Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18265 papers
Revealing the syntactic structure of sentences in Chinese poses significant challenges for word-level parsers due to the absence of clear word boundaries. To facilitate a transition from word-level to character-level Chinese dependency parsing, this paper proposes modeling latent internal structures within words. In this way, each word-level dependency tree is interpreted as a forest of character-level trees. A constrained Eisner algorithm is implemented to ensure the compatibility of character-level trees, guaranteeing a single root for intra-word structures and establishing inter-word dependencies between these roots. Experiments on Chinese treebanks demonstrate the superiority of our method over both the pipeline framework and previous joint models. A detailed analysis reveals that a coarse-to-fine parsing strategy empowers the model to predict more linguistically plausible intra-word structures.
While structure learning achieves remarkable performance in high-resource languages, the situation differs for under-represented languages due to the scarcity of annotated data. This study focuses on assessing the efficacy of transfer learning in enhancing dependency parsing for Javanese, a language spoken by 80 million individuals but characterized by limited representation in natural language processing. We utilized the Universal Dependencies dataset consisting of dependency treebanks from more than 100 languages, including Javanese. We propose two learning strategies to train the model: transfer learning (TL) and hierarchical transfer learning (HTL). While TL only uses a source language to pre-train the model, the HTL method uses a source language and an intermediate language in the learning process. The results show that our best model uses the HTL method, which improves performance with an increase of 10% for both UAS and LAS evaluations compared to the baseline model.
Second-order Recurrent Neural Networks (2RNNs) extend RNNs by leveraging second-order interactions for sequence modelling. These models are provably more expressive than their first-order counterparts and have connections to well-studied models from formal language theory. However, their large parameter tensor makes computations intractable. To circumvent this issue, one approach known as MIRNN consists in limiting the type of interactions used by the model. Another is to leverage tensor decomposition to diminish the parameter count. In this work, we study the model resulting from parameterizing 2RNNs using the CP decomposition, which we call CPRNN. Intuitively, the rank of the decomposition should reduce expressivity. We analyze how rank and hidden size affect model capacity and show the relationships between RNNs, 2RNNs, MIRNNs, and CPRNNs based on these parameters. We support these results empirically with experiments on the Penn Treebank dataset which demonstrate that, with a fixed parameter budget, CPRNNs outperforms RNNs, 2RNNs, and MIRNNs with the right choice of rank and hidden size.
Deep neural networks (DNNs) are widely used in fields like computer vision and natural language processing. A key component of DNN training is the optimizer. SGD-Momentum is popular in many DNN methodologies, such as ResNet and DenseNet, due to its simplicity and effectiveness. However, its slow convergence rate limits its use. To overcome this, we introduce inter-gradient collision into SGD-Momentum, inspired by the elastic collision model in physics. This new method, called ICSGD-Momentum, aims to improve convergence. We provide theoretical proof of convergence and establish a regret bound for ICSGD-Momentum. Experiments on benchmarks including function optimization, CIFAR-100, ImageNet, Penn Treebank, COCO, and YCB-Video show that ICSGD-Momentum accelerates training and enhances the generalization performance of DNNs compared to optimizers like SGD-Momentum, Adam, Radam, Adabound, and AdaBelief.
The paper reports on a series of experiments aiming at probing LeBenchmark, a pretrained acoustic model trained on 7k hours of spoken French, for syntactic information. Pretrained acoustic models are increasingly used for downstream speech tasks such as automatic speech recognition, speech translation, spoken language understanding or speech parsing. They are trained on very low level information (the raw speech signal), and do not have explicit lexical knowledge. Despite that, they obtained reasonable results on tasks that requires higher level linguistic knowledge. As a result, an emerging question is whether these models encode syntactic information. We probe each representation layer of LeBenchmark for syntax, using the Orféo treebank, and observe that it has learnt some syntactic information. Our results show that syntactic information is more easily extractable from the middle layers of the network, after which a very sharp decrease is observed.
