Search

The ability to assign meaning to perceptual stimuli forms the basis of human behavior and the ability to use language. The meanings of things have primarily been probed using behavioral production norms and corpus-derived statistical methods. However, it is not known to what extent the collection method and the language being probed influence the resulting semantic feature vectors. In this study, we compare behavioral with corpus-based norms, across Finnish and English, using an all-to-all approach. To complete the set of norms required for this study, we present a new set of Finnish behavioral production norms, containing both abstract and concrete concepts. We found that all the norms provide largely similar information about the relationships of concrete objects and allow item-level mapping across norms sets. This validates the use of the corpus-derived norms which are easier to obtain than behavioral norms, which are labor-intensive to collect, for studies that do not depend on subtle differences in meaning between close semantic neighbors. [ABSTRACT FROM AUTHOR]

Grounding language in vision is an active field of research seeking to construct cognitively plausible word and sentence representations by incorporating perceptual knowledge from vision into text-based representations. Despite many attempts at language grounding, achieving an optimal equilibrium between textual representations of the language and our embodied experiences remains an open field. Some common concerns are the following. Is visual grounding advantageous for abstract words, or is its effectiveness restricted to concrete words? What is the optimal way of bridging the gap between text and vision? To what extent is perceptual knowledge from images advantageous for acquiring high-quality embeddings? Leveraging the current advances in machine learning and natural language processing, the present study addresses these questions by proposing a simple yet very effective computational grounding model for pre-trained word embeddings. Our model effectively balances the interplay between language and vision by aligning textual embeddings with visual information while simultaneously preserving the distributional statistics that characterize word usage in text corpora. By applying a learned alignment, we are able to indirectly ground unseen words including abstract words. A series of evaluations on a range of behavioral datasets shows that visual grounding is beneficial not only for concrete words but also for abstract words, lending support to the indirect theory of abstract concepts. Moreover, our approach offers advantages for contextualized embeddings, such as those generated by BERT (Devlin et al, 2018), but only when trained on corpora of modest, cognitively plausible sizes. Code and grounded embeddings for English are available at (https://github.com/Hazel1994/Visually_Grounded_Word_Embeddings_2). [ABSTRACT FROM AUTHOR]

Speech enhancement performance has improved significantly with the introduction of deep learning models, especially methods based on the Long–Short-Term Memory architecture. However, these methods face challenges such as high computational complexity and redundancy of input features. To address these issues, we propose a U-Net-based approach that utilizes an encoder/decoder to extract more concise features, thereby enhancing single-channel speech performance and reducing computation complexity. The proposed method includes a Cross-Swin-Transformer block and a masked bottleneck module, which down-samples features while preserving the detailed representation through skip connections and carefully designed blocks. The bottleneck module extracts coarse representations of hidden features as masks. We evaluated our method against other U-Net-based approaches on VCTK and DNS corpora using CBAK, eSTOI, PESQ, STOI, and SI-SDR metrics. The results demonstrate that the proposed method achieves promising performance while significantly reducing computational complexity. [ABSTRACT FROM AUTHOR]

Based on the analysis of a linguistic corpus, this study uses relevance theory (Sperber and Wilson, 1986/1995) to demonstrate that the Tunisian Arabic particle ha :, which has never received any research attention, acts as an attitudinal marker in addition to its vocative function. Namely, ha: is restricted to utterances in which the speaker is expressing negative emotion or a subtype of positive emotion that can only be directed at the addressee. The data analysis shows that, unlike its relatively unmarked and well-studied counterpart ya: , ha: is restricted in terms of its frequency of distribution and the types of speaker emotions and attitudes expressed in the utterance in which it occurs. This study argues that ha: encodes a procedure which instructs the hearer to recognize the speaker's emotion or attitude expressed toward them in the utterance. The particle ha: therefore, serves as an explicit additional guarantee of relevance that encourages the hearer to look for additional contextual assumptions in order to achieve the speaker's desired interpretation. • The choice between the two vocative particles ha: and ya: is not arbitrary. • The Tunisian Arabic particle ha: acts as an attitudinal marker. • ha: encodes a procedure guiding the hearer to recognize the speaker's emotions. • The particle ha: serves as an explicit additional guarantee of relevance. • The particle ha : is no always compatible with the speaker's preferences. [ABSTRACT FROM AUTHOR]

The phenomenon of the 'complaint tradition', expressing the belief that language standards are deteriorating in some way, has had a long and vibrant existence in discussions of English in Hong Kong. The contribution of a Cultural Linguistics perspective here highlights the nature and significance of the cultural conceptualisations of language, for example, those that are revealed by the use of metaphor. The article begins with an overview of the discourse in Hong Kong, based on a corpus of media articles. The discourse structure and content of the articles, including their metaphorical aspects, are then analysed in more detail. To help understand the persistence of the discourse I draw on the insight that metaphors constitute 'mini‐narratives' which construct identities and identity boundaries. The discussion considers how the prevailing conceptualisations of English—some general, and others more culturally specific—serve to reproduce dominant, standardising views and inhibit the acceptance of local variation. [ABSTRACT FROM AUTHOR]

The learning quality of humans depends on the sense of contemplation. Textual documents are a huge part of the literature on contemplation which effortlessly creates perception. Automatic document class identification or organisation is a machine learning function to understand the psychological and emotional content of the text in a concise way. The problem of identification of documents falls in the field of library science, information science and artificial intelligence. The research progress of class identification of documents has been made in various most spoken languages. Numerous research works have been published in European and Asian languages. However, there is a gap in the literature when it comes to any less resource language, especially Bengali. Consequently, this work portrays an efficient topic modelling approach for Bengali document class identification. It proposes a Dirichlet-polynomial clustering model likelihood corpus distribution (LCD), which is based on a Bayesian numerical prototype. Experiments are done to prove the efficiency of LCD over various topic modelling algorithms, such as latent Dirichlet allocation (LDA), LDA with bag-of-words (LDA-BOW), latent semantic indexing (LSI), and hierarchical Dirichlet process (HDP). For performance evaluation, we considered five real-world datasets of Bengali corpora, such as science, sports, computer, season, and epic in this work. The coherence score of different modelling algorithms is compared to find the best model for each dataset separately. [ABSTRACT FROM AUTHOR]

The article discusses the integration of corpora into language teaching, emphasizing its benefits and the challenges in mainstream adoption. Topics include the underuse of corpora in teaching due to a lack of user-friendly resources, the potential for teachers to enhance their instructional decisions and materials through corpus-based insights, and the author's observations of students learning to use a corpus in an academic writing module.

