Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
In contrast to rational choice theory predicting humans to optimize expected utilities of choices, humans deviate from rational behavior in decision-making paradigms. Hewig et al. (2011) explored affective correlates of decision-making in the ultimatum game (UG) and the dictator game (DG). They found that feedback-related negativity (FRN), subjective valence ratings, and autonomic nervous system activity predicted rejection of monetary offers. This registered replication aimed to validate and extend these findings. Although behavioral patterns and results of subjective ratings closely matched the original study, not all psychophysiological effects were successfully replicated. Firstly, we could not replicate the reported effects of autonomic nervous system activity. Secondly, a quadratic instead of the originally proposed linear relation between the offer and the FRN emerged, possibly driven by the offer evaluation in economic games and the rewarding anticipation of successful punishment for low offers. Thirdly, P3 amplitudes mirrored the quadratic offer response pattern, generally peaking for the lowest offer. In contrast to the original study, P3 responses were larger in the UG compared with the DG. Finally, our findings indicate that participant-related higher midfrontal theta activation predicted lower acceptance behavior in the UG, with a systematic dampening effect for fairer offers. This highlights cognitive control as a crucial mechanism in economic decision-making to overcome behavioral defaults. Overall, our results conceptually support the original conclusion that decision-making in economic games is non-rational and dependent on the objective situation as well as emotional and neural markers, though not precisely as suggested by Hewig et al. (2011).
Abstract According to UNESCO. 2020. World atlas of languages, 3rd edn. United Nations Educational, Scientific and Cultural Organization. Available at: https://en.wal.unesco.org, the Tongan language is classified as ‘potentially vulnerable.’ This study aimed to identify a core Tongan vocabulary and develop a web-searchable database to support the preservation and accessibility of the language. Due to the lack of a large-scale corpus for calculating lexical frequencies, we utilized vocabulary from the Students’ English-Tongan and Tongan-English Dictionary (Thompson, Richard H. & ‘Ofa Thompson. 1992. The students’ English-Tongan and Tongan-English dictionary. Tonga: Faletohi ‘Otumotu Angl’ofa’), a resource commonly used in Tongan schools. This dictionary compiles words used in both Tongan and English in everyday life. By examining processing speed and accuracy rates in lexical decision tasks, we identified a core Tongan vocabulary. Lexical decision tasks conducted on 4,013 words from the dictionary demonstrated a mean accuracy of 95.40 % and a rapid mean reaction time of 575 ms. A comparative analysis with an equal number of high-frequency words in English (Balota, David A., Melvin J. Yap, Michael J. Cortese, Keith A. Hutchison, Brett Kessler, Bjorn Loftis, James H. Neely, Douglas L. Nelson, Greg B. Simpson & Rebecca Treiman. 2007. The English lexicon project. Behavior Research Methods 39. 445–459) and French (Ferrand, Ludovic, Boris New, Marc Brysbaert, Emmanuel Keuleers, Patrick Bonin, Alain Méot, Maria Augustinova & Christophe Pallier. 2010. The French Lexicon Project: Lexical decision data for 38,840 French words and 38,840 pseudowords. Behavior Research Methods 42. 488–496) revealed that Tongan words were processed more quickly, though with an average accuracy approximately 2 % lower than the English and French words. Despite this slight difference in accuracy, the overall comprehension level for Tongan words remained high, with accuracy exceeding 95 %. Given the faster processing speeds observed, we propose that these words represent a core vocabulary in Tongan. To facilitate access to this vocabulary, we developed a web-based search engine ( https://tonganlex.vercel.app/ ), which provides detailed information on letter (grapheme) count, phonemes, moras, parts of speech, meanings, as well as lexical decision times and accuracies. This study contributes to the accessibility of the Tongan language by making essential lexical information readily available online.
The objectives of this study are: discover the linguistic features in the opening strategy, closing strategy, and directness of the head-acts in the request letters; identify and analyze positive and negative politeness strategies used in request letters sent by students, organizations, and companies to the office of the Dean of Student Affairs; investigate the use of politeness strategies that indicates about linguistic conventions. This study employed a mixed-methods research design, incorporating both quantitative and qualitative analyses. The primary data for this study consists of request letters from students at St. Vincent's College Incorporated and various organizations within and outside the academic institution. The researcher selected 15 request letters submitted to the Dean of Student Affairs Office during the 2021-2022 academic year. This study employed three tiers of corpus analysis. The research reveals that request letters sent to the Dean of Student Affairs Office at the institution employ a diverse range of politeness strategies. The "other complimentary close and other pre-closing" is another element of the closure strategy that the researcher discovered. The head-acts shown that request letters are often direct on a degree of directness. The utilization of politeness strategies within a broader socio-cultural context illuminates established linguistic norms. The diverse inclinations towards employing politeness strategies and their multifaceted applications in request letter communication do not necessarily indicate a transition from impoliteness to politeness or vice versa. It demonstrates that people's communication styles and methods of politeness are evolving.
This paper investigates the rapidly advancing domain of Large Language Models (LLMs) and their growing potential in various fields. A central focus is the exploration of LLMs, e.g., LLaMA, as powerful tools for modeling and representing linguistic information, especially in the realm of syntax. We aim to evaluate the ability of these models to encode syntactic information, especially when explicitly supplied, through fine-tuning processes. Traditionally, Dependency Parsing has relied on specific techniques and dedicated architectures. Our research shifts this approach, conceptualizing it as a sequence-to-sequence task where Language Models interpret and transform syntax into bracketed structures that reflect dependency graphs. We introduce U-DepPLLaMA (Universal Dependency Parsing via auto-regressive LLMs based on LLaMA), a novel architecture optimized for multilingual, end-to-end Dependency Parsing. Our experimental evaluation, across 50 datasets in 26 languages from the Universal Dependency Treebank, shows that LLMs can be effectively trained for dependency parsing without the need for task-specific architectures. The results are on par with current state-of-the-art methods and demonstrate resilience across varying sentence complexities and lengths.