Temporal Convolutional Networks (TCNs) are one-dimensional convolutional neural networks for modelling sequential data. A key component in TCN is the dilation, that is used to increase the receptive field while keeping the number of parameters low. Dilation rates are predetermined in TCN. In this paper, an adaptive method is introduced for learning dilation rates by utilizing trainable binary masks with sparsity constraints (named as Adaptive TCN, AdaTCN). To select connections that are deemed important, the binary masks are applied to convolutional layers. We introduce structured sparsity into the mask using Gumbel Sharp softmax in order to control the number of active connections. Four different models, including TCN, random masked TCN, AdaTCN, and an AdaTCN that is initiated with TCN-like mask are trained and evaluated. With the Penn TreeBank (PTB) and WikitText-2 (WT2) datasets, experiments are conducted on word-level language models
This study examined the extent to which musical training, familiarity, and personality predict music preferences among Malaysian secondary school students. Subjects were 381 16-year-old Malaysian secondary school students from Pasir Gudang, Johor divided into two groups, i.e. those with musical training and those without musical training. The subjects listened to forty music excerpts from eight genres, including Malaysian Pop, Western Pop, Malaysian Hip-Hop, Western Hip-Hop, Malaysian R&B, Western R&B, Malaysian Rock, and Western Rock. Only vocal excerpts were utilised, with tempos ranging from 140 to 180 beats per minute. Subjects completed three questionnaire sections. Section A was demographic information; Section B consisted of the music preference inventory (MPI) and familiarity rating scale; and Section C was the Big Five Inventory (BFI). Results showed that musical training had a strong influence on music preference. Subjects with musical training obtained higher mean scores for each music genre preference. Familiarity was also strongly correlated with music preferences. Each of the Big Five personality dimensions was also positively correlated with every music genre preference. In conclusion, musical training, familiarity, and personality play a crucial role in music preference decisions.
The article reflects on the historical and socio-linguistic processes in Ukraine that contributed to the liberalization of language norms in the late 20th and early 21st centuries. This has led to the emergence of neologisms. There are factors that were shaping the linguo-cultural space in Ukraine during this period: granting Ukrainian the status of the sole state language, as the language of the titular nation of Ukraine, legislative and linguistic initiatives that supported the development of the language, globalization processes, and geopolitical changes. Abbreviative derivatives as neologisms of modern Ukrainian emerged during the process of liberalization and belong to four distinct parts of speech. Actually, speakers form adjectival derivatives. They are most frequently found in journalistic texts, as well as in conversational, literary, and academic styles. A socio-linguistic survey conducted in 2024 confirmed trends in the word formation of adjectival derivatives from abbreviative bases. Speakers add the confix (interfix + suffix) -ivs`k-. Today, the main issue in researching such neologisms is the identification, justification, and systematization of variant and invariant standard forms according to linguistic norms.
Previously used to refer to generic antecedents and antecedents of unknown gender, singular they has been found to increasingly occur with definite antecedents of known gender. This shift is associated with rising awareness of nonbinary gender identities and the expansion of they as a preferred pronoun. Usage of singular they has been previously examined only within Inner Circle Englishes (e.g., US English). In this study, we investigate sociolinguistic factors that influence the acceptability of singular they in Singapore English, an Outer Circle variety that is pivoting towards internal linguistic norms but also experiences frequent contact with non-local Englishes. We find that singular they is rated as significantly more grammatical by younger respondents; its rating is also constrained by definiteness and interactions between social factors, including gender and religiosity. These factors are found to be stronger predictors of singular they acceptability than linguistic prescriptivism. The diffusion of singular they to Singapore English illustrates the ongoing role of non-local contact in the evolution of this variety.
Languange holds a significant role in facilitating human thinking, serving as a foundation for understanding and accessing knowledge. However, the development of Indonesian language is currently experiencing a decline, mainly due to the widespread influence of social media. Social media users, often referred to as netizens, often use terms or vocabulary that are not in line with linguistic norms. This results in the communication patterns of Indonesian people in daily life, both in oral and written forms. This study uses an observation method with reading and note-taking techniques to describe the development of slang used by millennial teenagers on various social media platforms. The research findings show that the slang used by millennial teenagers comes from various sources, including regional languages, Indonesian, foreign languages, and a combination of Indonesian and foreign languages. Therefore, the use of slang by millennial teenagers is interpreted as a form of self-expression when building friendships and fostering close relationships between fellow teenagers
This report provides an exploration of the farming culture of the Achang people in Yunnan Province, China, and its translation practice. Through the Translation Workshop course, students collaboratively translated texts regarding the traditional agricultural practices and cultural significance of the Achang people. Participants not only enhance their individual translation skills but also deepen their understanding of the art and science of translation. The study employs the Functional Equivalence theory as translation guidance to ensure that the translated texts are both faithful to the original and adaptive to the cultural and linguistic norms of the target language. The report analyzes challenges encountered during the translation process, including cultural specificity and accurate expression of terminology, and proposes corresponding strategies. Methods such as re-creation, omission, and transliteration are used to improve the readability and accuracy of the target language. Finally, the report summarizes the gains from the translation practice and offers suggestions for future research directions and improvements.