The launch of the FrameNet project in 1997 was both a crystallisation point of decades worth of theoretical investigations into lexical meaning by Charles J. Fillmore and colleagues, as well as the seed of an ongoing line of corpus-based and computational research that seeks to implement Fillmore's theory of Frame Semantics in a way that both provides an interesting model relevant for further theorising and also is applicable practically for semantic analysis, lexicology, and lexicography. At the occasion of FrameNet's 25th birthday, we want to introduce the project to a new generation of researchers but also take stock of, and report on, what has been achieved. We revisit the origins of the FrameNet project, assess its development and various changes in the years since it was first reported on in a special issue in the pages of this journal twenty years ago. [ABSTRACT FROM AUTHOR]

This paper introduces readers to the TESOL Quarterly special topic issue: Assessing the effectiveness of corpus‐based approaches to English language teaching. The authors open by briefly defining and outlining data‐driven learning (DDL) research on second language (English) teaching. This is followed by an extended discussion of four themes, which pertain to recent developments and future directions in DDL. These themes include: (1) selecting corpora for corpus‐based language teaching; (2) pushing the boundaries of DDL; (3) preparing language educators for corpus‐based language teaching; and (4) reflecting on theory and methods. Throughout the piece, readers are also introduced to those contributions within the special issue, which include a diverse range of pieces that focus on aspects of theory, research, and pedagogy. [ABSTRACT FROM AUTHOR]

Data‐driven learning (DDL) is an increasingly popular area of research for language teaching and is gradually being incorporated into teacher education programs globally. However, assessing teacher trainees' (self‐reported) corpus literacy, and its impact on eventually (and successfully) incorporating corpora into language teaching, is an ongoing issue. This paper explores the experiences of in‐service English language teacher trainees in developing both corpus literacy and corpus‐based language pedagogy skills in Vietnam. The study finds a disconnect between participants' high self‐reported corpus literacy skills, high intention to conduct corpus‐based language teaching, and their ability to convert said intention and literacy into viable ideas for corpus‐based language teaching within their lesson planning. Potential reasons include difficulties conceptualizing DDL's integration into lesson objectives, selection of inappropriate corpora for the target audience, issues planning DDL activities for primary‐age learners, and Vietnamese‐specific issues mainly around access to technology. Accordingly, we present several recommendations for improving the assessment of DDL trainees' actual integration of DDL into their teaching practice. [ABSTRACT FROM AUTHOR]

The use of academic corpora in second language (L2) writing pedagogy has gained popularity in recent decades, particularly in genre‐specific contexts for graduate‐level L2 students (Charles, 2007; Lee & Swales, 2006). However, its overall effectiveness is mainly observed within classroom contexts, influenced by various contextual and participant‐related factors (Vyatkina & Boulton, 2017). Moreover, these investigations have predominantly relied on retrospective data useful for understanding learners' perception of corpus‐analysis experiences, but not for how learners strategically integrate corpora into personal toolkits for self‐directed writing tasks. Addressing these gaps, this study tracks five Chinese EFL graduate students' corpus use through a 5‐week concept‐ and corpus‐based tutoring intervention and investigates how they navigated a discipline‐specific corpus alongside other digital resources during post‐intervention self‐directed revisions. By triangulating screen recordings with retrospective data (questionnaires, interviews, draft revisions, and stimulated recalls), the results showed that while participants were generally favorable toward the specialized corpus and its complementary role to personal toolkits, individual problem‐solving patterns in revision revealed less overall reliance on the corpus and its effectiveness primarily in addressing known linguistic problems. Implications for a relational understanding of digital tools support and the integration of other pedagogical activities in L2 writing instruction are discussed. [ABSTRACT FROM AUTHOR]

Learner corpora have been used extensively in corpus research to identify gaps and errors within learner writing but have rarely been directly used in corpus‐informed instruction (CII). Importantly, scholars in CII have pointed out that "it is as important to see what learners can do as well as what they can't" when using learner corpora (Boulton & Thomas, 2012; p. 13). Learner corpora can also offer greater alignment with the genres students are writing (Seidlhofer, 2002). In our Brief Report, we take up Lu, Casal, and Liu's (2021) call for greater synergy of genre and corpus pedagogy by examining the impact of using a learner corpus to enhance students' language awareness and genre‐specific knowledge within an English as an Additional Language (EAL) first‐year writing classroom. We also move beyond error analysis to encourage an asset‐based model to learner CII (Staples, 2022). [ABSTRACT FROM AUTHOR]

This paper discusses the development and implementation of a locally‐sourced corpus to address the specific needs of an ESL writing program. The paper begins with a description of the motivation and development of the Corpus of Ohio Learner and Teacher English (COLTE), a large in‐house corpus consisting of assessed ESL student writing and teacher feedback from first‐year writing courses. The paper illustrates research conducted within the local context using the COLTE and discusses how the program's corpus‐based approach influenced curricular revisions and instructional practice. It addresses the evaluation of these efforts, challenges encountered, and considerations for building and using corpora at the program level. The paper concludes by arguing that stakeholders within programs can work collaboratively to build and use locally‐sourced corpora grounded in their specific setting to explore local practices and address the English language learning and teaching needs of their own particular context. [ABSTRACT FROM AUTHOR]

This paper analyses the methodologies in 148 empirical data‐driven learning studies for L2 English in prestige journals to examine best practice. Manual coding and corpus analysis of key words and n‐grams from the past 5 years (2018–2022) explore the field as a whole and how methodologies have evolved, suggesting improvements and future avenues for research. [ABSTRACT FROM AUTHOR]

