Search

Aviation English (AE), a specialized register of English, prioritizes precision, brevity and clarity to maximize aviation safety. While there has been a growing focus within the linguistics community on AE training and assessment since the release of a set of standards and recommended practices, its linguistic properties remain comparatively underexplored. Drawing upon Biber's (1988) multi-dimensional (MD) analysis framework, the present study conducted a corpus-based comparative MD analysis to investigate the multi-dimensional linguistic profile of AE vs. casual conversational English (CE) and exploratory factor analysis (EFA) to extract and interpret the co-occurring linguistic features of routine AE and non-routine AE. The comparative MD analysis shows that AE exhibits more information condensation, less authorial stance and technicality, and fewer features of online information elaboration compared to CE. The EFA shows variations in the linguistic and textual composition of the two sub-registers of AE across two dimensions. Non-routine AE demonstrates a present-focused, viewpoint/intention-oriented approach, involving higher levels of integrative information flow compared to routine AE. Routine AE is characterized by a higher degree of information condensation and is marked by a planned, procedural, and intensive use of standard phraseology. Some pedagogical implications are then proposed for enhancing AE training to cultivate pilots' and air traffic controllers' language competence for precise, unambiguous communication tailored to both routine and non-routine operational contexts. • AE: Dense information, less authorial stance and technicality, limit online elaboration. • Linguistic variations in routine and non-routine AE composition. • Non-routine AE: Present-focused, intention-oriented, integrated information flow. • Routine AE: Information condensed, planned, procedural, SP intensive. • Pedagogical implications for enhancing AE training. [ABSTRACT FROM AUTHOR]

Thesis writing (used here as an umbrella term to cover both master's and doctoral postgraduate-level writing) is a high-stakes genre for postgraduate students. This important student genre has been well-researched from a corpus-based perspective. Corpora of theses and also research articles have been used for data-driven learning (DDL) of this key genre. The purpose of this article is to critically examine key DDL initiatives, some of which take a 'research into practice' orientation. Importantly, the discussion is framed around the notion of 'specificity' in the context of needs analysis, and whether the initiatives take a wide-angle, narrow-angle or move from a wide-angle to a narrow angle approach. Accounts which focus on DIY (do-it-yourself) mini-corpus compilation and use by students are also reviewed. The final section of the article presents a critique of current pedagogic applications, taking a closer look at the issue of 'specificity' within the wider context of needs analysis and mapping out areas for future consideration. It is suggested that an ethnographic perspective may be particularly useful for conceptualising specificity relating to students' present situation needs. The article also considers the impact of AI/ChatGPT on future corpus-based pedagogy of thesis writing. • Disciplinary corpora can usefully be used for teaching thesis writing. • Thesis writing courses using corpora can be viewed along a continuum of a wide-angle and a narrow-angle approach. • Do-it-yourself corpora are an ideal means of achieving 'specificity' in thesis writing courses. • Ethnographic methods can also shed light on the issue of 'specificity' in thesis writing pedagogy. [ABSTRACT FROM AUTHOR]

Using the Central Andean Chavín Phenomenon as a case study, this paper explores the middle ground between top-down (big data) and bottom-up (Bayesian modeling) approaches to archaeological radiocarbon data. Compiling radiocarbon dates associated with the Chavín Phenomenon addresses questions of chronology, demonstrating that the relevant phases at interacting sites are relatively brief and broadly contemporary. In addition, the definition of a discrete span of time associated with the Chavín Phenomenon makes it possible to explore the context for that period of heightened interaction. Juxtaposing that timespan with a compilation of Central Andean radiocarbon dates identifies contemporary non-participatory sites, enabling characterization of the sociopolitical milieu within which the Chavín Phenomenon emerged, flourished, and faded. The identification of that corpus of sites also highlights the importance of a shift in focus from identification of interaction to characterization of interaction. Bottom-up approaches to radiocarbon chronology will be a key element of that effort, while high-level summary of radiocarbon assemblages can identify where additional dating and Bayesian modeling can have the greatest interpretive impact. [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]

Corpus retrieval is an important research direction in the field of corpora, which can help learners or applications search for relevant content and improve learning and application efficiency. Traditional retrieval methods have problems such as high development costs, complex structural design, and low data coverage. This article conducts research based on Lucene technology, leveraging its advantages such as inverted index structure, incremental indexing, object-oriented design, text analysis interfaces, and query engine services, to provide a complete solution for developing high-performance large-scale corpus full-text retrieval systems. Firstly, this article studies the foundation of Lucene technology, including system architecture, functional packages, and data flow analysis. Next, study probability retrieval models and directly model the relevance of user needs. Then, this article conducts system design and implementation, including system function design, inverted index design, and retrieval function implementation. Finally, this article conducts system testing and challenges, including building a testing environment, functional testing, and reliability testing. [ABSTRACT FROM AUTHOR]