Abstract In this explanatory study, we aim to determine the predictors of repetition avoidance or reproduction in the translation of reporting verbs from English into Italian and Polish, using a sample of 11 novels. We fit multiple negative binomial regression models with fixed effects to assess the impact of five predictor variables (i.e. frequency of a source-text verb, number of its translation equivalents in lexical databases, its number of senses, semantic-pragmatic category of verb, and length in characters) on the response variable, i.e., the number of different target language verb types into which a source text reporting verb is translated. The overall model fit per the lowest Akaike Information Criterion (AIC) value obtained through backward elimination reveals that verb category, frequency of a source-text reporting verb, and number of translation equivalents make the largest individual contributions to explaining the proportion of variation in the response variable in the Italian data; for the Polish translations the corresponding variables are verb category, frequency, the number of senses, and the interaction between the number of translation equivalents and the number of senses. The summaries of the final models provide a detailed multifactorial picture of when repetition of reporting verbs is maintained or avoided in literary translation.
The aim of this chapter is to design and implement the Interface of Old English dictionaries (IOED). The IOED is compiled, structured and exploited with database software, in such a way that each lemma or inflectional form is represented in a file with values for various types of lexical information. The IOED, therefore, constitutes a lexical database of Old English. The dictionaries inputted to the IOED include An Anglo-Saxon Dictionary, The student’s Dictionary of Anglo-Saxon, A Concise Anglo-Saxon Dictionary and The Dictionary of Old English in Electronic Form A–I. The design of the IOED is aimed to comparability at type and token level. At type level, the interface relates a given headword of a dictionary to its counterparts in the other dictionaries, including those with different spelling and format. At token level, headwords across dictionaries are linked when they share inflectional forms from the same lexical category. The extended version of the IOED also allows for direct comparability with the most widely used corpora of Old English. The main conclusion is that graph representations in which relations are filed as data turn the lexical database into a knowledge base.
PURPOSE: To evaluate the diagnostic accuracy of dual-energy CT (DECT) iodine maps in comparison to conventional CT series for the assessment of non-occlusive mesenteric ischemia (NOMI). MATERIAL AND METHODS: We evaluated data from 142 patients (72 men; 50.7%) who underwent DECT between 2018 and 2022, with surgically confirmed diagnosis of NOMI. One board-certified radiologist performed region of interest (ROI) measurements in bowel segments on late arterial (LA) and portal venous (PV) phase DECT iodine maps as well as LA conventional series, in both ischemic and non-ischemic bowel loops, using surgical reports as reference standard, and in a control group of 97 patients. Intra- and inter-reader agreement with a second board-certified radiologist was also evaluated. Receiver operating characteristic (ROC) curve analysis was performed to calculate the optimal threshold for discriminating ischemic from non-ischemic bowel segments. Subjective image rating of LA and PV iodine maps was performed. RESULTS: DECT-based iodine concentration (IC) measurements showed significant differences in LA phase iodine maps between ischemic (median:0.72; IQR 0.52-0.91 mg/mL) and non-ischemic bowel loops (5.16; IQR 3.45-6.31 mg/ml) (P <.0001). IC quantification on LA phase revealed an AUC of 0.966 for the assessment of acute bowel ischemia, significantly higher compared to both IC quantification based on PV phase (0.951) and attenuation values evaluated on LA conventional CT series (0.828). Excellent intra-observer and strong inter-observer agreements were observed for the quantification of iodine concentration. Conversely, weak inter-observer agreement was noted for conventional HU assessments. The optimal LA phase-based IC threshold for assessing bowel ischemia was 1.34 mg/mL, yielding a sensitivity of 100% and specificity of 96.48%. CONCLUSION: Iodine maps based on LA phase significantly improve the diagnostic accuracy for the assessment of NOMI compared to conventional CT series and PV phase iodine maps.
Previous research on whether source memory is enhanced for emotionally valenced sources yielded inconclusive results. To identify potential boundary conditions, we tested whether encoding instructions that promote affective versus different types of non-affective item-source-processing foster versus hamper source-valence effects. In both experiments, we used neutral words as items superimposed on emotional (positive & negative) or neutral pictures as sources. Source pictures were selected based on valence and arousal ratings collected in a pre-study such that only valence varied across sources. Source memory was measured via multinomial modelling. In Experiment 1, we applied an affective, item-focused orienting task (OT; i.e. word-pleasantness ratings) during item-source encoding and found enhanced source memory for emotionally valenced (positive & negative) compared to neutral sources. In Experiment 2, we systematically manipulated encoding instructions and again found enhanced source memory for emotionally valenced sources with an affective OT. No such effects occurred in the non-affective conditions, where participants were instructed to integrate item and source (item-source-fit judgments), to focus on the item (living-non-living judgments), or to encode the items only, respectively. With intentional item encoding, however, source memory was surprisingly better for positive than negative sources. We conclude that source-valence effects might unfold only under affective processing.
Tolosa Treebank for Occitan Tolosa Treebank is the first dependency treebank for Occitan, developed as part of the EFA 227/16 LINGUATEC Project, financed by the POCTEFA Interreg European funds. The current version of the treebank contains 25K tokens annotated for PoS tags, lemmas and syntactic dependencies. Linguistic annotation follows Universal Dependencies guidelines (https://universaldependencies.org/#language-u). The corpus files are stored in the ConLL-U format. Each sentence is preceded by a sentence ID and the original, non-tokenized text of the sentence. The annotation is provided in a column-based format defined as follows: 1. ID: Word index, integer starting at 1 for each new sentence; may be a range for multiword tokens.2. FORM: Word form or punctuation symbol.3. LEMMA: Lemma or stem of word form.4. UPOS: Universal part-of-speech tag.5. XPOS: Language-specific part-of-speech tag; underscore if not available.6. FEATS: List of morphological features from the universal feature inventory or from a defined language-specific extension; underscore if not available.7. HEAD: Head of the current word, which is either a value of ID or zero (0).8. DEPREL: Universal dependency relation to the HEAD9. DEPS: Enhanced dependency graph in the form of a list of head-deprel pairs.10. MISC: Any other annotation. The corpus is built out of data from four major Occitan dialects: Gascon, Lengadocian, Lemosin and Provençau. The corpus content is stratified by dialect. File naming conventions are indicated in the readme file. AUTHORS: Aleksandra Miletic, Myriam Bras, Marianne Vergez Couret, Jean Sibille, Louise Esher, Clamença Poujade LICENSE: The corpus is distributed under the Creative Commons BY-NC-SA 4.0 license (https://creativecommons.org/licenses/by-nc-sa/4.0/deed.en).REFERENCING: If you use this corpus in your work, please cite the following paper:Aleksandra Miletic, Myriam Bras, Marianne Vergez-Couret, Louise Esher, Clamença Poujade, and Jean Sibille. 2020. A Four-Dialect Treebank for Occitan: Building Process and Parsing Experiments. In Proceedings of the 7th Workshop on NLP for Similar Languages, Varieties and Dialects, pages 140–149, Barcelona, Spain (Online). International Committee on Computational Linguistics (ICCL).