Stylized neural machine translation (NMT) aims to translate sentences of one style into sentences of another style, it is essential for the application of machine translation in a real-world scenario. Most existing methods employ an encoder-decoder structure to understand, translate, and transform style simultaneously, which increases the learning difficulty of the model and leads to poor generalization ability. To address these issues, we propose a curriculum pre-training framework to improve stylized NMT. Specifically, we design four pre-training tasks of increasing difficulty to assist the model to extract more features essential for stylized translation. Then, we further propose a stylized-token aligned data augmentation method to expand the scale of pre-training corpus for alleviating the data-scarcity problem. Experiments show that our method achieves competitive results on MTFC and Modern-Classical translation dataset. [ABSTRACT FROM AUTHOR]

Few-shot relation classification (RC) is a crucial task that aims to identify the relationships between entity pairs with limited mentions. However, this task is challenging due to the insufficient amount of annotated data. Augmenting the quantity of annotated instances provides a reasonable solution, which is typically realized through distant supervision (DS). However, this approach may introduce substantial noise and negatively impact model performance. Conversely, prompt learning (PL) improves the quality of annotated instances by enabling the models to understand the input text thoroughly with templates. Unfortunately, these templates mainly neglect the semantics of labels. To address these issues, we propose coarse-to-fine filters with concept heuristic prompt (CoFF-CHP), which employs dual-stage filters to select and label reliable instances from unlabeled corpus. Specifically, we first employ the coarse-grained filters to preliminarily select the instances. Next, we design a fine-grained filter, concept heuristic prompt (CHP), which can integrate the semantics of the label concept to help the model understand the link between the input text and labels. Finally, to reduce the tilt toward positive instances, we incorporate the screen-out data and false-positive instances to form the negative set to correct the classifier. The experimental results on the public Fewrel dataset indicate that CoFF-CHP achieves state-of-the-art performance in low-resource scenarios, especially when the best baseline is upgraded by 15.06% in terms of F1 in the setting of 10 seed instances. [ABSTRACT FROM AUTHOR]

The Persian verb "xâstan" exhibits a dual nature, functioning both as an auxiliary verb and a lexical verb. As a lexical verb, it conveys a range of meanings expressed through various syntactic structures. This research delved into the syntactic and semantic aspects of the mentioned verb in different constructions. The authors proposed that Role and Reference Grammar (RRG), with its focus on complex structures, could provide a comprehensive framework for analyzing the constructions under consideration. RRG conceptualizes complex structures based on "Interclausal Relations Hierarchy", which comprises two syntactic and semantic continua. The syntactic continuum examines juncture -nexus relations, while the semantic continuum explores the type of semantic relations based on the main predicate. This study endeavored to describe the various applications of the verb "xâstan" within the mentioned framework. Additionally, the research findings were employed to evaluate the tenets of RRG. The findings revealed that, semantically, the verb "xâstan" fell within the categories of psych -action, jussive, and propositional attitude predicates. Syntactically, it manifested in the forms of core cosubordination and core subordination. The data for this research were extracted from the corpus of Hamshahri 2. [ABSTRACT FROM AUTHOR]

We report the discovery of a vast corpus of indigenous legal contracts (‘arraten’) in the Morocco Atlas Mountains. The documents comprise about 400 years of written records. A collection of this magnitude has never before been reported, and such documents have not as of yet been systematically amassed for examination, preservation and study. The amount of documents discovered by us, and remaining in situ, is likely to exceed 5000 items. This is in addition to our own private collection which is in excess of 4000 pieces. We estimate that the entire body of documents written on wood in the Anti-Atlas Mountains is likely to be no less than 100,000 pieces and may very well exceed 500,000. In addition to this, between 40,000 and 150,000 paper documents likely exist. Here we report our preliminary studies of those manuscripts. Our collection is available to anyone and donation to suitable institutions will be considered. [ABSTRACT FROM AUTHOR]

Boris Pahor was born in 1913 in Trieste, where he died at the age of 108. In all his works, he tries to reconstruct the conditions of exclusion experienced by himself, his family, his people, and finally all those who resisted the ideological pressures at the height of the Nazi terror, as described in the novel Necropolis (1967). The extreme violence with which assimilation was imposed on the Slovenian minority did not allow him to preserve his original identity. Pahor’s Trieste confirms that spaces exposed to a constant pressure of assimilation are spaces of undeniable vitality. This perspective should allow us to create a corpus of knowledge that reflects the resilience of cultures exposed to domination. Instead of seeing the margins as a place of exclusion and endemic difficulties, we need to change the narrative that defines them. Margins are the spaces where translation constantly materializes, the spaces of exchanges and transfers, thus the periphery has the capacity for mediation between complex cultural systems. [ABSTRACT FROM AUTHOR]

This study employs a corpus-based approach to examine and compare the use of two discourse markers (DMs), "you know" and "I mean", within the context of two mediatised English political interviews. The analysis encompasses frequencies, functions, co-occurrences, and positional distributions of these DMs. The study utilizes specialized corpora from two political interview programs: CGTN's The Point with Liu Xin and BBC's HARDtalk. The frequency analysis reveals that "you know" is statistically more prevalent than "I mean" in both programs, reflecting the spontaneity, interactivity, and need for clarification characteristic of political interviews. Notably, the Chinese interviewer (IR) uses "you know" more extensively, possibly due to a cultural preference for ensuring mutual understanding and engaging the audience, while the British IR employs "I mean" slightly more frequently, likely reflecting a tendency to clarify or reframe statements for precision. Functionally, these DMs serve diverse purposes such as hedging, agreeing, and monitoring across various domains including interpersonal, sequential, and rhetorical. Positional analysis shows "you know" typically appearing medially and "I mean" often in initial positions. These results underscore the distinctive interviewing styles of the two IRs and the pivotal role of these DMs in fulfilling a spectrum of communicative functions. This research offers valuable insights into the interviewer's perspective in political interviews. [ABSTRACT FROM AUTHOR]

