC curves were invalid. However, the average magnitude of intelligibility difference derived valid and useful thresholds. The MCID of intelligibility was determined to be about 7% for a small amount of difference and about 15% for a large amount of difference. This work demonstrates the feasibility of the novel experimental paradigm for collecting crowdsourced perceptual data to estimate MCIDs. Results provide empirical evidence that clinical tools for the perception of intelligibility by nonexpert listeners could consist of three categories, which emerged from the data ("no difference," "a little bit of difference," "a lot of difference"). The current work is a critical step toward development of a universal language with which to evaluate changes in intelligibility as a result of neurological injury, disease progression, and speech-language therapy.

Algorithm-Driven Robotic Discovery of Polyoxometalate-Scaffolding Metal-Organic Frameworks.

The experimental exploration of the chemical space of crystalline materials, especially metal-organic frameworks (MOFs), requires multiparameter control of a large set of reactions, which is unavoidably time-consuming and labor-intensive when performed manually. To accelerate the rate of material discovery while maintaining high reproducibility, we developed a machine learning algorithm integrated with a robotic synthesis platform for closed-loop exploration of the chemical space for polyoxometalate-scaffolding metal-organic frameworks (POMOFs). The eXtreme Gradient Boosting (XGBoost) model was optimized by using updating data obtained from the uncertainty feedback experiments and a multiclass classification extension based on the POMOF classification from their chemical constitution. The digital signatures for the robotic synthesis of POMOFs were represented by the universal chemical description language (χDL) to precisely record the synthetic steps and enhance the reproducibility. Nine novel POMOFs including one with mixed ligands derived from individual ligands through the imidization reaction of POM amine derivatives with various aldehydes have been discovered with a good repeatability. In addition, chemical space maps were plotted based on the XGBoost models whose F1 scores are above 0.8. Furthermore, the electrochemical properties of the synthesized POMOFs indicate superior electron transfer compared to the molecular POMs and the direct effect of the ratio of Zn, the type of ligands used, and the topology structures in POMOFs for modulating electron transfer abilities.

Speaking Competence of Grade 11 Students in using English as Second Language

English language is accepted as the universal language. This makes speaking competence in the English language a requisite to almost all professions. Speaking competence in the second language is one of the areas needing improvement amongst senior high school students in the Philippines. This study aimed to investigate the influence of some personal factors, speaking anxiety and linguistic competence to student’s speaking competence in the second language. The Second Language Classroom Anxiety Scale was used to measure student’s anxiety towards the use of English as a second language. Job interview simulation is also used involving selected Grade 11 students. Results showed that majority of the students are specialized in non-STEM strand whose parents are not college degree holders. These students demonstrated an average level of speaking anxiety reflecting the tendency to show low self- confidence when asked to speak using the English language yet unprepared for the conversation. Same students also showed remarkably low levels of linguistic competence in terms of grammar, comprehension, and vocabulary. Further analysis showed that linguistic competence significantly influenced student’s speaking competence in the job interview. Findings were used as bases for the development of an intervention program.

ЯППИ – универсальный язык программирования миссий АНПА

This article proposes a universal language for developing autonomous underwater vehicle (AUV) missions. The main features of the proposed language are: the presence of high-level commands to describe the main tasks of underwater research, the use of commands in Russian, the readability of the created missions and their understandability for personnel involved in the use of underwater robotics, the lack of strict typing and parameterization of commands, the possibility of expanding the language by creating new commands based on the old ones. The article describes the main parameters of the language, syntax, semantics and basic commands, as well as language tools - a parameter refinement module, which interactively allows you to resolve uncertainty arising due to non-strict parameterization, low-level command module that implements more complex commands based on basic ones, a language extension module responsible for adding new commands, and a translator. The language translator allows you to translate created missions into task program codes for various AUVs. The created language is called URPL – Underwater Research Programming Language and at the current stage of research it is being tested and debugged on the AUVs of the IPMT FEB RAS, taking into account the experience accumulated at the institute.