Abstract The epilogue rearranges the verse covered in the book as a whole back into a chronological narrative, from 1150 to 1500, to balance the non-chronological treatment of the topic necessary in the book as a whole. The time around 1400 is picked out as perhaps the hinge or pivot of the whole history of English poetry, the moment of the greatest potential and variety. The chapter reviews and emphasizes the uncentredness and possibility of early poetry: in its distribution, whether in writing or performance; in its lack of strong linguistic norms; and in the absence of a dominant prestige form. The task of reading early poetry is passed on to the book’s readers.
Introduction: Sleep loss is common during the perinatal period; however, few studies have assessed potential consequences of insufficient sleep for postnatal emotional responding, a key contributor to parenting behaviors with implications for parent-infant bonding and mental health. To generate hypotheses for future work assessing perinatal sleep and emotion-related outcomes, this pilot study explored whether prenatal sleep duration predicted postnatal emotional responding in a sample at risk for postpartum depression. Methods: Participants were nine birthing parents with a prior mood disorder who were not in a current episode at enrollment. We estimated sleep with actigraphy collected for 1 week at 33 weeks' gestation and at 2 and 6 weeks postpartum. Following each week, participants completed an emotional evaluation task, rating the valence and arousal of standardized images from the International Affective Picture System. We tested whether average prenatal (33 weeks) nighttime sleep duration predicted concurrent and future responsiveness to emotional images, quantified by participants' reaction times and arousal/valence ratings. Results: s ≤.043). Conclusions: In this small sample of birthing parents at risk for postpartum depression, shorter prenatal sleep duration predicted faster reactions to emotional stimuli and blunted arousal responses to negative images. Although preliminary, these findings justify further study of the role of prenatal sleep in postpartum emotional responses and how these factors may impact parent-infant outcomes.
The human mind automatically divides continuous experience into meaningful events (event segmentation). Despite abundant evidence that some kinds of situation changes (e.g., action, goal, or location changes) contribute to event segmentation, a component of experience that is critical for understanding and predicting others’ behavior, emotion, is rarely investigated. In two experiments, we sought to establish that viewers can track emotion changes while viewing naturalistic videos, and that these changes contribute to event segmentation. Participants watched commercial film excerpts while identifying either emotion changes or event boundaries (moments that separate two events) of different grains (Experiment 1: neutral-grain; Experiment 2: fine-grain or coarse-grain). We found that participants agreed with each other about when emotion changes occurred in the videos, demonstrating that viewers are able to track changes in the emotional content of dynamic naturalistic videos as they are experienced. Moreover, the emotion changes participants identified were temporally aligned with the event boundaries identified by other groups. In addition, valence and arousal ratings from separate groups of participants uniquely predicted the likelihood of identifying emotion changes and event boundaries, even after accounting for other types of change. However, emotion changes were more strongly tied to valence changes than arousal changes while coarse boundaries were more strongly associated with affective changes than were fine boundaries. These novel findings suggest that emotional information plays a substantial role in structuring ongoing experiences into meaningful events, providing a stronger basis for understanding how emotion shapes the perception and memory of everyday experiences.
Accurate terminology translation is crucial for the global dissemination and comprehension of Traditional Chinese Medicine (TCM). This study addresses the pressing issue of inconsistent TCM terminology, which has led to significant confusion and misinterpretation among scholars and practitioners worldwide. The primary objective of this research is to evaluate the effectiveness of Hu Gengshen’s theoretical framework in ensuring precise and culturally appropriate translations of TCM diagnostic method terms. Hu Gengshen’s framework emphasizes three essential dimensions of translation: communicative, linguistic, and cultural. The communicative dimension focuses on aligning translations with the target audience’s linguistic norms, ensuring the translated terms are accessible and understandable. The linguistic dimension preserves the original structure and style of TCM terminology while adapting it to the target language, maintaining technical accuracy. The cultural dimension guarantees that translations respect and reflect the original cultural context, enhancing cultural sensitivity and relevance. This study employs Hu Gengshen’s framework to assess translations from authoritative TCM sources. The analysis reveals that 146 out of 152 evaluated terms meet the rigorous standards set by the framework across all three dimensions. This finding demonstrates the framework’s effectiveness in producing accurate and culturally sensitive translations. The implications of this research are significant. By validating Hu Gengshen’s framework, the study provides a practical tool for improving the clarity and precision of TCM terminology in international academic and clinical contexts. This enhancement facilitates better global engagement with TCM practices, bridging communication gaps and fostering a deeper understanding of TCM principles and methodologies. The study highlights the framework’s value in addressing translation challenges and advancing the global dissemination of TCM.
Abstract The Holistic Hypothesis asserts that formulaic expressions (FEs) are processed more rapidly than non-formulaic expressions (non-FE items) by both native speakers (NSs) and L2 learners of English. This study utilized an acceptability judgment task and a self-paced reading task to investigate the online processing of FEs and non-FE items among bilingual speakers (L1 English, L2 Chinese) in both contextual and non-contextual conditions. Meanwhile, a familiarity rating task was employed to measure whether there is a familiarity effect in item processing. The results consistently provided support for the Holistic Hypothesis, indicating that learners of Chinese at each level exhibited faster processing of FEs compared to non-FE items, regardless of the presence or absence of context. However, the influence of item familiarity, rather than the proficiency effect, contributed to the improvement of L2 learners’ eventual processing abilities. Distinct patterns also emerged when comparing data from NSs and L2 learners of Chinese, highlighting L2 learners’ more pronounced processing advantage, characterized by faster response times (RTs) to FEs compared to non-FE items. Through an analysis of Chinese L2 data, this study sheds light on the interplay between the usage based approach and chunking within the cognitive approach to L2 learning.