As a corpus-assisted method for language pedagogy, DDL (data-driven learning) may have the potential to enhance language exposure and promote active learner engagement. Concordancing, or KWIC (Key Words in Context), has been the traditional method used in DDL to retrieve numerous language examples, while the method has limitations with overreliance on individual words to search. This paper aims to propose and promote concordancing alternatively based on grammar patterns, a multi-word concept in corpus linguistics. The conceptualised method is named as PIC (Patterns in Context), an extended form of KWIC. An empirical study was conducted to investigate whether the PIC method has any advantages over the traditional KWIC method, using two custom-built Android apps. The research involved 18 pre-university intermediate learners (and six pilot study participants), who used the apps in a self-directed way for two weeks. Then the assessment of the two apps and methods was conducted based on data from automatic logs and responses from questionnaires and interviews. The results suggest that, compared to KWIC, PIC could be slightly advantageous in efficiently helping learners find the target language use, while this approach seems not strong in user engagement and perceived effectiveness. The implications for DDL are discussed, and further investigation is also planned. [ABSTRACT FROM AUTHOR]

This article examines the depiction of the Belt and Road Initiative (BRI) in elite English media across Europe to discern sentiments and ideological disparities between China and the Western world. Using the PRISMA method, 1826 relevant research articles on the BRI were identified, and 42 were selected for detailed analysis. Additionally, 1357 news articles from the Financial Times, The Guardian and BBC from 2015 to 2023 underwent linguistic corpus analysis. Employing a three-dimensional critical discourse analysis (3d-CDA), the study scrutinized sample news articles, revealing ideological underpinnings across various dimensions and frames. The analysis indicates a growing negativity towards the BRI in European English media, with sentiments averaging 38% from 2015 to 2023 and peaking at 56% and 54% in 2022 and 2023, respectively. Ideological differences often discussed in academic circles include a contrast between the emphasis on individualism in the West and the values of collectivism prominent in Chinese culture. However, it is important to recognize the diversity within these broad categories. This research offers insights for BRI stakeholders and enriches academic discourse with updated data. Its novelty lies in its dual-tiered approach to news relevance for content analysis and the application of three-dimensional CDA to BRI news discourse, revealing ideological disparities not previously explored. [ABSTRACT FROM AUTHOR]

While slope is a topic in the algebra curriculum, having a robust understanding of slope is needed for students to truly understand several single and multivariable calculus topics with any depth. We begin with a review of the topic of slope and present what is known from its existing corpus of literature. We then outline the tenets of APOS theory. Building from there, we suggest what a robust, flexible understanding of slope involves, as well as how slope is used, with the APOS-slope framework acting as a theoretical lens. This is followed by the cases of two hypothetical students built from amalgamations of research and experience to emphasize why moving easily between different ways of thinking about and the various uses of slope is vital to successfully transition into calculus. We offer suggestions as to how university instructors might consider slope understanding when teaching calculus, then conclude with suggestions for future research on slope. [ABSTRACT FROM AUTHOR]

Idiomatic expressions are not just neutral equivalents of their literal counterparts, but exert some sort of evaluative force. This study employed the appraisal system (AS) to explore the functions of idiomatic expressions and their variations in English as a lingua franca context, and to compare different academic domains in terms of use of evaluative language in the form of idiomatic expressions. To this end, the corpus of English as a lingua franca in academic settings and the Vienna–Oxford International Corpus of English were searched for idiom tokens through whole‐phrase and keyword search. The idioms and their variations were classified based on the AS. The findings indicated that social and behavioral scientific texts included higher proportions of evaluative idiomatic expressions whereas medical and natural scientific texts generated lower percentages of appraisal idioms. There was a stronger tendency in both corpora toward using idioms as appreciation as compared to judgment and affect. Given the employed methodology, the AS could not be applied to some idioms for their compositionality nature. [ABSTRACT FROM AUTHOR]

Objective: This study was undertaken to delineate 21‐year sex‐specific trends in recurrence and postrecurrence mortality. Methods: Between 2000 and 2020, first‐ever ischemic stroke (IS) patients, ascertained from the population‐based BASIC (Brain Attack Surveillance in Corpus Christi) project in South Texas, were followed for recurrent stroke and all‐cause mortality until December 31, 2020. Multivariable regression models with an interaction between calendar year and sex were used to estimate sex‐specific trends and sex differences in recurrence and postrecurrence mortality. Results: Of the 6,057 IS patients (median age = 69 years, 49.8% women), 654 (10.8%) had a recurrence and 399 (47.7%) had postrecurrence mortality during 5 years of follow‐up. In 2000, women had 2.5% higher albeit non‐statistically significant 5‐year risk of recurrence than men in absolute scale. With the trend declining in women by 7.6% (95% confidence interval [CI] = −10.8 to −4.5%) and in men by 3.6% (95% CI = −6.5% to −0.7%), the risk at the end of the study period was 1.5% (95% CI = −0.3% to 3.6%) lower among women than men. For postrecurrence mortality, the risk was 10.2% lower among women in 2000, but the sex difference was 3.3% by the end of the period, which was due to a larger overall increase in the risk among women than men over the entire time period. Interpretation: The declines in recurrent stroke suggest successful secondary stroke prevention, especially in women. However, the continued high postrecurrence mortality among both sexes at the end of study period emphasizes the need for ongoing interventions to improve prognosis in those who have had recurrent cerebrovascular events. ANN NEUROL 2024;96:332–342 [ABSTRACT FROM AUTHOR]