Music is often regarded as the 'language of the emotions' (Cooke, 1995), and since the early 20th century, empirical research into musical emotions has been conducted to explore the mystery of how they are evoked by music. In this context, Frank et al. presented an experimental study that examined the scarcely researched field of historical listening. Their study aimed to investigate the question, "Do modern listeners hear the emotional content in Baroque music that the composer intended to portray?". The results indicated that modern listeners placed the three modern excerpts in the expected quadrants of the valence-arousal space. However, there were significantly different valence and arousal ratings among the Baroque examples and the modern excerpts, and significant differences between paired examples (where Baroque and modern examples were expected to fall into the same quadrants) occurred. This commentary summarizes Frank et al.'s experimental study, discusses methodological considerations, and suggests possible refinements for future (experimental) studies on historical listening.
This paper proposes incorporating linguistic semantic information into discourse relation recognition and constructing a Semantic Augmented Chinese Discourse Corpus (SACA) comprising 9546 adversative complex sentences. In adversative complex sentences, we suggest a quadruple (P, Q, R, Qβ) representing internal semantic elements, where the semantic opposition between Q and Qβ forms the basis of the adversative relationship. P denotes the premise, and R represents the adversative reason. The overall annotation approach of this corpus follows the Penn Discourse Treebank (PDTB), except for the classification of senses. We combined insights from the Chinese Discourse Treebank (CDTB) and obtained eight sense categories for Chinese adversative complex sentences. Based on this corpus, we explore the relationship between sense classification and internal semantic elements within our newly proposed Chinese Adversative Discourse Relation Recognition (CADRR) task. Leveraging deep learning techniques, we constructed various classification models and the model that utilizes internal semantic element features, demonstrating their effectiveness and the applicability of our SACA corpus. Compared with pre-trained models, our model incorporates internal semantic element information to achieve state-of-the-art performance.
This paper introduces and demonstrates MWE-Finder, an application to search for flexible multiword expressions (MWEs) in Dutch text corpora, starting from an example. If the example is in canonical form, the application automatically generates three queries to search for sentences that contain an occurrence of the MWE and thus enables efficient analysis of its properties. The application offers canonical forms for more than 11k MWEs. Searching is done in treebanks, so the grammatical structure of the sentences is taken into account.
The article is devoted to the analysis of simple sentences’ structure of Russian and Uzbek languages. We propose an algorithm that solves crucial problem for machine translation of these unrelated languages, and the linguistic database that gives the possibility to implement the process of machine translation.
During the initial speech contact the speaker has the opportunity to freely choose various language means described in the speech etiquette. But the deviation from the linguistic norm is observed quite often. In a relaxed dialogue interaction, the initiator of communication can use various thematic statements as phatic units that successfully carry out the act of establishing initial speech contact. In fiction, such statements are characterised by a rather wide semantic diversification.
This paper investigates how linguistic norms are negotiated in German-speaking localities in Rio Grande do Sul, Brazil. The aim of the article is to find out whether linguistic norms still play a role in a heterogeneous multilingual context in which German ceased to be used as a written language while still being transferred to children in spoken varieties. The data is based on 58 semi-structured interviews from ten locations. The analyses take their starting point in language labels, providing key words for identifying and contrasting varieties. The results show that different varieties of German are still clearly perceived and labeled, and that they are evaluated according to the vertical dimension. Norms are negotiated displaying an ideology of linguistic homogeneity and relating to speaker age, the value of written language, and norm instances like schools. Comparing the older and the younger generations, a tendency of norm varieties being less associated with written language and more and more based on surrounding spoken German varieties is perceived, with West Central German (Hunsrückisch) showing some dominance.
This study introduces a pretrained large language model-based annotation methodology for the first dependency treebank in Ottoman Turkish.Our experimental results show that, iteratively, i) pseudo-annotating data using a multilingual BERT-based parsing model, ii) manually correcting the pseudo-annotations, and iii) fine-tuning the parsing model with the corrected annotations, we speed up and simplify the challenging dependency annotation process.The resulting treebank, that will be a part of the Universal Dependencies (UD) project, will facilitate automated analysis of Ottoman Turkish documents, unlocking the linguistic richness embedded in this historical heritage.
Objective This study aimed to assess the impact of the different concentrations of iodine contrast agents used on the quality of computed tomography (CT) images obtained intraindividually in hepatocellular carcinoma patients. Methods In this retrospective study, data from a cohort of 29 patients diagnosed with primary hepatocellular carcinoma who had undergone two preoperative CT-enhanced examinations within a 3-month timeframe were analyzed. Each patient was randomly assigned to receive either a low-concentration contrast agent (300 mg I/mL iohexol) or a high-concentration contrast agent (350 mg I/mL iohexol) for the first scan and the alternative contrast agent for the second scan. CT images of different liver regions of each patient were compared between low-and high-concentration scans using their before-and-after control design. Subjective image quality scores for portal vein images were also assessed. Results The findings of this study indicate that patients in the high-concentration group presented significantly elevated CT values across various anatomical regions, including the liver parenchyma, abdominal aorta, and hepatic portal vein, compared to those in the low-concentration group ( p < 0.05). Moreover, the high-concentration group demonstrated superior subjective image ratings ( p < 0.05). Nevertheless, there was no statistically significant difference in the CT values observed in liver cancer parenchyma scans at different phases between the two groups ( p > 0.05). Conclusion In summary, using a high-concentration iodine contrast agent is efficient in enhancing the visual clarity of the liver parenchyma, the aorta, and the portal vein in individuals diagnosed with primary hepatocellular carcinoma.