Automatically detecting mental state such as stress from video images of the face could support evaluating stress responses in applicants for high risk jobs or contribute to timely stress detection in challenging operational settings (e.g., aircrew, command center operators). Challenges in automatically estimating mental state include the generalization of models across contexts and across participants. We here aim to create robust models by training them using data from different contexts and including physiological features. Fifty-one participants were exposed to different types of stressors (cognitive, social evaluative and startle) and baseline variants of the stressors. Video, electrocardiogram (ECG), electrodermal activity (EDA) and self-reports (arousal and valence) were recorded. Logistic regression models aimed to classify between high and low arousal and valence across participants, where "high" and "low" were defined relative to the center of the rating scale. Accuracy scores of different models were evaluated: models trained and tested within a specific context (either a baseline or stressor variant of a task), intermediate context (baseline and stressor variant of a task), or general context (all conditions together). Furthermore, for these different model variants, only the video data was included, only the physiological data, or both video and physiological data. We found that all (video, physiological and video-physio) models could successfully distinguish between high- and low-rated arousal and valence, though performance tended to be better for (1) arousal than valence, (2) specific context than intermediate and general contexts, (3) video-physio data than video or physiological data alone. Automatic feature selection resulted in inclusion of 3-20 features, where the models based on video-physio data usually included features from video, ECG and EDA. Still, performance of video-only models approached the performance of video-physio models. Arousal and valence ratings by three experienced human observers scores based on part of the video data did not match with self-reports. In sum, we showed that it is possible to automatically monitor arousal and valence even in relatively general contexts and better than humans can (in the given circumstances), and that non-contact video images of faces capture an important part of the information, which has practical advantages.
Robert Pugh, Francis Tyers. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.
Music-evoked autobiographical memories (MEAMs) are typically elicited by music that listeners have heard before. While studies that have directly manipulated music familiarity show that familiar music evokes more MEAMs than music listeners have not heard before, music that is unfamiliar to the listener can also sporadically cue autobiographical memory. Here we examined whether music that sounds familiar even without previous exposure can produce spontaneous MEAMs. Cognitively healthy older adults (N = 75, ages 65–80 years) listened to music clips that were chosen by researchers to be either familiar or unfamiliar (i.e., varying by prior exposure). Participants then disclosed whether the clip elicited a MEAM and later provided self-reported familiarity ratings for each. Self-reported familiarity was positively associated with the occurrence of MEAMs in response to familiar, but not the unfamiliar, music. The likelihood of reporting MEAMs for music released during youth (i.e., the “reminiscence bump”) relative to young adulthood (20–25 years) included both music released during participants’ adolescence (14–18 years) and middle childhood (5–9 years) once self-reported familiarity was accounted for. These developmental effects could not be accounted for by music-evoked affect. Overall, our results suggest that the phenomenon of MEAMs hinges upon both perceptions of familiarity and prior exposure.
Recent advances in deep learning and Vision-Language Models (VLM) have enabled efficient transfer to downstream tasks even when limited labelled training data is available, as well as for text to be directly compared to image content. These properties of VLMs enable new opportunities for the annotation and analysis of images. We test the potential of VLMs for landscape scenicness prediction, i.e., the aesthetic quality of a landscape, using zero- and few-shot methods. We experiment with few-shot learning by fine-tuning a single linear layer on a pre-trained VLM representation. We find that a model fitted to just a few hundred samples performs favourably compared to a model trained on hundreds of thousands of examples in a fully supervised way. We also explore the zero-shot prediction potential of contrastive prompting using positive and negative landscape aesthetic concepts. Our results show that this method outperforms a linear probe with few-shot learning when using a small number of samples to tune the prompt configuration. We introduce Landscape Prompt Ensembling (LPE), which is an annotation method for acquiring landscape scenicness ratings through rated text descriptions without needing an image dataset during annotation. We demonstrate that LPE can provide landscape scenicness assessments that are concordant with a dataset of image ratings. The success of zero- and few-shot methods combined with their ability to use text-based annotations highlights the potential for VLMs to provide efficient landscape scenicness assessments with greater flexibility.
Abstract The literature are abound with studies on the impact of environmental, social, and governance (ESG) factors on a company‘s value, or more broadly, on its financial performance. However, most analyses concern developed markets, mainly because the largest rating agencies operate in these markets, as well as because these are markets where ESG awareness and regulations have developed much faster. In developing markets, the number of studies in this area is disproportionately smaller. Therefore, the purpose of this article is to examine the relationship between the environmental, social, and governance ratings (ESGR) of Polish listed companies included in the WIG-ESG index and their value. This study covered 36 companies listed in WIG-ESG in the period of 2019–2023. We used market data, financial data from examined companies and ESG data provided by Refinitive. The empirical results were negative but a non-statistically significant influence of ESGR and a company’s value. Further analysis indicated that none of the sub-ratings (environmental rating (ER), social rating (SR) and governance rating (GR)) had significant impact on value. The Polish market does not seem to recognize the potential of ESG factors in building the long-term value of companies and believes that the costs of ESG factors outweigh the benefits. Investors seem to disregard or underestimate ESG criteria when valuing companies, which may seem irrational when looking at the long-term effects of ESG factors. This article contributes to the existing literature by being part of the research on ESG factors and company value. The article expands the field of analysing the relationship between ESGRs and corporate value by examining this relationship not only using the overall ESGR, but also its individual sub-ratings. We also attempt to answer the question of where the channels of transmission of ESGRs on the value of the company are located, and which areas affect ratings. To the best of our knowledge, this is the first study of this type for the Polish market.
It is known that the norms of writing in any language are subject to changes under the influence of internal linguistic processes and external factors of influence. In this regard, the works aimed at studying the issues of linguistic norms, spelling, punctuation, regulation and consolidation of new linguistic phenomena identified in the context of modern realities, legalizing orthograms that have become familiar in the practice of writing, seem very relevant and promising. In accordance with the purpose of our research work, the article examines certain aspects of the issue of applying the rules and norms established by the rules of punctuation and spelling of the Kazakh language in modern writing practice, focuses on a number of rules that are often not taken into account when writing by writers, types of orthograms that introduce inconsistencies in the unity of word transmission, their relation to the norms of the language and the reasons for their use in practice. The research uses methods of collection, selection, systematization of linguistic material, as well as descriptive, statistical, comparative methods and the method of orthological analysis. The results of the study and the recommendations presented make a definite contribution to the unification of words with various variations in modern writing practice, additions and clarifications of spelling rules, and improvement of language literacy of students.