Analysis of online user-generated reviews has attracted extensive attention with broad applications in recent years. However, the high-volume and low-value density of online reviews bring challenges to a timely and effective data utilization. To address that challenge, this work proposes an unsupervised review filtering method based on the inherent tree-structured hierarchies among review data that reflect the general-to-specific characteristics of various quality aspects discussed in reviews. In particular, the reviews with aspects distributed near the leaf nodes of the tree are capable of providing more specific and detailed information about the examined product, which is more likely to be reserved after the tree-based filtering. To enable an effective extraction of aspect hierarchies from a broad variety of review corpora, a Bayesian nonparametric hierarchical topic model has been constructed and incorporated with an enhanced Pólya urn scheme. The approximate inference of model parameters is obtained by an efficient collapsed Gibbs sampling procedure. The proposed method can enhance the layered effect of individual reviews according to their general-to-specific characteristics and reserve an information-rich subset filtered from the raw review corpus. The merits of the proposed method have been elaborated by case studies on two real-world data sets and an extensive simulation study. [ABSTRACT FROM AUTHOR]

This study aims to examine Jordanian males’ comments on females’ physical appearance on Facebook to identify the impoliteness strategies that Jordanian males utilize in online communication. The socio-cultural factors affecting the selection of impoliteness behavior in males’ comments are also investigated. As for the corpus of the study, a total of 470 comments and replies were chosen from 32 females’ picture posts posted on two different Jordanian groups on Facebook. This study draws on Culpeper’s (1996) impoliteness framework. All Culpeper’s (1996) impoliteness strategies except withholding impoliteness (i.e., bald-on-record, sarcasm/mock, negative, positive) were employed. It was also found that making a clear boundary between different impoliteness strategies is not easy; more than one impoliteness strategy can be realized by the same expression. Furthermore, males were found to use other additional impoliteness strategies not mentioned in Culpeper’s (1996) impoliteness framework, namely, third-party implicit impoliteness, supplications, and negative interjections. The data analysis also revealed that some comments were subsumed under new categories not mentioned in previous (im)politeness frameworks: ambiguous (im)politeness and adversative (im)politeness. The use of such various strategies can be ascribed to the effect of socio-cultural and religious background on males’ behavior. [ABSTRACT FROM AUTHOR]

An English-Bengali machine translation (MT) application can convert an English text into a corresponding Bengali translation. To build a better model for this task, we can optimize English-Bengali MT. MT for languages with rich resources, like English-German, started decades ago. However, MT for languages lacking many parallel corpora remains challenging. In our study, we employed back-translation to improve the translation accuracy. With back-translation, we can have a pseudo-parallel corpus, and the generated (pseudo) corpus can be added to the original dataset to obtain an augmented dataset. However, the new data can be regarded as noisy data because they are generated by models that may not be trained very well or not evaluated well, like human translators. Since the original output of a translation model is a probability distribution of candidate words, to make the model more robust, different decoding methods are used, such as beam search, top-k random sampling and random sampling with temperature T, and others. Notably, top-k random sampling and random sampling with temperature T are more commonly used and more optimal decoding methods than the beam search. To this end, our study compares LSTM (Long-Short Term Memory, as a baseline) and Transformer. Our results show that Transformer (BLEU: 27.80 in validation, 1.33 in test) outperforms LSTM ( 3.62 in validation, 0.00 in test) by a large margin in the English-Bengali translation task. (Evaluating LSTM and Transformer without any augmented data is our baseline study.) We also incorporate two decoding methods, top-k random sampling and random sampling with temperature T, for back-translation that help improve the translation accuracy of the model. The results show that data generated by back-translation without top-k or temperature sampling ("no strategy") help improve the accuracy (BLEU 38.22 , + 10.42 on validation, 2.07 , + 0.74 on test). Specifically, back-translation with top-k sampling is less effective ( k = 10 , BLEU 29.43 , + 1.83 on validation, 1.36 , + 0.03 on test), while sampling with a proper value of T, T = 0.5 makes the model achieve a higher score ( T = 0.5 , BLEU 35.02 , + 7.22 on validation, 2.35 , + 1.02 on test). This implies that in English-Bengali MT, we can augment the training set through back-translation using random sampling with a proper temperature T. [ABSTRACT FROM AUTHOR]

Previous academic research has argued that videogames are represented as violent in the media, and that these representations provide a justification to study videogames as a social object. However, little systematic research has been done to confirm whether such representations occur frequently, or whether this is a non-representative assumption based on a small number of news stories. Through using methods associated with Modern-Diachronic Corpus-Assisted Discourse Studies (md-cads), we explore two corpora of newspaper articles from the British press containing references to videogames: one contains data from 2000 and the other from 2020. We find that whilst newspapers do represent videogames as promoting violence, these representations occur much more frequently in 2000 than in 2020, and that in 2020 this idea is more regularly challenged. Furthermore, in 2020 there is a greater focus on the benefit of videogames for the economy compared to 2000. The research highlights the need to critically revaluate assumptions about how the media represents not only videogames, but also comparable topics. [ABSTRACT FROM AUTHOR]

In complex stimuli, there are many different possible ways to refer to a specified target. Previous studies have shown that when people are faced with such a task, the content of their referring expression reflects visual properties such as size, salience, and clutter. Here, we extend these findings and present evidence that (i) the influence of visual perception on sentence construction goes beyond content selection and in part determines the order in which different objects are mentioned and (ii) order of mention influences comprehension. Study 1 (a corpus study of reference productions) shows that when a speaker uses a relational description to mention a salient object, that object is treated as being in the common ground and is more likely to be mentioned first. Study 2 (a visual search study) asks participants to listen to referring expressions and find the specified target; in keeping with the above result, we find that search for easy-to-find targets is faster when the target is mentioned first, while search for harder-to-find targets is facilitated by mentioning the target later, after a landmark in a relational description. Our findings show that seemingly low-level and disparate mental "modules" like perception and sentence planning interact at a high level and in task-dependent ways. [ABSTRACT FROM AUTHOR]