This paper presents an innovative, custom Hindi Speech Corpus dedicated to assessing the positive intensity in Hindi language. By meticulously gathering voice samples from individuals across diverse demographics, this corpus is well crafted for acoustic research in Sentiment Analysis. Several sentences spanning a spectrum of sentiments (extremely positive, positive, neutral, negative) were acquired from participants. The resulting corpus comprises of 225 audio files. To guarantee the consistency and to accommodate domain knowledge based nuances, the sentiments and its intensity (high, medium, low) were classified by Expert annotation. We used PowerBI for attaining insights into emotional landscape of the corpus. Employing exploratory analysis and Sentiment Analysis, the corpus is analyzed to unveil linguistic features indicative of positive intensity in Hindi speech. This study advances the understanding of positivity expression in Hindi and furnishes valuable resources for linguistic investigations in the scope of Hindi language. It is observed that positive sentiment has the highest count at 58 and the lowest count is for very negative at 27. The main contribution of current work lies in creating the custom dataset in Hindi which to our knowledge is first of its kind. Also, we did not come across any dataset focused on intensity of positivity. These contributions shall definitely be beneficial to future researchers in exploring positive affect related research. [ABSTRACT FROM AUTHOR]

An oxymoron is a linguistic phenomenon in which a pair of opposite or antonymous words are combined to convey a new meaning. Sometimes, it is used to express figurative, irony, or rhetoric within the text. This issue has received relatively less attention in the realms of linguistics and computational disciplines. Oxymorons play a significant role in various language-processing applications. This study represents a pioneering effort in the exploration of oxymorons in the Bengali language. A corpus-based study of oxymoron is a fundamental issue that has not been explored so far. A system has been proposed for the automated recognition of oxymorons from a given corpus. Frequency analysis, semantic similarity, and an antonym dictionary have been employed to discern oxymorons within the corpus. The system achieved promising results when tested on a Bengali corpus, and found 308 distinct oxymorons. A corpus-based descriptive statistics is measured in two different corpora. The most common oxymorons are ranked based on their frequency. Their notable presence underscores the importance of the Bengali language. This study aimed to explore fundamental questions concerning oxymorons, such as the automated detection of oxymorons within a corpus, descriptive statistics regarding oxymorons across languages, and the process of their construction and creation. Additionally, efforts were made to extract oxymorons from large language models using zero-shot prompts, but the results were not as promising compared to our proposed system. [ABSTRACT FROM AUTHOR]

This paper explores what I call Dissociative Syntactic Amalgam (DSA) constructions, observed in conversation parts of fictional varieties of English, such as film/TV scripts and novels. I provide 27 cases of DSAs, most of which come from Corpus of Contemporary American English and Movie Corpus. A paradigm example is I get off at ten past I'm never going out with you (as an answer to What time do you get off (work) tonight?), where I'm never going out with you is clausal but fills a nominal slot. In DSAs, the speaker initially lures the hearer to create a positive expectation (e.g., I can go out with her), but the speaker 'dissociates' herself from this expectation, that is, expresses her negative attitude (e.g., irritation) towards it. DSAs have received little (if any) analytic attention, and this paper aims to fill this gap in the literature. More specifically, I describe interpretive aspects of DSAs in comparison with the standard cases of syntactic amalgams (e.g., Ken sent you'll never guess how many letters to Naomi) and develop a cognitive–pragmatic account in relevance theory, explicating the interpretive path a hearer takes in comprehending a DSA utterance. • Identifying the Dissociative Syntactic Amalgam (DSA) construction. • Considering the ways in which a speaker conveys a variety of emotive meanings. • Presenting a theory-neutral description of interpretive aspects of DSAs. • Developing a cognitive–pragmatic account of DSAs in relevance theory. [ABSTRACT FROM AUTHOR]