Word identification accuracy is modulated by many factors including linguistic characteristics of words (frequent vs. infrequent), listening environment (noisy vs. quiet), and listener-related differences (older vs. younger). Nearly, all studies investigating these factors use high-familiarity words and noise signals that are either energetic maskers (e.g., white noise) or informational maskers composed of competing talkers (e.g., multitalker babble). Here, we expand on these findings by examining younger and older listeners' speech-in-noise perception for words varying in both frequency and familiarity within a simulated hospital noise that has important non-speech information. The method was inspired by the real-world challenges aging patients can face in understanding less familiar medical terminology used by healthcare professionals in noisy hospital environments. Word familiarity data from older and young adults were collected for 800 medically related terms. Familiarity ratings were highly correlated between the two age groups. Older adults' transcription accuracy for sentences with medical terminology that vary in their familiarity and frequency was assessed across four listening conditions: hospital noise, speech-shaped noise, amplitude-modulated speech-shaped noise, and quiet. Listeners were less accurate in noise conditions than in a quiet condition and were more impacted by hospital noise than either speech-shaped noise. Sentences with low-familiarity and low-frequency medical words combined with hospital noise were particularly detrimental for older adults compared to younger adults. The results impact our theoretical understanding of speech perception in noise and highlight real-world consequences of older adults' difficulties with speech-in-noise and specifically noise containing competing, non-speech information.
Summary While consumers are increasingly used to purchasing online, virtual reality (VR) is well acknowledged to enhance consumer experience via immersive and interaction systems. In our research, we recruited 160 participants (80 per experiment) to compare their consumer experiences of choosing Fast‐Moving Consumer Goods (FMCGs) in a virtual store ( via VR) and on an online webpage. Specifically, each participant was asked to choose four FMCGs (foods in Experiment 1 and non‐foods in Experiment 2) for the next 4 weeks, and they filled out the Positive and Negative Affect Scale before and after completing choice tasks. Results of the anova showed that there was a significant decrease in negative affect scores following both food and non‐food choices ( M before = 1.3 vs. M after = 1.2). Positive affect ratings were significantly higher for food choices in VR than online ( M VR = 2.5 vs. M online = 2.1), with no significant difference for non‐food choices. This highlights the impact of VR on food choice and its potential to transform human–food interactions. The study reveals that VR can narrow the gap between consumers' expectations and actual perceptions by creating an immersive and interactive shopping experience, thus significantly influencing consumer behaviour.
Abstract Naturalistic paradigms can assure ecological validity and yield novel insights in psychology and neuroscience. However, using behavioral experiments to obtain the human ratings necessary to analyze data collected with these paradigms is usually costly and time-consuming. Large language models like GPT have great potential for predicting human-like behavioral judgments. The current study evaluates the performance of GPT as a substitute for human judgments for affective dynamics in narratives. Our results revealed that GPT’s inference of hedonic valence dynamics is highly correlated with human affective perception. Moreover, the inferred neural activity based on GPT-derived valence ratings is similar to inferred neural activity based on human judgments, suggesting the potential of using GPT’s prediction as a reliable substitute for human judgments.
Previous research has established that determining lexical sophistication (i.e., the percentage of sophisticated words in a text) through the judgment of teachers on a corpus of words is a more accurate method than relying on word frequency-based lists. However, this approach can be time-consuming. To overcome this drawback, a new method is proposed in this study, which involves rating specific words out of context. A list of 68 words that appeared in approved high-school textbooks of teaching Hebrew to Arabic speakers was given to six experienced Hebrew teachers, who then categorized the words into four levels of lexical sophistication: (1) very basic words to (4) very advanced words. From this, a list of 28 words was created, with seven words from each level, and the lexical sophistication level was agreed upon by two-thirds of the teachers. Nineteen Arabic-speaking learners of Hebrew were asked to define the chosen words (passive vocabulary) and compose a sentence including each (controlled-active vocabulary) in a test-retest study at two time-points: the 11th and 12th grade. The results indicated that although there was no significant increase in lexical sophistication over time, significant differences emerged between the four levels of lexical sophistication, with students’ accuracy decreasing as the level of lexical sophistication increased. Additionally, only in the 11th grade was passive vocabulary found to be significantly larger than controlledactive vocabulary. However, as acquisition time increased, the gap between these two vocabulary types narrowed, due to improved performance in the controlled-active task. Furthermore, a significant correlation was found between passive and controlled-active vocabulary, which became stronger with more acquisition time.
Augmented Reality (AR) technology has begun to significantly influence human interaction with language, particularly in English language use. Hence, this research examined the influence of Augmented Reality (AR) on communication patterns, the emergence of specialized AR-related language, and its impact on English language norms. For this purpose, 10 Students from the MA TEFL workshop at Allama Iqbal Open University, Islamabad, were selected to conduct detailed interviews. Interview data and its content were analyzed linguistically. These students were also teachers, so the data were used both ways. Through linguistic analysis, it was found that Augmented Reality (AR) significantly influences English language communication patterns by notable shifts in vocabulary, syntax, and discourse structures. Integrating AR-specific terms into everyday language among MA TEFL students signifies a transformative impact on linguistic norms, reflecting a deeper entwinement of language and technology. The study's findings recommended that language educators and curriculum developers should actively integrate AR technology and its associated linguistic elements into English language teaching and learning frameworks.
The teaching material of a student can come from anywhere, one of which is an oration text. In an oration text, it is necessary to pay attention to the quality of content, effective sentences, and language suitability. This research aims to know the quality of content and language suitability contained in the Mendikbudristek oration text on National Education Day 2023 for a tenth-grade SMA student's teaching material. The method used in this research is a qualitative descriptive method. This qualitative descriptive method is used to investigate and explain matters related to a fact that occurs in the surrounding environment. Additionally, the data analysis in the study is conducted using the method of agih. The method of agih is a method where the determining tool is part of the language concerned. The result of this research is that there are still several errors in the Mendikbudristek oration text on National Education Day 2023 and 2024, including spelling errors, incorrect words, and word economy. This causes some sentences to be ineffective and the information provided is not well conveyed to the reader. It is hoped that this research can provide knowledge to readers so that they can understand language rules and linguistic norms, thus ensuring that the purpose of the provided information is well conveyed to the public.