The aim of this study was to investigate the performance of eight digital radiography systems and to optimise the dose-image quality relationship for digital pelvis radiography. The study involved eight digital radiography systems used for general examinations at Vilnius University Hospital Santaros Klinikos. An anthropomorphic pelvic phantom (CIRS, US) was used to simulate a patient undergoing clinical pelvis radiography. Dose quantities entrance surface dose, dose area product (DAP) and exposure parameters (kVp, mA, mAs) were measured and the effects on the images were evaluated, considering physical contrast to noise ratio (CNR) and observer-based evaluations as image quality metrics. Increasing the tube voltage by 5 kVp from standard protocol led to a reduction in radiation dose (DAP) by 12%-20% with a slight impact on image quality (CNR decreases by 2%-10%). There was an inter-observer variability in image rating across different equipment (kappa value between 0 and 0.3); however, both observers agreed that increasing kVp up to 85-90 kV had no effect on perceived image quality. The results indicate that optimisation strategies should be tailored specifically for each x-ray system since significant performance differences and wide variations in radiation dose exist across various digital radiography systems used in clinical settings. The use of high kVp can be used for dose optimisation in digital pelvis radiography without compromising image diagnostic accuracy.
Currently, Sentiment Analysis (SA) has been gradually applied in a variety of fields and has become one of the most researched topics in adolescent education. However, since the interaction between cognition and emotion is involved in every learning process, it is possible to intervene with students based on the emotions they express in classroom or extracurricular environments, in order to assist teachers in assessing the overall state of students. This is conducive to improving teaching effectiveness, facilitating personalized learning, improving the emotional state and mental health of students, and promoting development and progress in the field of education. Emotion recognition is usually studied using electroencephalography (EEG), which is not practical for the adolescent population that spends most of their time at school almost every day. Therefore, in this paper, we propose an SA method based on a modified transformer network combined with convolutional neural network (CNN), aiming to utilize language for emotion recognition. The experiments were conducted using the Standford Sentinent Treebank (SST) dataset for training and validation of the model, which categorizes emotions into two categories based on positive and negative emotions, and ultimately obtains an overall accuracy of 95.00%. The experimental results demonstrate the recognition ability of our proposed model in sentiment analysis and show the potential for application in adolescent education.
Previous work has shown that isolated non-canonical sentences with Object-before-Subject (OSV) order are initially harder to process than their canonical counterparts with Subject-before-Object (SOV) order. Although this difficulty diminishes with appropriate discourse context, the underlying cognitive factors responsible for alleviating processing challenges in OSV sentences remain a question. In this work, we test the hypothesis that dependency length minimization is a significant predictor of non-canonical (OSV) syntactic choices, especially when controlling for information status such as givenness and surprisal measures. We extract sentences from the Hindi-Urdu Treebank corpus (HUTB) that contain clearly-defined subjects and objects, systematically permute the preverbal constituents of those sentences, and deploy a classifier to distinguish between original corpus sentences and artificially generated alternatives. The classifier leverages various discourse-based and cognitive features, including dependency length, surprisal, and information status, to inform its predictions. Our results suggest that, although there exists a preference for minimizing dependency length in non-canonical corpus sentences amidst the generated variants, this factor does not significantly contribute in identifying corpus sentences above and beyond surprisal and givenness measures. Notably, discourse predictability emerges as the primary determinant of constituent-order preferences. These findings are further supported by human evaluations involving 44 native Hindi speakers. Overall, this work sheds light on the role of expectation adaptation in word-ordering decisions. We conclude by situating our results within the theories of discourse production and information locality.