Copyright of Jahrbuch für Germanistische Sprachgeschicht is the property of De Gruyter and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Jahrbuch für Germanistische Sprachgeschicht is the property of De Gruyter and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Deep learning has made significant advancements in speech enhancement, which plays a crucial role in improving the quality of speech signals in noisy conditions. In this paper, we propose a new approach called M-DGAN, which introduces a time (T)-domain encoder-decoder structure with rich channel representations into the time-frequency (TF)-domain generator framework, resulting in a new generator structure with mixed magnitude and phase representations in the T and TF-domains. The proposed mixed T-domain and TF-domain generator, incorporating the cascaded reworked conformer (CRC) structure, exhibits improved modeling capability and adaptability. Test results on the Voice Bank + DEMAND public dataset show that our method achieves the highest score with P S E Q = 3.52 and performs well on all the remaining metrics when compared to the current state-of-the-art methods. In addition, tests on the NISQA_TEST_LIVETALK real dataset of the NISQA Corpus show the breadth and robustness of our model on speech enhancement tasks. [ABSTRACT FROM AUTHOR]

While it is well known today that J.S. Bach’s The Well-Tempered Clavier (WTC) was not composed with equal temperament in mind, it is unknown exactly which, if any, well temperament Bach had intended for the work. While equal temperament produces 12 uniform intervals of each type, other systems of well temperament slightly stretch and compress the size of various fifths, resulting in alterations to other intervals. A number of competing hypotheses have been published as to which temperament Bach intended, or whether he intended a specific temperament at all. While we do not have a definitive historical answer as to Bach’s preference of temperament in this work, this paper applies corpus methodologies to compare these systems. This paper develops and utilises Detuning-Emphasis Correlation (DEC): the relationship between scale-degree pairs' deviation from just intonation and their frequency of use. Frequency of use is compared to a ‘baseline’ frequency, found by averaging a scale degree pair’s prevalence across the corpus. This paper analyses the 48 fugues of Bach’s WTC by computing DECs for selected well temperaments (Werckmeister III, Bach-Lehman, Niedhardt, Vallotti, and Young), aiming to find a significant negative DEC for some temperament. While no temperament was found to fit the WTC in this way, strong correlations between these well temperaments indicated that the work would highlight similar affects for the various keys regardless of which (non-equal) well temperament is used to realise this work. [ABSTRACT FROM AUTHOR]

Despite the high creation cost, annotated corpora are indispensable for robust natural language processing systems. In the clinical field, in addition to annotating medical entities, corpus creators must also remove personally identifiable information (PII). This has become increasingly important in the era of large language models where unwanted memorization can occur. This paper presents a corpus annotated to anonymize personally identifiable information in 1,787 anamneses of work-related accidents and diseases in Spanish. Additionally, we applied a previously released model for Named Entity Recognition (NER) trained on referrals from primary care physicians to identify diseases, body parts, and medications in this work-related text. We analyzed the differences between the models and the gold standard curated by a physician in detail. Moreover, we compared the performance of the NER model on the original narratives, in narratives where personal information has been masked, and in texts where the personal data is replaced by another similar surrogate value (pseudonymization). Within this publication, we share the annotation guidelines and the annotated corpus. [ABSTRACT FROM AUTHOR]

Citation networks enable analysis of author groups, defining in-group dynamics, and mapping out inter-group relationships. While intellectual diversity and inclusiveness is one of the important principles of modern scholarship, it is intriguing to explore the extent to which these principles apply to historical communities of leaders and intellectuals. This paper introduces a novel methodological framework aimed at assessing the degree of viewpoint plurality and diversity of historical scholarship communities, through an in-depth analysis of the citations used in their literature, which has become possible due to the recently developed advanced computational analysis techniques. To achieve this goal, we have devised a set of new network-based indicators grounded in standard network metrics. These indicators can be applied at both the individual author and community levels. The developed methodology was applied to a citation network automatically constructed from a corpus of Rabbinic Halachic literature spanning the 10th to 15th centuries. This corpus includes over 5,000 citations from hundreds of books authored by approximately 140 Rabbinic scholars from six diverse geographic communities. We found that most of the authors and communities cite many more external resources from other communities than their own reflecting a willingness to engage with a diverse range of viewpoints. A more in-depth analysis based on the novel proportional diversity measures unveils more intriguing insights. Contrary to expectations, communities with the greatest number of external citations, such as Spain and Ashkenaz, surprisingly exhibit lower levels of viewpoint plurality compared to others, such as Italy and North Africa, elucidating a key finding of the study. [ABSTRACT FROM AUTHOR]

Metaphorical expressions are widely present in natural language, posing significant challenges to a variety of natural language processing tasks such as machine translation. How to obtain richer contextual representations is an urgent problem to be solved. To address this issue, this paper proposes a model that combines syntax-aware local attention (SLA), a simple contrastive sentence embedding framework (SimCSE), and linguistic theories, called a combination of syntax-aware and semantic methods (CSS). Specifically, we apply linguistic theory in metaphor detection. Additionally, we simultaneously conduct metaphor identification and contrastive learning tasks. The SimCSE contrastive learning framework effectively captured more information, and the concurrent execution of these two tasks helped increase the sensitivity of the semantic space to metaphors. The integration of SLA with the pre-trained language model BERT enhanced the attention weights between grammatically relevant words, assisting the encoder in focusing more on grammar-related words. Overall, CSS prioritizes the sentence itself, avoiding the introduction of excessive additional information. Experimental results on the VU Amsterdam-verb (VUA), TroFi, and MOH-X metaphorical corpora show that our method is superior to state-of-the-art models. [ABSTRACT FROM AUTHOR]

This paper approaches variable rhoticity in Scottish Standard English (SSE) from a methodological, data‐oriented perspective. The main focus is on how to integrate within a single sociolinguistic framework data that have been elicited under different conditions (sociolinguistic interviews vs. corpus data) and may therefore be incompatible when we take a relatively simplistic approach to stylistic variation. The article first models variation in such an unbalanced ‘patchwork’ dataset in a holistic, unified analysis and then proceeds to deconstruct the results by taking a closer look at the behaviour of the two main components of the data. While previously reported trends in the development of coda‐/r/ in SSE accents are confirmed, it is also shown that, depending on the precise nature of registers (or genres), productions may be considerably more diverse and may go beyond a simple projection of middle‐class identities. The integration of different styles in a single interpretive framework is challenging and far from straightforward. [ABSTRACT FROM AUTHOR]