Implementasi Penggunaan Media Canva dalam Pembelajaran Kurikulum Merdeka di SMA Pendidikan Keluarga Pematangsiantar

As one aspect of the Tri Dharma of Higher Education, community service activities are an obligation that must be carried out by every lecturer in higher education. English is a universal language because it is used by most countries in the world as the main language. Apart from that, English is an international language that is important to master or learn. The low use of digital media in independent curriculum learning, especially the use of Canva digital content, is influenced by internal and external factors. This activity aims to see whether the Pematang Siantar Family Education High School teachers can use the Canva application well in making Independent Curriculum teaching materials and whether the teachers are satisfied with the training provided by the Service Team. From the results of the training it can be concluded that the Canva application can help teachers in creating teaching materials based on the Independent Curriculum and almost all teachers (90.9%) who received training in using the Canva application were very satisfied with the material they received.

The Volapük Qur’an: language, scripture, and nineteenth-century German universalist provincialism

ABSTRACT This article focuses on the translation of selected segments from the Qur’an into the auxiliary ‘world language’ Volapük that was published by Johann Martin Schleyer, the inventor of the language, in Constance, Germany, in 1890. I investigate the place of this translation within Schleyer’s vision of creating a world language, and use it as a case study through which to explore the ways that literature was framed and recreated within the relatively short-lived endeavour to promote Volapük as the new universal language. Analysing the impact that Schleyer’s German Catholic background had on his presentation of the Qur’an, this article sheds light on the provincialism of this specific brand of nineteenth-century European universalism that sparked the construction of Volapük, especially when contrasted with a later project of translating the Qur’an into Esperanto. These findings will allow us to situate the Volapük translation of the Qur’an within the broader field of Qur’an translations that have primarily emerged from language activism, as opposed to a missionary or academic engagement with the Qur’an.

Сопоставительный анализ категории обращения в русском и японском языках (новая редакция)

This article undertakes a comparative analysis of address forms in the Russian and Japanese languages. As address constitutes a linguistic universality, numerous commonalities in speech functioning, syntactic peculiarities, as well as phonetic, morphological, and semantic developments of linguistic units used in vocative positions were discovered in Russian and Japanese. However, some significant differences were identified. For instance, address forms in Japanese, to a greater extent than in Russian, are associated with expressing social hierarchy relations. Consequently, Japanese speakers tend to avoid using address forms when communicating with unfamiliar people (whose social status is unclear). In familial interactions, juniors are addressed by their names, while elders are referred to using kinship terms. In business sphere, superiors are addressed by their titles, while inferiors are addressed by their surnames with added necessary politeness suffixes. In the Russian language, a wide range of figurative and metonymic means are employed as address forms, and their list is not limited. There even exist specific grammatical constructions characteristic of the vocative position. In the Japanese language, the freedom in employing vocative means, akin to Russian, is only present in poetic style. In colloquial Japanese, the utilization of figurative and metonymic means is restricted.

Language Statistics at Different Spatial, Temporal, and Grammatical Scales.

In recent decades, the field of statistical linguistics has made significant strides, which have been fueled by the availability of data. Leveraging Twitter data, this paper explores the English and Spanish languages, investigating their rank diversity across different scales: temporal intervals (ranging from 3 to 96 h), spatial radii (spanning 3 km to over 3000 km), and grammatical word ngrams (ranging from 1-grams to 5-grams). The analysis focuses on word ngrams, examining a time period of 1 year (2014) and eight different countries. Our findings highlight the relevance of all three scales with the most substantial changes observed at the grammatical level. Specifically, at the monogram level, rank diversity curves exhibit remarkable similarity across languages, countries, and temporal or spatial scales. However, as the grammatical scale expands, variations in rank diversity become more pronounced and influenced by temporal, spatial, linguistic, and national factors. Additionally, we investigate the statistical characteristics of Twitter-specific tokens, including emojis, hashtags, and user mentions, revealing a sigmoid pattern in their rank diversity function. These insights contribute to quantifying universal language statistics while also identifying potential sources of variation.