Three Minute Thesis (3MT) presentations have received global popularity, which develop the discursive competence of postgraduate students to communicate their disciplinary research to a wide audience. Despite increased research into aspects of 3MT presentations, little is known about how student presenters react to the audience's knowledge base and recognize the possibility of rejecting their arguments. Therefore, this study focuses on the use of negation in 3MT presentations, suggesting how presenters introduce an alternative positive position into the dialogue, and hence acknowledging it, so as to reject it. We explored the frequencies, forms, and functions of negation at the discourse-semantic level with other appraisal resources in a corpus of 124 3MT presentations across hard and soft disciplines. Our results revealed that not , no, and negative contractions were the most frequent, and notably preferred by soft-disciplinary speakers. We also identified intriguing cross-disciplinary variations in the way negation was used alongside pronounce resources that expressed explicit authorial emphases to co-articulate the rhetorical effect in discourse. Pedagogical implications are raised as to the strengthening of students' rhetorical awareness of linguistic resources to achieve effective academic presentations and wide communication of disciplinary knowledge. • We studied negation at lexical and co-articulating levels using appraisal system. • Disciplinary rhetorical use of negation was explored in 3MT presentations. • Negation more often occurred in 3MT presentations than in written genres overall. • 3MT speakers from soft disciplines used notably more negation resources. • Cross-field variation was found in co-articulations between negation and pronounce. [ABSTRACT FROM AUTHOR]

The need to promote research and make it visible to various audiences has led to the emergence of various digital genres which seek to draw attention to research publications. Tweetorials, long Twitter threads to communicate complex concepts, are increasingly being used by medical researchers to report on and promote their own published articles and preprints, in the competive context of academic publishing. The main purpose of this article is to examine the strategies employed by researchers to recontextualize scientific discourse in these tweetorials. The analysis of a corpus of 50 biomedical publication-promoting tweetorials has revealed five categories of strategies: strategies to establish the authors' authority and credibility; strategies to make claims and arguments convincing; strategies to engage the reader, by creating intimacy and dialogic involvement or by attracting their attention to the tweetorials; strategies to facilitate quick processing of information; and strategies to deal with space constraints. The results suggest that the recontextualization strategies used in the composition of these tweetorials are determined by the promotional purpose of the genre, the audience, the affordances and constraints of the medium, and the genre contextual features. • Recontextualization strategies in publication-promoting tweetorials are analyzed. • These tweetorials recontextualize the discourse of RAs to promote them online. • Frequent use of strategies to engage the audience and attract their attention. • Strategies to facilitate quick processing and deal with space constraints. • The medium and the genre contextual features have an influence on the strategies. [ABSTRACT FROM AUTHOR]

This paper analyzes discourse acts in an ESP context from a local grammar perspective. We illustrate this analysis by constructing a local grammar for the discourse act of commenting on information in graphics, i.e., graphic data commentary in the discourse of Economics. Based on a corpus of research articles of Economics, we extracted instances of graphic data commentary that contain lexical items indicating different types of graphics (e.g., table, figure). Eight functional terminologies (e.g., 'Graphic', 'Data') were identified from the instances for the subsequent local grammar analysis, which in turn led to 17 frequently occurring semantic patterns, i.e., local grammar patterns. Each pattern has been observed to have a distinctive way to organize meanings and a corresponding communicative function, fitting the variation of the context. The study indicates that local grammar can provide a precise and systematic description of the lexical-semantic means that economists rely on to communicate graphic information and can offer pedagogical insights into the teaching of academic writing in an ESP context. • The study develops a local grammar of graphic data commentary in economic discourse. • The analyses identified 17 local grammar patterns for graphic data commentary. • The patterns systematically quantify the lexical-semantic realization of this act. [ABSTRACT FROM AUTHOR]

The assembly of corpus-based discipline-specific word lists for pedagogical purposes has recently been on the rise (e.g., Arndt, 2022 ; Fraser, 2007 ; O'Flynn, 2019 ; Yang, 2015). In order to cater to the needs of learners in veterinary medicine (VM) and for field-specific academic literacy, this paper analyses a reiteration of the Veterinary Medicine Corpus (Özer and Akbaş, 2023 ; hereafter the VMC), assembling a specified list of academic words used in published research articles (RAs) in VM. So far, VM has remained largely unexplored with the exception of Durrant's (2009) limited coverage of VM texts. The analysis was conducted using LancsBox 6.0, prompted to yield a list of PoS-tagged lemmas (P-lemmas) sorted by frequency (Coxhead, 2000 ; Yang, 2015) and dispersion (Arndt, 2022). The most frequent 1,000 lemmas for each content word category were collated into a candidate list. We then manually eliminated proper nouns and GSL (West, 1953) words. The final list, the Veterinary Medicine Academic Word List (VMAWL), contains 835 PoS-tagged lemmas (P-lemmas), a reiteration of which was produced by adding the types in Bauer and Nation's (1993) taxonomy at the second level to test coverage. The VMAWL was profiled and validated against the four parent categories independently on AntWordProfiler (Anthony, 2022a) with analysis showing the VMAWL comprises 13.75 % of the VMC and diverges greatly from generic word lists like the AWL, NGSL, new-GSL, and AVL. The list can be used to develop teaching materials for EAP or ESP academic writing courses. • We assemble an academic word list in veterinary medicine with pedagogical value. • The machine-generated veterinary corpus was reconfigured for reliable dispersion statistics. • The justified list consists of 835 P-lemmas granting a satisfying coverage of 11.92 % in VMC. • Coverage by our justified list in the veterinary corpus is much higher than the generic lists. • The lexis in our disciplinary list is thus significant for better literacy in veterinary medicine. [ABSTRACT FROM AUTHOR]