Literature in music theory and psychology shows that, even in isolation, musical sounds can reliably encode gender-loaded messages. Musical material can be imbued with many ideological dimensions and gender is just one of them. Nonetheless, studies of the gendering of music within multimodal communicative events are sparse and lack an encompassing theoretical framework. The present study attempts to address this literature gap by employing a critical quantitative analysis of music in gendered toy marketing, which integrated a content analytical approach with multimodal affective and music-focused perceptual responses. Ratings were collected on a set of 606 commercials spanning a ten-year time frame and strong gender polarization was observed in nearly all of the collected variables. Gendered music styles in toy commercials exhibit synergistic design choices, as music in masculine-targeted adverts was substantially more abrasive-louder, more inharmonious, and more distorted-than in feminine-targeted ones. Thus, toy advertising music appeared deliberately and consistently in line with traditional gender norms. In addition, music perceptual scales and voice-related content analytical variables explain quite well the heavily polarized affective ratings. This study presents a empirical understanding of the gendering of music as constructed within multimodal discourse, reiterating the importance of the sociocultural underpinnings of music cognition. We provided a public repository with all code and data necessary to reproduce the results of this study on github.com/marinelliluca/music-role-gender-marketing.
In the ever-evolving landscape of language education, this sociolinguistic study covers the discursive positioning of Indonesian adolescents in the intricate process of learning English. Language acquisition is a multifaceted phenomenon, often intertwined with complex social, cultural, and identity dynamics. This study covers the linguistic and sociocultural aspects of how Indonesian adolescents position themselves and are positioned by others within the discourse of English language learning. It used a qualitative design and conducted in-depth interviews with a diverse group of Indonesian adolescents aged 14-18 from various socio-economic backgrounds and educational settings. The analysis is rooted in positioning theory, a theoretical framework to examine how individuals construct their identities through language and interaction. The findings reveal a multifaceted picture of how Indonesian adolescents position themselves within the context of English language learning. Participants often engage in discursive practices that reflect their aspirations and struggles in mastering English. These practices are influenced by their social backgrounds, the role of English in their lives, and the educational settings they are part of. Furthermore, the study found distinct patterns of positioning, such as the "striver" positioning, where participants actively seek to align themselves with proficient English speakers, and the "resistant" positioning, where they push back against the pressure to conform to English language norms. The analysis also highlights the role of English language materials and classroom dynamics in shaping the discursive positioning of Indonesian adolescents. The materials often perpetuate certain linguistic norms and ideologies, impacting how students perceive themselves as language learners. Classroom interactions, on the other hand, provide a platform for students to negotiate their identities in relation to English. There is a need for a more nuanced and culturally sensitive approach to English language education in Indonesia.
The (inter-)dental non-sibilant fricatives, consonants articulated with the tongue tip or blade against or between the front teeth, are rare among the world’s languages but, nevertheless, are present in the sound inventories of some of the most spoken languages in existence. Here we try to shed light on the reason(s) for their distribution using multiple approaches, ranging from examining large cross-linguistic databases and phylogenetic reconstructions to the analysis of speech production data and anatomical measurements of rigid oral cavity structures obtained using intraoral 3D optical scanning. With these, we don’t only confirm that dental fricatives are rare among the present-day languages, but also that they have likely been so as far back as language families can be reconstructed, and that they are rarely borrowed between languages. The experimental data from L2 English speakers seem to suggest that details of the anatomy of the anterior vocal tract may play a role in the success of their acquisition. Therefore, dental fricatives are rare speech sounds for a multitude of reasons touching upon their articulation, acoustics, and confusability with other speech sounds, including the difficulty to produce in both L1 and L2 acquisition, the difficulty to perceive in L2 and in borrowing situations, and the rarity of attested sound changes producing them. Nevertheless, their frequent loss and merging with other phonemes in language change. Moreover, our data suggest that tiny, continuous, and overlapping patterns of variation in the anatomy of the anterior oral vocal tract may help explain their instability and geographic patterning.
The modulation of Autonomic Nervous System (ANS) dynamics is a fundamental aspect of emotional response. However, the dynamics of the interplay between ANS activity and subjective perception of emotional states are still an open question. In this preliminary study, we explored the causal relationship between continuously annotated emotional ratings and physiological time series. We used a subset of the publicly-available Continuously Annotated Signals of Emotions (CASE) dataset, focusing our analysis on time-varying self-assessed arousal ratings (AR), electrocardiographic signals, and electrodermal activity (EDA), simultaneously acquired from 30 healthy participants during two different video-based emotional stimuli, i.e. scary and relaxing. We applied the Convergent Cross Mapping (CCM) approach to investigate the causal links between EDA, AR, and heart rate variability (HRV) time series. The results were compared against surrogate series, generated using the Amplitude Adjusted Fourier Transform, to assess their significance. Regardless of the stimulation type, a statistically significant <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathbf{p}-\mathbf{value} < \mathbf{1}\mathbf{e}-\mathbf{14})$</tex> causal inference was found between the HRV and both the EDA and AR signals, and a statistically significant <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathbf{p}-\mathbf{value} < \mathbf{1}\mathbf{e}-\mathbf{18})$</tex> bidirectional coupling was observed between EDA and AR. Our findings support previous literature about physiologically-driven emotional processes, testing, for the first time, this hypothesis on the three-variable system of real-time arousal, HRV, and EDA recordings.