Emotional stimuli are usually remembered with high confidence. Yet, it remains unknown whether—in addition to memory for the emotional stimulus itself—memory for a neutral stimulus encountered just after an emotional one can be enhanced. Further, little is known about the interplay between emotion elicited by a stimulus and emotion relating to affective dispositions. To address these questions, we examined (1) how emotional valence and arousal of a context image preceding a neutral item image affect memory of the item, and (2) how such memory modulation is affected by two hallmark features of emotional disorders: trait negative affect and tendency to worry. In two experiments, participants encoded a series of trials in which an emotional (negative, neutral, or positive) context image was followed by a neutral item image. In experiment 1 ( n = 42), items presented seconds after negative context images were remembered better and with greater confidence compared to those presented after neutral and positive ones. Arousal ratings of negative context images were higher compared to neutral and positive ones and the likelihood of correctly recognizing an item image was related to higher arousal of the context image. In experiment 2 ( n = 59), better item memory was related to lower trait negative affect. Participants with lower trait negative affect or tendency to worry displayed higher confidence compared to those with high negative affect or tendency to worry. Our findings describe an emotional “carry-over” effect elicited by a context image that enhances subsequent item memory on a trial-by-trial basis, however, not in individuals with high trait negative affect who seem to have a general memory disadvantage.
In this paper, we aimed to develop a neural parser for Vietnamese based on simplified Head-Driven Phrase Structure Grammar (HPSG). The existing corpora, VietTreebank and VnDT, had around 15% of constituency and dependency tree pairs that did not adhere to simplified HPSG rules. To attempt to address the issue of the corpora not adhering to simplified HPSG rules, we randomly permuted samples from the training and development sets to make them compliant with simplified HPSG. We then modified the first simplified HPSG Neural Parser for the Penn Treebank by replacing it with the PhoBERT or XLM-RoBERTa models, which can encode Vietnamese texts. We conducted experiments on our modified VietTreebank and VnDT corpora. Our extensive experiments showed that the simplified HPSG Neural Parser achieved a new state-of-the-art F-score of 82% for constituency parsing when using the same predicted part-of-speech (POS) tags as the self-attentive constituency parser. Additionally, it outperformed previous studies in dependency parsing with a higher Unlabeled Attachment Score (UAS). However, our parser obtained lower Labeled Attachment Score (LAS) scores likely due to our focus on arc permutation without changing the original labels, as we did not consult with a linguistic expert. Lastly, the research findings of this paper suggest that simplified HPSG should be given more attention to linguistic expert when developing treebanks for Vietnamese natural language processing.
Abstract In recent years, the study of codeswitching has made great strides by leveraging a multidisciplinary approach that integrates insights from experimental psycholinguistics, corpus linguistics, cognitive psychology, neurolinguistics, and other allied fields. We provide an overview of 2 main strands of this research, focusing on the control process model of codeswitching (e.g., Green, 2018), the variable equivalence hypothesis (e.g., Torres Cacoullos, 2020), and the ways in which these accounts of codeswitching behavior intersect and complement one another. We argue that the common insight of these 2 distinct approaches is their focus on conventionalized linguistic norms at the level of the speech community. The second portion of the article sheds light on how various aspects of codeswitching experience give rise to adaptive change through bilingual phenotyping, highlighting the centrality of our mentor Judy Kroll's contributions to this research. Finally, we present a social network analysis of Judy's research publications and argue that the high degree of interconnectivity in Judy's research network, combined with the many positive community norms that she has been instrumental in establishing, have greatly benefited both the individual members of the network and the research enterprise itself.
Traditional morphological methods have long been at the heart of Arabic language processing. They exploit the complex morphological structure of the language, characterized by roots, prefixes, infixes, and suffixes used for analyzing and generating words. These methods have been effective for tasks such as lemmatization, and rooting. However, they can be resource-intensive and do not always take semantics into account. Recent advancements in deep learning, particularly transformer-based models, offer new prospects for Arabic language processing. Despite their success in languages such as Turkish, Latin, English, and Slavic, to our knowledge, no transformer-based method has been implemented for the Arabic lemmatization task. In this study, we propose a transformer-based lemmatizer using T5, specifically AraT5, MT5, and BERT. We evaluate the models using the ArabicPADT UD Treebank corpus and benchmark their performance against standard metrics, including accuracy, precision, recall, and F1 score. Our results show an accuracy of $88.58 \%$, demonstrating competitive performance and highlighting the potential of transformer-based models for advancing Arabic lemmatization.
Activity and parameter sparsity are two standard methods of making neural networks computationally more efficient. Event-based architectures such as spiking neural networks (SNNs) naturally exhibit activity sparsity, and many methods exist to sparsify their connectivity by pruning weights. While the effect of weight pruning on feed-forward SNNs has been previously studied for computer vision tasks, the effects of pruning for complex sequence tasks like language modeling are less well studied since SNNs have traditionally struggled to achieve meaningful performance on these tasks. Using a recently published SNN-like architecture that works well on small-scale language modeling, we study the effects of weight pruning when combined with activity sparsity. Specifically, we study the tradeoff between the multiplicative efficiency gains the combination affords and its effect on task performance for language modeling. To dissect the effects of the two sparsities, we conduct a comparative analysis between densely activated models and sparsely activated event-based models across varying degrees of connectivity sparsity. We demonstrate that sparse activity and sparse connectivity complement each other without a proportional drop in task performance for an event-based neural network trained on the Penn Treebank and WikiText-2 language modeling datasets. Our results suggest sparsely connected event-based neural networks are promising candidates for effective and efficient sequence modeling.
We propose the Hyperbolic Tangent Exponential Linear Unit (TeLU), a neural network hidden activation function defined as TeLU(x)=xtanh(exp(x)). TeLU's design is grounded in the core principles of key activation functions, achieving strong convergence by closely approximating the identity function in its active region while effectively mitigating the vanishing gradient problem in its saturating region. Its simple formulation enhances computational efficiency, leading to improvements in scalability and convergence speed. Unlike many modern activation functions, TeLU seamlessly combines the simplicity and effectiveness of ReLU with the smoothness and analytic properties essential for learning stability in deep neural networks. TeLU's ability to mimic the behavior and optimal hyperparameter settings of ReLU, while introducing the benefits of smoothness and curvature, makes it an ideal drop-in replacement. Its analytic nature positions TeLU as a powerful universal approximator, enhancing both robustness and generalization across a multitude of experiments. We rigorously validate these claims through theoretical analysis and experimental validation, demonstrating TeLU's performance across challenging benchmarks; including ResNet18 on ImageNet, Dynamic-Pooling Transformers on Text8, and Recurrent Neural Networks (RNNs) on the Penn TreeBank dataset. These results highlight TeLU's potential to set a new standard in activation functions, driving more efficient and stable learning in deep neural networks, thereby accelerating scientific discoveries across various fields.