Hematoma of corpus callosum is a very rare phenomenon and is caused by severe trauma to head. Most common traumatic injury to corpus callosum is seen in diffuse axonal injury in form of small hemorrhagic foci and associated prolonged unconsciousness. Trivial trauma causing well defined corpus callosal hematoma in absence of coagulation defects or neurological deficits in conscious patient has not been reported in the literature. We present such a unique case and the review the corpus callosal hematoma due to trauma. [ABSTRACT FROM AUTHOR]

The upper ranks of corpus word frequency lists are primarily populated by functional items, as well as by other small words characterised by their polysemy that reward focussed attention and research. Exploring these items has resulted in rich and fruitful insights from a corpus pragmatic perspective. This paper focuses on the nature and value of there as a discourse-pragmatic marker in Irish English (IrE). We focus on intimate discourse (i.e., the language used between family and friends), and use corpora of IrE to trace and investigate the varietal nuances of there in this context. Our analysis acknowledges the expected existential and locative functions and illustrates that there has specific discourse–pragmatic properties in the intimate sphere in IrE. We argue that there has acquired additional pragmatic dimensions and is used in the language of family and friends as, for example, an involvement marker or mitigator. In the discussion of the development of there as a discourse-pragmatic marker, we consult historical data that matches our focal context, and discuss the processes by which there has come to fulfil a range of discourse-pragmatic functions in IrE. • Presents the first focussed study of discourse-pragmatic there in contemporary Irish English. • Takes a corpus pragmatic form-to-function approach, moving from raw frequency to detailed discussion of use in context. • Demonstrates a high level of functional variation for there in a contemporary Irish English corpus. • Patterns of DPM there include its use with various time-stamps, as a mitigator and in certain conversational contexts. • There provides a shortcut to a notional, imagined shared space creating an accelerated involvement in interaction. [ABSTRACT FROM AUTHOR]

Numerous attempts to establish a practicable means of non-surgical embryo transfer in pigs were unsuccessful until instruments were developed to permit deposition of embryos deep inside a uterine horn. Recently, however, successful non-surgical transfer of embryos to the uterine body in non-sedated recipients has been reported. These reports have prompted us to present data from a study that addresses the question of whether the commonly experienced low pregnancy rates following transfers to the corpus uteri results from the inability of endometrial receptors to recognize the presence of embryos or to the prevailing unfavorable local conditions. Blastocysts were transferred to the corpus uteri of recipients in which maintenance of pregnancy was assured by the transfer of additional embryos to the tip of a uterine horn. Embryos from pigmented and unpigmented donors were used to be able to differentiate between fetuses from embryos transferred to different sites. Fewer fetuses developed from embryos transferred to the corpus uteri than from embryos transferred to the uterine horn. This leads to the conclusion that prolonged retention of blastcysts in the uterine body under conditions inappropriate for their stage of development is responsible for an unsatisfactory pregnancy rate. • Porcine blastocysts were simultaneously transferred to uterine body and uterine horn. • Transuterine migration is not warranted in embryos transferred to the corpus uteri. • Embryo transfer to the corpus uteri was less successful than to a uterine horn. [ABSTRACT FROM AUTHOR]

This study examines the pragmatic functions of fuck among British English teenagers in casual conversation in two youth language corpora from the 1990s and 2010s. It applies a corpus-pragmatics approach to explore how the ongoing weakening of the taboo strength of fuck in the perception of young speakers is realised in usage data. The major functions observed involve a predominance of idiomatic, emphatic and emotionally expressive functions. Conversely, usage associated with potentially abusive functions, including literal reference to sexual intercourse, is infrequent. Our observations are interpreted in the context of delexicalization and related long-term diachronic processes, whereby contemporary usage of fuck among teenagers is characterised in terms of semi-delexicalized, pragmatically strengthened usage with weakened taboo status. The article also evaluates the interpretation of idiomatic usage from a functional perspective, and contributes to methodological considerations of the use of spoken corpora for pragmatic research. • The taboo strength of fuck has weakened in British English conversation among teens • Predominant usage involves idiomatic, emphatic and emotionally expressive functions • Literal and potentially abusive usage of fuck is infrequent • The taboo status of swear words can be interpreted through delexicalization • Microstructural corpus-pragmatic analysis offers perspectives on taboo status [ABSTRACT FROM AUTHOR]