CyberUnits Bricks: An Implementation Study of a Class Library for Simulating Nonlinear Biological Feedback Loops

Feedback loops and other types of information processing structures play a pivotal role in maintaining the internal milieu of living organisms. Although methods of biomedical cybernetics and systems biology help to translate between the structure and function of processing structures, computer simulations are necessary for studying nonlinear systems and the full range of dynamic responses of feedback control systems. Currently, available approaches for modelling and simulation comprise basically domain-specific environments, toolkits for computer algebra systems and custom software written in universal programming languages for a specific purpose, respectively. All of these approaches are faced with certain weaknesses. We therefore developed a cross-platform class library that provides versatile building bricks for writing computer simulations in a universal programming language (CyberUnits Bricks). It supports the definition of models, the simulative analysis of linear and nonlinear systems in the time and frequency domain and the plotting of block diagrams. We compared several programming languages that are commonly used in biomedical research (S in the R implementation and Python) or that are optimized for speed (Swift, C++ and Object Pascal). In benchmarking experiments with two prototypical feedback loops, we found the implementations in Object Pascal to deliver the fastest results. CyberUnits Bricks is available as open-source software that has been optimised for Embarcadero Delphi and the Lazarus IDE for Free Pascal.

Translation, cross-cultural adaptation and validation of the universal Welch Emotional Connection Screen using primary and bilingual Spanish-speaking coders of videotaped mother-child interactions

IntroductionUsing clear explicit translatable language, we translated the Welch Emotional Connection Screen into a new universal language instrument, the English uWECS. In this study, we had two aims: Aim 1 was to establish concurrent validity of the uWECS by comparing scores coded by primary Spanish-speaking coders using the Spanish translation of the uWECS to scores coded by bilingual, secondary Spanish-speaking coders using the oWECS. Aim 2 was to establish the criterion-related validity in terms of oWECS and uWECS performance in tracking change in autonomic emotional connection (AEC) during the course of an intervention among preschool aged children.MethodsWe created a library of 52 five-minute Spanish-speaking mother-child videos that were collected during a randomized controlled trial of Mother-Child Emotional Preparation intervention (MCEP). The videos were collected at two time points, at enrollment and at a 6-month follow-up. The subsample of Primary Spanish-Speaking dyads from the MCEP study were coded by two independent teams of coders. We trained primary English-speaking (bilingual Spanish) coders on the oWECS, using the original training program. A different team of primary Spanish-speaking coders coded the same cases using the novel uWECS guide and trained briefly for reliability with the Spanish uWECS translation materials.ResultsWe found that the Spanish oWECS and Spanish uWECS ratings from the baseline and 6-month follow-up observations were robustly correlated, with intraclass correlations ranging from .81 to .84 and all p-values<.001, thus demonstrating sound concurrent validity for the uWECS. The oWECS and uWECS scores also achieved parallel results when evaluating the efficacy of the MCEP for primary Spanish-speaking dyads. Both the AEC scores of the oWECS [F(1, 27) = 4.31, p < .05] and the scores of the uWECS [F(1,27) = 4.06, p < .05] similarly demonstrated significant change post intervention, thus demonstrating sound criterion-related validity of the uWECS.DiscussionThese findings demonstrate that the uWECS can be used to measure parent/child AEC in linguistically diverse populations and cultures.

Development of interactive English e-learning video entertainment teaching environment based on virtual reality and game teaching emotion analysis

With the rapid development of information technology, virtual reality (VR) and game teaching have gradually become an important tool in modern education. As a universal language, the innovation of English teaching methods is particularly important. This study aims to develop an interactive English e-learning video entertainment teaching environment based on virtual reality and game teaching to enhance students’ learning interest and effect. By combining virtual reality technology with game teaching method, this paper designs and develops a set of interactive English e-learning video entertainment teaching platform. The platform uses emotion analysis technology to monitor students’ emotional state in real time and dynamically adjust teaching content according to emotional feedback. The results showed that the experimental group of students in English learning performance is significantly better than the control group, learning interest and participation is also significantly improved. Emotion analysis data showed that students in the experimental group showed more positive emotions when using the platform, and their learning pressure was significantly reduced. The interactive English e-learning video entertainment teaching environment based on virtual reality and game teaching not only improves the learning effect of students, but also significantly improves their learning interest and participation.