This article presents a transfer-learning-based method to improve the synthesized speech quality of the low-resource Dungan language. This improvement is accomplished by fine-tuning a pre-trained Mandarin acoustic model to a Dungan language acoustic model using a limited Dungan corpus within the Tacotron2+WaveRNN framework. Our method begins with developing a transformer-based Dungan text analyzer capable of generating unit sequences with embedded prosodic information from Dungan sentences. These unit sequences, along with the speech features, provide pairs as the input of Tacotron2 to train the acoustic model. Concurrently, we pre-trained a Tacotron2-based Mandarin acoustic model using a large-scale Mandarin corpus. The model is then fine-tuned with a small-scale Dungan speech corpus to derive a Dungan acoustic model that autonomously learns the alignment and mapping of the units to the spectrograms. The resulting spectrograms are converted into waveforms via the WaveRNN vocoder, facilitating the synthesis of high-quality Mandarin or Dungan speech. Both subjective and objective experiments suggest that the proposed transfer learning-based Dungan speech synthesis achieves superior scores compared to models trained only with the Dungan corpus and other methods. Consequently, our method offers a strategy to achieve speech synthesis for low-resource languages by adding prosodic information and leveraging a similar, high-resource language corpus through transfer learning. [ABSTRACT FROM AUTHOR]

Featured Application: The model is primarily utilized for the task of entity relationship extraction during the construction process of an aquatic disease knowledge graph. Entity–relationship extraction plays a pivotal role in the construction of domain knowledge graphs. For the aquatic disease domain, however, this relationship extraction is a formidable task because of overlapping relationships, data specialization, limited feature fusion, and imbalanced data samples, which significantly weaken the extraction's performance. To tackle these challenges, this study leverages published books and aquatic disease websites as data sources to compile a text corpus, establish datasets, and then propose the Fd-CasBGRel model specifically tailored to the aquatic disease domain. The model uses the Casrel cascading binary tagging framework to address relationship overlap; utilizes task fine-tuning for better performance on aquatic disease data; trains on specialized aquatic disease corpora to improve adaptability; and integrates the BRC feature fusion module—which incorporates self-attention mechanisms, BiLSTM, relative position encoding, and conditional layer normalization—to leverage entity position and context for enhanced fusion. Further, it replaces the traditional cross-entropy loss function with the GHM loss function to mitigate category imbalance issues. The experimental results indicate that the F1 score of the Fd-CasBGRel on the aquatic disease dataset reached 84.71%, significantly outperforming several benchmark models. This model effectively addresses the challenges of ternary extraction's low performance caused by high data specialization, insufficient feature integration, and data imbalances. The model achieved the highest F1 score of 86.52% on the overlapping relationship category dataset, demonstrating its robust capability in extracting overlapping data. Furthermore, We also conducted comparative experiments on the publicly available dataset WebNLG, and the model in this paper obtained the best performance metrics compared to the rest of the comparative models, indicating that the model has good generalization ability. [ABSTRACT FROM AUTHOR]

Many research articles have explored the impact of surgical interventions on voice and speech evaluations, but advances are limited by the lack of publicly accessible datasets. To address this, a comprehensive corpus of 107 Spanish Castilian speakers was recorded, including control speakers and patients who underwent upper airway surgeries such as Tonsillectomy, Functional Endoscopic Sinus Surgery, and Septoplasty. The dataset contains 3,800 audio files, averaging 35.51 ± 5.91 recordings per patient. This resource enables systematic investigation of the effects of upper respiratory tract surgery on voice and speech. Previous studies using this corpus have shown no relevant changes in key acoustic parameters for sustained vowel phonation, consistent with initial hypotheses. However, the analysis of speech recordings, particularly nasalised segments, remains open for further research. Additionally, this dataset facilitates the study of the impact of upper airway surgery on speaker recognition and identification methods, and testing of anti-spoofing methodologies for improved robustness. [ABSTRACT FROM AUTHOR]

In academic writing, exemplification plays a crucial role in contextualizing complex ideational material through instances the readers can understand. In addition to illustrating ideas through concrete instances, the act of providing examples serves the purpose of helping the readers grasp the writer's intentions. However, strategically performing exemplification to elaborate the propositional material seems to be a challenge for many novice student writers. Although some studies have mentioned that students use significantly less frequently the exemplification resources in their writing, fewer studies have probed into EFL student writing to determine the problems they face in elaborating the ideas. Using the marker approach, which examines the discourse functions bottom‐up from markers to moves, the learner corpus of essay writing (LCEW) was analyzed for three major forms of exemplification: representation, argumentation, and analogy. The results indicate that the examples are strictly limited to certain patterns like specifying concepts through a subcategory and illustrating the arguments through everyday experiences. Moreover, many examples deviate from the usual patterns of exemplification causing confusion. These findings have pedagogic implications for academic writing courses in the EFL context. [ABSTRACT FROM AUTHOR]

Objective Methods Results Conclusion In narrow anterior urethral strictures, the combined buccal mucosa graft (BMG) with pedicled penile skin flap (PSF) represents a well‐known effective alternative to staged urethroplasty. We hypothesized that if the native urethral plate and adjacent corpus spongiosum were preserved, a narrower flap would be needed, and reinforced ventral stability could be achieved without compromising the surgical outcome.Twelve patients with narrow penile urethral strictures underwent single‐stage augmentation urethroplasty using a combined technique. A BMG was quilted to the corpora cavernosa in a dorsal onlay approach, and a longitudinal ventral PSF was transposed ventrally and sutured to the scarred native urethral mucosa on one side and to the BMG on the other side to form a neourethra of triangular form. The preserved corpus spongiosum was wrapped and fixed around the flap ventrally.The median age was 47 years (IQR 35–59), and the median stricture length was 5 cm (IQR 3, 8–7). The median surgical time was 205 min (IQR 172–236). The overall success rate (SR) was 91.7% without sacculation or diverticula formation after a median follow‐up period of 38 months (IQR 33–40). Three transient fistulas healed through prolonged urinary diversion. Five patients (41.7%) reported postvoid dribbling following urethroplasty.Preservation of the native urethral plate is a valuable adjunct to the combination of graft and flap for single‐stage augmentation urethroplasty for narrow urethral strictures, with satisfactory mid‐term success and an acceptable complication rate. [ABSTRACT FROM AUTHOR]