Metadiscourse has been subject of linguistic inquiry for a few decades. Nevertheless, the role and behaviour of metadiscourse in various non-academic discourses is still on the research agenda. This study contributes to this niche by examining interactional metadiscourse in informational-persuasive discourse, an under-researched promotional sub-genre taking a move analysis approach. Aware of the co(n)textual nature of metadiscourse, we conducted an English-Spanish (EN-ES) contrastive analysis of interactional markers taken from an ad-hoc comparable corpus of online tea descriptions. Benefiting from the rhetorical tagging of the corpus, devices that seemed to build a direct relation between writer-reader were identified manually, per move, and then classified as 'commentary' markers (Dafouz-Milne, 2008). Findings reveal variation of interactional metadiscourse within the genre moves and across languages. First, the more overtly persuasive and/or instructional a move is, the higher the occurrence of commentary markers. Likewise, while 'direct address' and 'directives' stand out both in EN and in ES, EN relies on 'self-mentions' to a great extent, while ES does so with 'inclusive we'. Overall, the study attests to the role of interactional metadiscourse in building rapport to persuade potential customers, be it through ethos, as in EN, or logos, as in ES. • Interactional metadiscourse is frequent in non-academic discourse. • Analysing move-bound commentary markers in English-Spanish online tea descriptions. • 'Direct address' and 'directives' are typical in the genre across languages. • 'Self-mentions' in English promotional texts build rapport through closeness. • 'Inclusive we' in Spanish promotional texts build rapport through group membership. [ABSTRACT FROM AUTHOR]

A long tradition of research in theoretical, experimental and computational pragmatics has investigated over-specification and under-specification in referring expressions. Along broadly Gricean lines, these studies compare the amount of information expressed by a referring expression against the amount of information that is required. Often, however, these studies offer no formal definition of what "required" means, and how the comparison should be performed. In this paper, we use a simple set-theoretic perspective to define some communicatively important types of over-/under-specification. We argue that our perspective enables an enhanced understanding of reference phenomena that can pay important dividends for the analysis of reference in corpora and for the evaluation of computational models of referring. To illustrate and substantiate our claims, we analyse two corpora, containing Chinese and English referring expressions respectively, using the new perspective. The results show that interesting new monolingual and cross-linguistic insights can be obtained from our perspective. • A new prospective on reference is proposed; • Referring Expressions are sub-categorised; • An annotation scheme is proposed based on the new perspective; • Using the scheme, a enhanced analysis is conducted on two reference corpora (in English and Chinese, respectively). [ABSTRACT FROM AUTHOR]

The study is devoted to the problem of label quality assurance and labeling cost minimization. It is important to have high quality labels in order to build efficient supervised machine learning pipelines. The aforementioned labels have proved to be costly to acquire; however, one can work with weaker instances of supervision that may be helpful to create a reliably set of ground truth labels for classifiers. The authors present a semi-supervised approach using pretrained language models to increase supervision information quality from general-purpose as well domain-specific document sets where weak labels and unlabeled data is present. [ABSTRACT FROM AUTHOR]

This paper analyzes formulaic language in conference abstracts in Applied Linguistics and tracks an application of the findings to a second language English academic writing context. Specifically, the authors identify and profile phrase-frames (a form of discontinuous formulaic sequence) in terms of their frequency, internal variation, predictability, structure, and rhetorical functions in a corpus of approximately 1,600 conference abstracts in Applied Linguistics. The analysis informed a corpus- and genre-based second language writing pedagogical intervention on formulaic language, and the paper also presents the experiences of learners through sample student writing, survey, and interview data. Overall, findings highlight the pedagogical potential for including frame-based formulaic language in second language writing pedagogy due to the prevalence and inherent productivity of such features, the clear role they play in signaling functional intentions in rhetoric, and their ability to stimulate reflection and discussion of formulaicity in writing more broadly. • Nearly two-fifths of phrase-frames had strong relationships with lead variants. • Phrase-Frames were notably productive, with about one variant per two occurrences. • Phrase-Frames served as strong cues of rhetorical moves. • Students reported and demonstrated confidence with and awareness of formulaicity. • Challenges of corpus-based teaching surfaced and limited changes to writing found. [ABSTRACT FROM AUTHOR]