Linguistics for Translators

What makes a translator truly exceptional? Is it their grasp of both source and target languages, or perhaps their sensitivity to cultural nuances? Linguistics for Translators by Ali Almanna and Juliane House delves into these questions with the ambition to bridge the gap between linguistic theory and translation practice. This ambitious textbook does not just present linguistic concepts but intricately weaves them into the fabric of translation studies, making it an indispensable resource “for a diverse audience, including students of translation and linguistics, practising translators, and readers from neighbouring disciplines who are interested in the subject” (Alamnaa and House, 2024, p.4). Almanna and House (2024) begin with a captivating premise: understanding the intricate mechanisms of language is fundamental for any translator. They set the stage by discussing why linguistic knowledge is crucial, not just as an academic exercise, but as a practical toolkit for enhancing translation quality. “Our aim with Linguistics for Translators is to emphasize that the linguistic dimension of translation cannot be overlooked,” Almanna and House (p.1) assert, highlighting the book’s relevance from the outset. This approach aligns with the views of Hatim and Munday (2004), who also emphasize the significance of linguistic competence in translation. The book’s twelve chapters cover a broad spectrum of linguistic areas, each crucial for the aspiring translator. Each chapter is meticulously designed to build on the previous one, from the basics of phonetics and phonology to the complexities of cognitive linguistics. The authors provide an engaging overview of linguistics, linking it directly to translation through practical exercises and discussion points. They introduce linguistic universals and translation universals, sparking curiosity about their application in real-world translation. This discussion echoes the foundational work of Catford (1965), who first articulated the need for linguistic theory in translation.

The Traditional Intramedullary Axis Underestimates the Medial Tibial Slope Compared to Transmalleolar Sagittal Axis in Image-based Robotic-Assisted Unicompartimental Knee Arthroplasty.

The medial unicompartmental knee arthroplasty (mUKA) has been recognized as an excellent treatment for medial knee osteoarthritis. The posterior tibial slope (PTS) is measured radiographically with the intramedullary axis (IMA) to the tibial baseplate on the sagittal plane radiograph. However, in most computer-navigated or robotic mUKAs, the PTS is set from a transmalleolar axis (TMA).The PTS difference was evaluatedbetween the sagittal TMA and the sagittal IMA of patients undergoing a CT-based primary robotic-assisted mUKA.We retrospectively reviewed the preoperative computed tomography (CT) scans taken according to the MAKO system protocol (Stryker) of 67 patients undergoing mUKAs. We measured the angular difference between the IMA and the TMA in the sagittal plane.Using the TMA to set the PTS the estimation of the slope of the medial tibial plateau would increase by an average of 1.9 ± 3.2 degreescompared to the IMA. Furthermore, in nineknees, PTS was decreased.Tibial components implanted with the help of a CT scan-based preoperative planning MAKO will show an average of 1.9 degrees more than those measured on sagittal radiographs potentially of concern for knee kinematics. A universal language is needed to standardize the slope calculation and the respective reference axis used.

How to Bake a Quantum Π

We construct a computationally universal quantum programming language Quantum Π from two copies of Π, the internal language of rig groupoids. The first step constructs a pure (measurement-free) term language by interpreting each copy of Π in a generalisation of the category Unitary in which every morphism is “rotated” by a particular angle, and the two copies are amalgamated using a free categorical construction expressed as a computational effect. The amalgamated language only exhibits quantum behaviour for specific values of the rotation angles, a property which is enforced by imposing a small number of equations on the resulting category. The second step in the construction introduces measurements by layering an additional computational effect.