An acute bout of exercise in the moments after learning benefits the retention of new memories. This finding can be explained, at least partly, through a consolidation account: exercise provides a physiological state that is conducive to the early stabilisation of labile new memories, which supports their retention and subsequent retrieval. The modification of consolidation through non-invasive exercise interventions offers great applied potential. However, it remains poorly understood whether effects of exercise translate from the laboratory to naturalistic settings and whether the intensity of exercise determines the effect in memory. To this end, adult endurance runners were recruited as participants and completed two study sessions spaced two weeks apart. In each session, participants were presented with a list of words and asked to recall them on three occasions: (i) immediately following their presentation, (ii) after a 30-minute retention interval, and (iii) after 24 hours. Crucially, the 30-minute retention interval comprised our experimental manipulation: higher intensity exercise (running) in the first session and lower intensity exercise (walking) in the second, both completed in a naturalistic setting around participants' existing physical activity training programmes. Exertion was recorded through heart rate and rate of perceived exertion data. Alertness, mood, and arousal ratings were also collected before and after the 30-minute retention interval. Immediate memory for the two wordlists was matched, but participants retained significantly more words after 30 minutes and 24 hours when encoding was followed by higher than lower intensity exercise. Exertion data revealed that participants experienced vigorous and light exercise in the higher and lower intensity conditions, respectively. Significant improvements in alertness, mood, and arousal were observed following both exercise conditions, but especially in the higher intensity condition. These outcomes reveal that experiencing higher intensity physical activity in the field is conducive to declarative memory retention, possibly because it encourages consolidation.
Abstract Syntactic parsing is one of the areas in Natural Language Processing. The development of large-scale multilingual language models has enabled cross-lingual parsing approaches, which allows us to develop parsers for languages that do not have treebanks available. However, these approaches rely on the assumption that languages share orthographic representations and lexical entries. In this article, we investigate methods for developing a dependency parser for Xibe, a low-resource language that is written in a unique script. We first investigate lexicalized monolingual dependency parsing experiments to examine the effectiveness of word, part-of-speech, and character embeddings as well as pre-trained language models. Results show that character embeddings can significantly improve performance, while pre-trained language models decrease performance since they do not recognize the Xibe script. We also train delexicalized monolingual models, which yield competitive results to the best lexicalized model. Since the monolingual models are trained on a very small training set, we also investigate lexicalized and delexicalized cross-lingual models. We use six closely related languages as source language, which cover a wide range of scripts. In this setting, the delexicalized models achieve higher performance than lexicalized models. A final experiment shows that we can increase performance of the cross-lingual model by combining source languages and selecting the most similar sentences to Xibe as training set. However, all cross-lingual parsing results are still considerably lower than the monolingual model. We attribute the low performance of cross-lingual methods to syntactic and annotation differences as well as to the impoverished input of Universal Dependency Part-of-Speech tags that the delexicalized model has access to.
Purpose: The general purpose of this study was to explore the impact of social media on language evolution. Methodology: The study adopted a desktop research methodology. Desk research refers to secondary data or that which can be collected without fieldwork. Desk research is basically involved in collecting data from existing resources hence it is often considered a low cost technique as compared to field research, as the main cost is involved in executive’s time, telephone charges and directories. Thus, the study relied on already published studies, reports and statistics. This secondary data was easily accessed through the online journals and library. Findings: The findings reveal that there exists a contextual and methodological gap relating to the impact of social media on language evolution. Preliminary empirical review revealed significant changes in language use driven by the rapid spread of new words, phrases, and communication styles on social media platforms. It found that social media democratized language change, allowing diverse users to influence linguistic trends, and highlighted the emergence of micro-languages within online communities. The integration of visual elements like emojis and memes into text-based communication added nuance and expressiveness, aligning digital interactions more closely with face-to-face communication. The study emphasized the need for digital literacy in education and ongoing research into digital communication's impact on language, noting social media's role as a powerful catalyst for language change. Unique Contribution to Theory, Practice and Policy: The Social Network Theory, Diffusion of Innovations Theory and Speech Community Theory may be used to anchor future studies on social media on language evolution. The study recommended expanding traditional linguistic theories to incorporate the dynamics of digital communication, incorporating digital literacy into education to prepare students for new linguistic norms, and for businesses to align their messaging with emerging social media trends. It also advised policymakers to use social media for language preservation, bridge the digital divide to promote linguistic diversity, and support interdisciplinary research to understand the long-term effects of social media on language. These recommendations aimed to adapt theory, practice, and policy to the evolving linguistic landscape.
Fine-tuning pre-trained language models for specific natural language processing tasks often leads to suboptimal performance due to the vast parameter space and the challenge of finding the most effective model configurations. Traditional fine-tuning methods can be computationally expensive and may not fully exploit the potential of these models, particularly for tasks like sentiment analysis. This work explores the application of metaheuristic techniques, focusing on sentiment analysis using the Stanford Sentiment Treebank 2 (SST2) dataset.We utilize two approaches: Genetic Algorithms (GA) and the Whale Optimization Algorithm (WOA). These methods aim to efficiently search the parameter space and optimize the fine-tuning process of pretrained language models. To enhance efficiency, we incorporate layer freezing techniques along with different crossover strategies. In the results obtained, GA with Simulated Binary Crossover (SBX) and 30% layer freezing achieved 94.02% accuracy, while WOA with SBX crossover and 30% layer freezing achieved 94.67% accuracy. These outcomes highlight the potential of integrating evolutionary algorithms, nature-inspired optimization techniques, and strategic layer freezing with deep learning finetuning processes for NLP tasks.