How limitations are acknowledged and discussed has a profound impact on the extent the research is evaluated and accepted by its intended readers. However, little attention has been drawn to the presentation of limitations in the EAP literature. This study seeks to remedy the oversight by exploring how this discursive practice is mediated by metadiscourse, how limitations are rhetorically contextualized and how much these rhetorical investments differ between PhD dissertations and research articles in applied linguistics. A corpus-based analysis of 100 PhD dissertations and 200 published articles in applied linguistics shows that PhD dissertation writers make more use of frame markers but less use of code glosses, evidentials, and hedges in the acknowledgment of limitations than published writers do in limitations steps. It is also found that limitations pertaining to the overall quality of research and writers' competence are far more often self-reported in PhD dissertations than in research articles, and PhD dissertation writers tend to attribute the limitations to situational constraints in research context and unmanageable complexity of research subjects. The results support the two-genre perspective (El-Dakhs, 2018; Kawase, 2015) and demonstrate that discussing limitations is a strategically self-critical but promotional effort in conclusion sections. • Discussing limitations are rhetorically contextualised in EAP discourse. • Rhetorical practice of acknowledging limitations differ between PhD dissertations and research articles. • The differences in metadiscourse markers used in Conclusion sections are investigated. • Different self-evaluative foci of limitations are identified and the mitigation of their impacts is discussed. [ABSTRACT FROM AUTHOR]

Innovation can be a key mechanism to address some of society's greatest challenges, or it can contribute to them. There is extensive conceptual academic literature focused on how policy can be used to create more positive societal and environmental impacts through innovation, however, little empirical evidence exists to understand to what extent innovation policy in particular embeds the principles of social and environmental sustainability into its discourse. We begin to address this lack of evidence by using a critical discourse analysis and corpus linguistics approach to explore how UK Innovation Policy embeds the concepts of societal and environmental impact, and how it balances these at times conflicting paradigms into policy documents. We find that although there is some inclusion of key environmental and societal words these are predominately secondary to economic themes, signalling a 'business as usual' approach to innovation policy. • We provide a comprehensive definition of sustainable innovation policy building on types of alternative innovation. • Our methodology contributes a novel way to assess the underlying sustainability paradigms in policy documents. • Social and environmental paradigms are underrepresented in innovation policy and are dominated by the economic paradigm. • We make recommendations for ways in which policymakers can integrate sustainable innovation policy principles further. [ABSTRACT FROM AUTHOR]

Discourse parsing for multiparty dialogue aims to detect the discourse structure and relations in a dialogue to obtain a discourse dependency graph. Existing models for this task have proven the effect of speaker information. However, no model explicitly learns representations of the speaker for parsing the discourse structure of dialogues. In this paper, to further exploit the effect of speaker information, we propose a novel model HG-MDP , which uses a heterogeneous graph neural network to encode dialogue graphs and we use an iterative update for the aggregation of speaker nodes and utterance nodes. Finally, we adopt updated utterance nodes to predict discourse dependency links and relations using the biaffine module. To validate our HG-MDP model, we perform experiments on the two existing benchmarks STAC and Molweni corpus. The results prove the effectiveness of the speaker modeling module on two datasets and we achieve the state-of-the-art on the Molweni dataset. [ABSTRACT FROM AUTHOR]

Research in the area of ​​the interconnection of lexical resources represents a real challenge, because it addresses the difficult problem of semantic understanding and, more precisely, the disambiguation of the meaning of the words - Word Sense Disambiguation (WSD). In the current and prospective context of the information society, the existence of the digital format of the fundamental works of a national culture is strictly necessary. It is a topical issue throughout the world of creating a representative corpus of a language accessible through the Internet, the corpus being a concrete, clear picture of the use of that language. In this study we will describe the development of a Romanian language GOLD corpus, related to the multiple meanings existing for various words. We propose a corpus annotation standard, based on three lexical resources as follows: the Thesaurus Dictionary of the Romanian Language in electronic format (eDTLR), from which we extracted a list of words with multiple meanings; from the Reference Corpus for Contemporary Romanian Language (CoRoLa) we extracted contexts in which these words were founded and from the the Romanian WordNet (RoWN) resource, we took into account the sense meaning of the word from the corpus context. [ABSTRACT FROM AUTHOR]