Research Highlights There is substantial evidence that children's apparent omission of grammatical morphemes in utterances such as “She play tennis” and “Mummy eating” is in fact errors of commission in which contextually licensed unmarked forms encountered in the input are reproduced in a context‐blind fashion. So how do children stop making such errors? In this study, we test the assumption that children's ability to recover from error is related to their developing sensitivity to longer‐range dependencies. We use a pre‐registered corpus analysis to explore the predictive value of different cues with regards to children's verb‐marking errors and observe a developmental pattern consistent with this account. We look at context‐independent cues (the identity of the specific verb being used) and at the relative value of context‐dependent cues (the identity of the specific subject+verb sequence being used). We find that the only consistent effect across a group of 2‐ to 3‐year‐olds and a group of 3‐ to 4‐year‐olds is the relative frequency of unmarked forms of specific subject+verb sequences being used. The relative frequency of unmarked forms of the verb alone is predictive only in the younger age group. This is consistent with an account in which children recover from making errors by becoming progressively more sensitive to context, at first the immediately preceding lexical contexts (e.g., the subject that precedes the verb) and eventually more distant grammatical markers (e.g., the fronted auxiliary that precedes the subject in questions). We provide a corpus analysis investigating input effects on young children's verb‐marking errors (e.g., Mummy go) across development (between 2 and 4 years of age). We find evidence that these apparent errors of omission are in fact input‐driven errors of commission that persist into the third year of life. We compare the relative effect on error rates of context‐independent (e.g., verb) and context‐dependent (e.g., subject+verb sequence) cues across developmental time. Our findings support the proposal that children recover from making verb‐marking errors by becoming progressively more sensitive to preceding context. [ABSTRACT FROM AUTHOR]

Historical markers offer important insights into the portrayal of place, social dynamics, and untold narratives within various landscapes. The vast collection of over 195,000 historical markers scattered throughout the United States provide a wealth of data that can shed light on distinctive representation patterns and narratives associated with different individuals and events. Yet, only a few studies examined historical markers as a valuable data source. The goal of this study is to investigate how Texas women are commemorated through historical markers and to uncover the spatial and temporal patterns and contexts associated with these commemorations. We utilize qualitative GIS method and a suite of corpus linguistics techniques to analyze the spatial, temporal, and thematic patterns of the historical markers in Texas that are dedicated to women. Our findings address the existing gender imbalance and shed light on the underrepresentation of women in Texas' historical narratives. [ABSTRACT FROM AUTHOR]

This study examines patterns of disaster commemoration in Texas as witnessed in 1,590 markers erected by the state between 1936 and 2019. In the first half of the analysis, we highlight spatiotemporal trends of commemoration, which focus on a few well-known historic disasters that occurred between 1875 and 1916. From a spatial perspective, markers concentrate in the coastal cities of Galveston, Houston, and Corpus Christi, but no discernible temporal patterns emerge concerning the timing of commemoration—that is, when a marker is erected. After looking at the when and where of commemoration, in the second part of the analysis we look at how each disaster term is described in the text of the markers. Here, the narrative is one of initial destruction quickly followed by recovery and the reestablishment of communities. We also note how the selective nature of commemoration extends to disaster. [ABSTRACT FROM AUTHOR]

"Smart (Studies) Now" rethinks the significance of Christopher Smart for authors, critics, and students. While writers from Frances Burney to Allen Ginsberg have rated this elusive author highly, critical appreciation has lagged, often focusing on his madness, or treating him as a hack who got lucky once. This essay extends attention from the linguistic and formal power of Smart's religious verse to highlight his generic range and cross-corpus coherence. Three imperatives to advance Smart studies are formulated: clarity about Smart's influence on authors across various periods, attention to the value and relevance of his complete corpus, and availability of a robust sample of his poetry and prose for classroom use. [ABSTRACT FROM AUTHOR]

This article seeks to compare two of the most significant mystical corpora of Judaism and Islam, Zoharic literature and the oeuvre of Muḥyī al-Dīn Ibn al-ʿArabī (1165–1240) respectively. Following a few pioneering studies on relations between Jewish and Islamic mysticism in the medieval Iberian Peninsula from recent years, this article intends to contribute further to the understanding of such relations. It compares one motif or concept shared by both corpora, that of the spiritual garment, according to which the different realms of creation are divine "garments" that cover the Godhead or veil the primordial divine light. It suggests that the similarities between Zoharic literature and Ibn al-ʿArabī's writings can be explained by their shared roots, which can be traced to the tradition of Arabic Neoplatonism. Some possible Neoplatonic sources for the similarity between the two corpora are also discussed. [ABSTRACT FROM AUTHOR]

Most dialects of Breton have largely penultimate stress, and are also said to exhibit stress on certain clitics when they precede monosyllabic content words. However, data suggest that this prosodic process may not be maintained consistently by modern Breton speakers. This study explores the nature of clitic stress in Breton and investigates its potential loss by examining the indefinite article, low numerals and the adverb re 'too' in three pre‐existing corpora: two linguistic atlases and a more recent online repository of Breton recordings. The findings show that the loss of stressed clitics is greater in south‐eastern Brittany, and more advanced for the indefinite article than for other contexts. It is suggested that this is due to the fact that the process does not occur with the definite article, and that the clitics are being reanalysed such that they have a less cohering relationship with their hosts. Given the unstable nature of Breton as an endangered language with a high proportion of older speakers, it is unclear whether this ongoing language change will continue along the same path, or if revitalisation is likely either to halt or accelerate it. [ABSTRACT FROM AUTHOR]