The administration of painful primes has been shown to influence the perception of successively presented semantic stimuli. Painful primes lead to more negative valence ratings of pain-related, negative, and positive words than no prime. This effect was greater for pain-related than negative words. The identities of this effect's neural correlates remain unknown. In this EEG experiment, 48 healthy subjects received noxious electrical stimuli of moderate intensity. During this priming, they were presented with adjectives of variable valence (pain-related, negative, positive, and neutral). The triggered event-related potentials were analyzed during N1 (120-180 ms), P2 (170-260 ms), P3 (300-350 ms), N400 (370-550 ms), and two late positive complex components (LPC1 [650-750 ms] and LPC2 [750-1000 ms]). Larger event-related potentials were found for negative and pain-related words compared to positive words in later components (N400, LPC1, and LPC2), mainly in the frontal regions. Early components (N1, P2) were less affected by the word category but were by the prime condition (N1 amplitude was smaller with than without painful stimulation, P2 amplitude was larger with than without painful stimulation). Later components (LPC1, LPC2) were not affected by the prime condition. An interaction effect involving prime and word category was found on the behavioral level but not the electrophysiological level. This finding indicates that the interaction effect does not directly translate from the behavioral to the electrophysiological level. Possible reasons for this discrepancy are discussed.
This paper analyzes the translation of “Xing Qing” in two seminal English versions of A Dream of Red Mansions by Yang Hsien-yi, Gladys Yang, and David Hawkes. It examines the translation strategies employed across different contexts using corpus analysis. The term “Xing Qing” is translated in three principal ways: literal translation, which faithfully replicates the characters’ personality traits; free translation, which adapts to specific situations and emotional contexts; and a version aligning with English linguistic norms. Yang’s translation tends toward literal fidelity, emphasizing the preservation of cultural nuances. In contrast, Hawkes’ approach is more flexible, favoring free translation and prioritizing reader engagement. The study argues that translators should select contextually appropriate strategies that convey the characters’ traits and emotions accurately and ensure both the translation’s fluidity and the retention of cultural information. This analysis underscores the importance of strategic choice in bal-ancing fidelity to the source text with the accessibility of the translation.
To analyze English discourse more accurately and provide more detailed feedback information, this study applies Rasch measurement and Conditional Random Field (CRF) models to English discourse analysis. The Rasch measurement model is widely used to evaluate and quantify the potential traits of individuals, and it has remarkable advantages in measurement and evaluation. By combining the CRF model, the Rasch model is employed to model the structural and semantic information in the discourse and use this model to carry out sequence labeling, to enhance the ability to capture the internal relations of the discourse. Finally, this study conducts comparative experiments on integrating the Rasch measurement and CRF models, comparing the outcomes against traditional scoring methods and the standalone CRF model. The research findings indicate that: (1) The discourse component syntactic analysis model on the Penn Treebank (PTB) database obtained Unlabeled Attachment Score (UAS) values of 94.07, 95.76, 95.67, and 95.43, and Labeled Attachment Score (LAS) values of 92.47, 92.33, 92.49, and 92.46 for the LOC, CRF, CRF2O, and MFVI models, respectively. After adding the Rasch measurement model, the UAS values of the four models on the PTB database are 96.85, 96.77, 96.92, and 96.78 for the LOC, CRF, CRF2O, and MFVI models, respectively, with LAS values of 95.33, 95.34, 95.39, and 95.32, all showing significant improvement. (2) By combining contextual information with CRF models, students can better understand their discourse expression, capture the connections between English discourse sentences, and analyze English discourse more comprehensively. This study provides new ideas and methods for researchers in English language education and linguistics.
This article is about Semantic Role Labeling for English partitive nouns (5%/REL of the price/ARG1; The price/ARG1 rose 5 percent/REL) in the NomBank annotated corpus. Several systems are described using traditional and transformer-based machine learning, as well as ensembling. Our highest scoring system achieves an F1 of 91.74% using "gold" parses from the Penn Treebank and 91.12% when using the Berkeley Neural parser. This research includes both classroom and experimental settings for system development.
Teaching English to university students requires a balance between fluency and accuracy.Both aspects are crucial for effective communication, but they serve slightly different purposes and should be addressed appropriately in a universitylevel English course.The paper explores the importance of accuracy and fluency in teaching and learning English and emphasises they are both essential components of mastering a foreign language.Teaching both accuracy and fluency requires a structured approach that addresses students' academic needs and prepares them for real-world communication.Within the scope of this paper, we focus on effective strategies for promoting and integrating fluency and accuracy in English language classrooms.Fluency and accuracy in language teaching are important since they complement each other.We believe it is necessary to foster fluency and accuracy in English language classrooms that result in higher levels of communicative competence.These two aspects help learners become effective communicators who can express themselves clearly and confidently while adhering to the linguistic norms of the language.The paper aims to provide the best ideas and effective practices to be implemented in an English language classroom: using authentic materials, providing constructive feedback, integrating task-based learning, and encouraging peer collaboration.The responsibility of every teacher is to develop a balanced and structured approach when every student gets a chance to work on fluency and accuracy.The teacher should provide constructive feedback that addresses both accuracy and fluency.By balancing the development of accuracy and fluency in English, university students can effectively communicate in academic settings and beyond, preparing them for future academic success and professional endeavours.
The cognitive effort required for humans to perceive similarities and relationships between words is considerable. Measuring similarity and relatedness between text components such as words, texts, or documents is challenging, and it continues to be an active area of research across various domains. The complexity of language and the diverse factors that influence similarity and relatedness make this task an ongoing research focus. Researchers are exploring diverse approaches, to improve the accuracy and effectiveness of measuring similarity and relatedness in text. The utilization of knowledge sources, such as WordNet, has been a popular approach for modeling semantic relationships between words. However, Recently, distributional semantic models, such as Word2Vec, have demonstrated their ability to effectively capture semantic information and outperform lexiconbased methods in terms of unidirectional contextual similarity outcomes. In contrast to lexicon-based approaches, which rely on structure, distributional models leverage context to capture semantics. This study proposes a novel approach that linearly combines the lexical databases WordNet and Word2Vec to measure semantic similarity, focusing on improving upon previous techniques. The proposed approach is thoroughly detailed and evaluated using popular datasets to determine its effectiveness. The experimental results indicate that the proposed approach achieves highly satisfactory results and surpasses the performance of individual methods.