We present a novel corpus for French dialect identification comprising 413,522 French text samples collected from public news websites in Belgium, Canada, France and Switzerland. To ensure an accurate estimation of the dialect identification performance of models, we designed the corpus to eliminate potential biases related to topic, writing style, and publication source. More precisely, the training, validation and test splits are collected from different news websites, while searching for different keywords (topics). This leads to a French cross-domain (FreCDo) dialect identification task. We conduct experiments with four competitive baselines, a fine-tuned CamemBERT model, an XGBoost based on fine-tuned CamemBERT features, a Support Vector Machines (SVM) classifier based on fine-tuned CamemBERT features, and an SVM based on word n-grams. Aside from presenting quantitative results, we also make an analysis of the most discriminative features learned by CamemBERT. Our corpus is available in open-source format at https://github.com/MihaelaGaman/FreCDo. [ABSTRACT FROM AUTHOR]

This study investigates all- cleft constructions in the London–Lund Corpora of spoken British English. It has two aims: an empirical and a methodological one. The empirical aim is to contribute to research on spoken discourse by analysing the form–meaning properties of all- clefts along with their pragmatic functions and their development since the 1950s drawing on insights from Cognitive Linguistics with special focus on Diachronic Construction Grammar. We show that all -clefts are used both as assertions and directives. They focalise an element that speakers find particularly relevant at the same time as they have the effect of blocking alternative perspectives and thereby contracting the discursive space for addressees. The methodological aim is to critically assess the comparability level of the diachronic corpora based on a detailed investigation of the distribution of all -clefts in the data. We show that there is a high degree of similarity between the designs of the corpora, which makes them suitable for diachronic investigations of recent change. • The pragmatic functions of all -clefts were assertions and directives. • There is evidence for increased intersubjectivity of all -clefts in llc –2. • All-clefts have a focalising and dialogically contractive function. • There is a high level of similarity between the designs of the London–Lund Corpora. [ABSTRACT FROM AUTHOR]

Previous research on pragmatic overlap and change suggests that in cases in which two linguistic items evolve one after the other to express similar functions, the new form may substitute the older one, the older expression may further evolve, thereby differentiating from the newer item, or the two forms may become more alike. In this paper, we present a case attesting to a different process. In this type of path, it is the newer form that further evolves, thus "outpassing" the older expression in the subjective-intersubjective cline. The case in point is the consecutive development of the Hebrew phrases yeʃ/en sikuy/matsav ('there is/isn't a chance/situation'). In Present-Day Hebrew, these pairs express similar subjective-modal functions referring to possibilities, yet only the relatively new pair, yeʃ/en matsav , also functions as a conventionalized intersubjective marker. We here provide a descriptive account of their development in the cline and the factors involved in the further evolution of yeʃ/en matsav. Our findings suggest that yeʃ/en matsav pragmatically outpassed their counterparts due to a semantic advantage relating to the persistence of a circumstantial meaning. Overall, this research illuminates an understudied type of development, pointing to the multifaceted nature of pragmatic change. • An examination of consecutive change and overlap of Hebrew intersubjective markers. • Corpus data reveal the new expression has further evolved pragmatically. • Findings suggest new items may outpass older ones due to semantic advantages. [ABSTRACT FROM AUTHOR]

Previous lexical bundle research has stimulated heated discussions on disciplinary variations and disciplinary specificity-generality spectrum. The current study explores whether (and how) bundle extraction criteria (i.e. frequency, dispersion, and bundle length) may affect the conclusions on disciplinary variations and specificity-generality. Focusing on eight disciplines, it used an 11-million corpus of academic journal articles with a balanced design (the same number of texts and similar average text length for each subcorpus). The results indicated a clear picture of disciplinary variations and a strong tendency towards disciplinary specificity. More importantly, the results suggested that different methodological criteria played an important part, especially concerning the analysis of disciplinary specificity or generality. The choice of '4-word' bundles over '3-word' bundles, in particular, would tend to be associated with a result of a higher degree of disciplinary specificity. Regarding disciplinary variations, it was found that the effect from different choices of bundle extraction criteria was relatively smaller. Only the choice on different dispersion requirements could predict the significant differences between certain disciplines. These findings provide support for disciplinary variations and new perspectives on the disciplinary specificity-generality debate. • Explores the effect of identification thresholds on bundle analysis results. • Uses a balanced corpus of academic journal articles from eight disciplines. • Indicates a strong disciplinary variation and disciplinary specificity. • Shows dispersion as a negative predicator for significant disciplinary variations. • Shows bundle length as a crucial variable on the conclusion of specificity-generality. [ABSTRACT FROM AUTHOR]