BACKGROUND: We analyzed online rating scores and comments of head and neck surgeons to understand factors that contribute to higher ratings. METHODS: Numerical ratings and comments for American Head and Neck Society physicians were extracted from Healthgrades, Vitals, RateMDs, and Yelp, with narrative comments categorized based on content. Physician practice location, education, and residency training were also compiled. RESULTS: Patient ratings were significantly higher with supportive staff and affable physician demeanor but showed significant drops with longer wait times and difficulties scheduling appointments or follow-ups. Physician education and postgraduate training did not significantly affect ratings. CONCLUSION: Online ratings and comments correlated to modifiable factors in clinical practice and may be informative in understanding patient needs.
Previous models for learning the semantic vectors of items and their groups, such as words, sentences, nodes, and graphs, using distributed representation have been based on the assumption that the basic sense of an item corresponds to one vector composed of dimensions corresponding to hidden contexts in the target real world, from which multiple senses of the item are obtained by conforming to lexical databases or adapting to the context. However, there may be multiple senses of an item, which are hardly assimilated and change or evolve dynamically following the contextual shift even within a document or a restricted period. This is a process similar to the evolution or adaptation of a living entity with/to environmental shifts. Setting the scope of disambiguation of items for sensemaking, the author presents a method in which a word or item in the data embraces multiple semantic vectors that evolve via interaction with others, similar to a cell embracing chromosomes crossing over with each other. We obtained a preliminary result: the role of a word that evolves to acquire the largest or lower-middle variance of semantic vectors tends to be explainable by the author of the text.
Nowadays, roadway infrastructures are designed in order to satisfy technical and economical requirements, as well as to guarantee advanced environmental performance. Focusing on that, this paper deals with an innovative procedure for the characterization of pavement materials, both asphalt and cement-bound mixtures. The methodology takes its cue from a previous study in which the so-called Environmental Asphalt Rating (EAR) was firstly introduced as a reference parameter for asphalt pavements to evaluate technical offers and for the assignment of scores, in terms of environmental impacts, during the tender phase. In this work, the EAR methodology is revised with a focus on the main variations and improvements related to the new version of the ISO standard. By applying the same approach to rigid or concrete pavements, a preliminary version of the Environmental Concrete Rating (ECR) is presented. For ECR, a correction is provided regarding functionality through a fatigue-related parameter and the surface characteristics related to the IRI value. Despite its strong applicability to the pavement sector, the strength of the proposed method is its ability to be fine-tuned to different fields by varying the associated performance coefficients.
Can bi-encoders, without additional fine-tuning, achieve a performance comparable to fine-tuned BERT models in classification tasks? To answer this question, we present a simple yet effective approach to text classification using bi-encoders without the need for fine-tuning. Our main observation is that state-of-the-art bi-encoders exhibit varying performance across different datasets. Therefore, our proposed approaches involve preparing multiple bi-encoders and, when a new dataset is provided, selecting and ensembling the most appropriate ones based on the dataset. Experimental results show that, for text classification tasks on subsets of the AG News, SMS Spam Collection, Stanford Sentiment Treebank v2, and TREC Question Classification datasets, the proposed approaches achieve performance comparable to fine-tuned BERT-Base, DistilBERT-Base, ALBERT-Base, and RoBERTa-Base. For instance, using the well-known bi-encoder model all-MiniLM-L12-v2 without additional optimization resulted in an average accuracy of 77.84%. This improved to 89.49% through the application of the proposed adaptive selection and ensemble techniques, and further increased to 91.96% when combined with the RoBERTa-Base model. We believe that this approach will be particularly useful in fields such as K-12 AI programming education, where pre-trained models are applied to small datasets without fine-tuning.
Verbs form the backbone of language, providing the structure and meaning to sentences.Yet, their intricate semantic nuances pose a longstanding challenge.Understanding verb relations through the concept of lexical entailment is crucial for comprehending sentence meanings and grasping verb dynamics.This work investigates the capabilities of eight Large Language Models in recognizing lexical entailment relations among verbs through differently devised prompting strategies and zero-/few-shot settings over verb pairs from two lexical databases, namely WordNet and Hy-perLex.Our findings unveil that the models can tackle the lexical entailment recognition task with moderately good performance, although at varying degree of effectiveness and under different conditions.Also, utilizing fewshot prompting can enhance the models' performance.However, perfectly solving the task arises as an unmet challenge for all examined LLMs, which raises an emergence for further research developments on this topic.