Membrane Physiology Symposium April 22nd-23rd, 2024, Napa California, USA.

The Membrane Physiology Symposium was created with the goal of joining basic research with technology companies, where questions and conversations are open and welcomed in a universal language. For many years, academic physiology research areas have been naturally siloed into their own niche communities, which can surely be beneficial. Linking different technological application areas with varied research sectors is an integral formula for successful scientific breakthroughs. The meeting covers a wide variety of topics related to channelopathies, neurological and cardiac disease, drug development, and therapeutic applications, with research programs represented by core academic facilities, medical science institutions, small and large pharmaceutical enterprises, as well as novel cell-based and reagent providers. For this reason, gathering the brightest minds of all relevant fields in one integrative forum is essential for new avenues of discovery, development, and process optimization to occur.

Language bias, not knowledge shortfall, underestimates the evidence of megafauna kill sites in South America

The Megafauna Kill Sites (MKS), with irrefutable empirical evidence of human-megafauna interactions over the late Quaternary, have widely been recognized in North America and historically overlooked in South America. Here, we systematically reviewed the MKS from the Americas and showed that the neglected evidence from South American MKS is as robust as or even more robust, in both quantity and quality, than the evidence from North America. However, the South American studies are often limited to low-reach publications (gray literature or local journals) written in Latin languages, which are less cited than English-written publications. Our findings clearly indicate that the neglect of South American MKS is due to its limited scientific impact caused by language bias, and not due to the lack of robust knowledge, inadequate data, or questionable evidence. To reverse this context and put the rich evidence of South American MKS on an international audience, we strongly recommend that South American researchers prioritize publishing their studies in English, the universal language of science since sec. XX. We also suggest making all archaeological evidence readily available in online databases, such as the KillSite database (www.killsitedatabase.com). We believe our recommendations align with the current scope of science, which predominantly operates in English and benefits from the widespread accessibility provided by the internet, and will help South American publications reach international audiences interested in human-megafauna interaction and animal exploitation.

Learning Transfer and Universal Grammar: The Case of Arabic as a Foreign Language

This Arabic as a Foreign Language (AFL) is considered a parsimonious field in applied linguistics. Therefore, this article tries to highlight how language transfer influences American students in their Arabic learning process. Additionally, this work attempts to explore whether Arabic foreign learners rely on their pre-existing knowledge to learn Arabic and determine the role of transfer regarding language universals. The study employs a quantitative methodology based on a Grammaticality Judgment Task (GJT) to achieve its objective. According to the study's results, L2 Arabic learners initially transfer their L1 setting to L2. In other words, Universal Grammar (UG) exists, but learners only have indirect access to it via their L1. By studying and investigating the influence of language transfer on learning outcomes, experts' and educators’ interventions can be tailored to more effectively assist foreign students in overcoming errors connected to their first language and improving their proficiency in learning Arabic language.

Enhancing Music Mood Recognition with LLMs and Audio Signal Processing: A Multimodal Approach

Abstract: Music Mood Recognition aims to allow computers to understand the emotions behind music the way humans do, in order to facilitate better perception of media by computers to aid in enhanced services like music recommendations, therapeutic interventions, and Human Computer Interaction. In this paper, we propose a novel approach to improving Music Mood Recognition using a multi-modal model that uses lyrical and audio features of a song. Lyrical features are analysed using stateof-the-art open-source Large Language Models like Microsoft Phi-3 to classify lyrics from one of the four possible emotion categories as per the James Russel Circumplex Model. Audio features are used to train a Deep Learning (ConvNet) model to predict emotion classes. A multimodal combiner model with Audio and Lyrics is then trained and deployed to enable accurate predictions. The dataset used in this research is “MoodyLyrics”, a collection of 2000+ songs classified with one of 4 possible emotion classes as per the James Russel Circumplex Model. Due to compute limitations, we are using a balanced set of 1000 songs to train and test our models. The workin this paper outperforms most other multimodal researches by allowing higher accuracies with universal language support