The article is an analysis of national stereotypes formed in Italy regarding the Chinese people. Italian anecdotes containing the lexemes “China”, “Chinese”, “Chinaman/Chinese woman” are selected as materials for the study. The author identifies the general characteristics of Italian anecdotes about the Chinese, studies the linguistic features of these texts and makes their thematic classification depending on the stereotype involved in them. The main themes of the anecdotes are peculiarities of gastronomic traditions and food habits of the Chinese; highly developed industry and a wide range of consumer goods; anthropological features of the Chinese people; copying of foreign brands by the Chinese; the number of Chinese in relation to the world population; personal qualities of the Chinese people; the influence of native language norms on the pronunciation of Italian words.The author concludes that the comic effect of Italian anecdotes is based mainly on the hyperbolisation of certain character traits of the Chinese, their patterns of behaviour and certain cultural and linguistic norms inherent in this nation, but not causing understanding among Italians. At the same time, it is noted that the humorous discourse does not contain aggression or insults, which makes it possible to conclude that Italians have a friendly perception of the Chinese people.
Fine-tuning Large Language Models (LLMs) for specific domains is crucial. However, lack of Thai open dialogues presents a major challenge. For the major challenge, this study proposes a novel methodology for extracting and constructing multi-turn conversational data from existing Thai large social platform, named Pantip. Our approach implements semantic matching algorithms to identify and compile both single-turn and multi-turn dialogues. By employing a cosine similarity threshold ≥ 0.3, we yields contextually coherent conversation pairs directly from the source data. The outcome dataset represents real Thai conversation styles, which could improve how accurately fine-tuned language models reflect Thai social and linguistic norms. Our approach introduces a new way to use existing publicly available data to create training datasets, which is especially valuable for languages with limited resources. Chaotic Pantip datasets can be contributed to the development of more culturally attuned and linguistically precise Thai language models, potentially advancing culturally-specific natural language processing.
Gender-fair language has been the subject of much recent research and insufficient consideration has been given to the negotiations of inclusive linguistic norms in everyday interactions. We build on the concept of grassroots linguistic activism to propose that community norms and affiliation around shared values influence the use of gender-fair language. We test these expectations following a corpus-assisted discourse approach, analysing ten YouTube videos on sustainable period products and their comment sections. We focus on the highly cisgendered period discourse to explore the extent to which gender-fair language has infiltrated mainstream usage. We find great variation in the awareness of gender diversity in the data. Only when gender-fair language is used in the videos, trans and non-binary people openly participate in the commentary, suggesting that linguistic invisibility leads to actual exclusion. Shared communal values of inclusivity, on the other hand, form the basis of successful grassroots linguistic activism and foster change in language use.
In the investigation of musical features that influence musical affect, timbre has received relatively little attention. Investigating affective timbres as they vary between instrument families can lead to inconsistent results, because one instrument family can produce a wide variety of timbres. Here, we consider timbre descriptors, as fine-grained acoustic representations of a sound. Using identical methods, we re-analyzed and synthesized results from three previously published studies: Eerola et al. (2012, Mus. Percept.), McAdams et al. (2017, Front. Psychol.), and Korsmit et al. (2023, Front. Psychol.). In doing so, we aimed to reveal robust timbre descriptors that consistently predict the affective response and to explain any discrepancies in results arising from differences in experimental methodology. We computed spectral, temporal, and spectro-temporal descriptors from all stimuli and used these to predict the affect ratings using linear and nonlinear methods. Our most consistent finding was that the fundamental frequency or higher-frequency energy of a sound predicted pleasant affect (i.e., positive valence, happiness, sadness) in one direction and unpleasant affect (i.e., tension, anger, fear) in the opposite direction. Clear discrepancies in previous findings may be attributable to differences in experimental design. When pitch variation was present in a stimulus set, energy arousal was predicted by pitch and inharmonicity, whereas when attack variation was present in the stimulus set, energy arousal was predicted by a faster attack and shorter sustain.
The aim of the study is to determine the quality of machine translation of English press releases into Russian in terms of semantic, stylistic and communicative adequacy. The article considers examples of communicative failures and analyses the reasons for the inadequacy of machine translation. The scientific novelty of the research consists in revealing violations of linguistic norms of the Russian language in the translated text of informative governmental press releases characterized by the high diplomatic style. The study found that communicative failures – violations of the norms of the Russian language – occur as a result of literal translation without analyzing the wider linguistic context, selection of an adequate meaning or significant translation transformations at the level of syntactic constructions, such as modulation, displacement, omission, addition, and others. Avoiding such communicative failures requires training of the machine translation system in the rules of the Russian language, English-Russian translation equivalents corresponding to the official business style, and replenishment of the language base.
Proportions are one of the primary components of successful image composition during the visual art creation process, which, in turn, is determinant of the variety of effects of images on the viewer, including emotional reactions, attention, and aesthetic preference. The importance of image width and height ratio is especially visible in the current trend to adopt the widest possible screens in a variety of modern creative media applications: photo, video, computer games, etc. In the present study emotional and aesthetic evaluations of the three most popular aspect ratios that are used in digital media devices were compared. This was achieved by assessing emotional arousal and valence ratings together with the interest and appeal evaluations of realistic photos presented in 4:3, 16:9, and 21:9 aspect ratios. The results demonstrated that the widest images did not have an inherent advantage – photos presented in the mid-wide aspect ratio of 16:9 could be considered as more effective, because they were rated as evoking the most positive emotional reactions and as the most liked pictures. This demonstrated that single design features can have an independent emotional effect, which needs to be considered in visual design aiming to evoke emotional reactions to the viewer.
Previous research has variably indicated the role of working memory in error detection by which working memory played a role in rhythmic error detection but not melodic error detection. Here, we devised a longer melodic error detection task for college musicians in an auditory, rather than visual, condition using classical excerpts, which we compared to briefer visual and auditory control conditions. These tests were compared to performance on a test of verbal working memory (forward digit span test) and an experimenter-created tonal working memory test. The tonal working memory test was positively related to the forward digit span test, the melodic error detection, and the visual control but not to the auditory control. Performance on the error detection test was not significantly related to year in school, level of aural skills class, years of private piano, or level of group piano class. Our participants performed similarly on the aurally presented melodic error detection of classical excerpts and the briefer visual control but not on the briefer aural control. Among other variables, years of experience on a second instrument was a significant predictor of error detection skill. High familiarity ratings with a classical excerpt did not yield a relationship to error detection performance.
The strongest formulations of grounded cognition assume that perceptual intuitions about concepts involve the re-activation of sensorimotor experience we have made with their referents in the world. Within this framework, concreteness and imageability ratings are indeed of crucial importance by operationalising the amount of perceptual interaction we have made with objects. Here we tested such an assumption by asking whether visual intuitions about concepts are provided accurately even when direct visual experience is absent. To this aim, we considered concreteness and imageability intuitions in blind people and tested whether these judgments are predicted by Image-based Frequency (IF, i.e. a data-driven estimate approximating the availability of the word referent in the visual environment). Results indicated that IF predicts perceptual intuitions with a larger extent in sighted compared to blind individuals, thus suggesting a role of direct experience in shaping our judgements. However, the effect of IF was significant not only in sighted but also in blind individuals. This indicates that having direct visual experience with objects does not play a critical role in making them concrete and imageable in a person's intuitions: people do not need visual experience to develop intuition about the availability of things in the external visual environment and use this intuition to inform concreteness/imageability judgments. Our findings fit closely the idea that perceptual judgments are the outcome of introspection/abstraction tasks invoking high-level conceptual knowledge that is not necessarily acquired via direct perceptual experience.
Shallow Parsing is an important step for many Natural Language Processing tasks. Although shallow parsing has a rich history for resource rich languages, it is not the case for most Indian languages. Shallow Parsing consists of POS Tagging and Chunking. Our study focuses on developing shallow parsers for Indian languages. As part of shallow parsing, we included morph analysis as well. For the study, we first consolidated available shallow parsing corpora for seven Indian Languages (Hindi, Kannada, Bangla, Malayalam, Marathi, Urdu, Telugu) for which treebanks are publicly available. We then trained models to achieve state-of-the-art performance for shallow parsing in these languages for multiple domains. Since analyzing the performance of model predictions at sentence level is more realistic, we report the performance of these shallow parsers not only at the token level, but also at the sentence level. We also present machine learning techniques for multi-task shallow parsing. Our experiments show that fine-tuned contextual embedding with multi-task learning improves the performance of multiple as well as individual shallow parsing tasks across different domains. We show the transfer learning capability of these models by creating shallow parsers (only with POS and Chunk) for Gujarati, Odia, and Punjabi for which no treebanks are available. As a part of this work, we will be releasing the Indian Languages Shallow Linguistic (ILSL) benchmarks for 10 Indian languages, including both the major language families Indo-Aryan and Dravidian as common building blocks that can be used to evaluate and understand various linguistic phenomena found in Indian languages and how well newer approaches can tackle them.
Abstract We examine how firm life cycle affects ratings and costs of debt for public offers. We find that ratings for issuers in the introduction and decline stages are lower than those for growth and mature issuers. A similar U‐shaped relation between life stage and yield spread, after controlling for credit rating, indicates that life stage affects cost of debt through multiple channels. Costs of debt are lower for growth and mature issuers than for introduction and decline issuers. Analyses of high‐yield bonds and term to maturity suggest that the adverse effect on costs of debt for introduction and decline firms is associated with their elevated riskiness and greater information asymmetry.