• Two experiments demonstrated how RC and material features affect the performance of error annotation tasks. • Annotators demonstrated higher efficiency and accuracy when the task involved lower RC. the difficulty of the error annotation task was reduced by a reduction in the RC of the annotation task through the decomposition method. • In addition to RC, the implied features of material were also effective predictors of error annotation performance. the implied features were: error severity, sentence length and number of errors. With low inter-annotator agreement and large time consumption, phonetic annotation of non-native speech is complicated. However, having a large-scale non-native speech corpus with error annotation is important for second language phonology and computer-assisted pronunciation training research. To achieve large-scale annotation, we must clarify the causes of difficulty in error annotation. This study identified the factors that can predict annotation performance of non-native Mandarin speech by considering two perspectives.1) We used a decomposed method (that involved dividing a traditional error annotation task into two different subtasks, namely the error location task and error description task) to determine whether human performance, from the cognitive viewpoint, can be predicted using the decomposed annotation method (Experiment 1). 2) We determined whether the features of annotation task materials (e.g., error severity) can be employed to predict annotation performance (Experiment 2). The results of Experiment 1 revealed that the decomposed method leads to a shorter annotation time and a higher consistent hit rate than does the traditional method. The regressions of Experiment 2 revealed that the features of error severity and sentence length accounted for 96% of the hit rate and the features of sentence length and number of errors accounted for 95.5% of the annotation time. These results suggest that the performance of manual annotation can be predicted through both cognitive factors and material features. [ABSTRACT FROM AUTHOR]

This paper examines metalinguistic negation of proper names on the data from the Russian National Corpus, including its pragmatic functions, its performance with regards to other cases of metalinguistic negation, and the opposition between explicit and implicit use of the metalinguistic predicate. The data showed that the speaker's most common aim is to reject a particular representation of the name owner in favor of another representation. Next, we discovered that metalinguistic negation is more natural in corrections of content ascribed to speaker than in reactions to a preceding interlocutor's comment. Finally, the metalinguistic predicate of naming can be expressed not only lexically, but also be part of the meaning of some syntactic constructions. • Metalinguistic negation of proper names is analysed on the Russian data. • We found that it indicates the unsuitability of the name in the current situation. • It is more commonly used for correcting contents ascribed to speaker that contents ascribed to hearer. • Metalinguistic predicate of naming can be grammaticalized. [ABSTRACT FROM AUTHOR]

This work proposes a multimodal Hindi-to-English machine translation on a news corpus by integrating multiple input modalities. The experimental dataset comprises of an image from a news article and its caption in the English-Hindi language. The parallel English-Hindi dataset is constructed by translating the image caption from English-language news articles into Hindi. Unlike standard images used for NLP tasks, images in news articles frequently depict significant events, locations, or individuals. The caption provides a more extensive description of the image, resulting in the inclusion of multiple identified entities. Our methodology consists of encoders for each input modality, allowing for simultaneous consideration of both text and image. We undertake two tests to examine how effectively the image content improves MT systems when 1) the entire image is considered first and 2) the largest object recognized in the image is considered. We employ Byte Pair Encoding (BPE) to represent the text, and the visual attributes are extracted from the image with VGG-19, a pre-trained CNN model. In terms of BLEU and chrF, our Multimodal Machine Translation (MMT) systems are superior than the baseline unimodal NMT system. Our MMT system outperforms the NMT model by +1.8 BLEU and +0.03 chrF. [ABSTRACT FROM AUTHOR]

Distant supervision (DS) can automatically generate annotated data for relation extraction (RE) with knowledge bases and corpora. The existing DS methods that train on bags selected by attention mechanism are susceptible to noisy bags and neglect useful information in noisy bags. In this paper, we propose DCSR, a novel DS method which utilizes deep clustering to obtain refined superbag representations for solving the wrong labeling problem. we substitute deep clustering for selective attention to construct superbags, capturing helpful information between spatially-close bags, including noisy bags. Moreover, we implement data augmentation on the input sentences to handle the long-tail problem. Experiments on the NYT2010 and NYT-H datasets show that our method can effectively improve RE and significantly outperforms state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Grammatical metaphor (GM) has been shown to perform a central function in academic prose, enabling writers to reorganise grammatically intricate texts into lexically dense expression. It follows, therefore, that some facility with grammatical metaphor is an important facet of academic literacy. However, due to its contextual nature, the study of GM has almost entirely been conducted at small scale, precluding a broad understanding of how it is employed in published writing and the extent to which its use varies across disciplines. The current study begins to address these issues with an analysis of GM use in a 200,000-word corpus drawn from samples of 200 research articles across four disciplinary groupings. The 3,480 GMs identified in the analysis were ranked by frequency and dispersion and compared across the four disciplinary groupings. Some variation was detected between disciplines, both in frequency and in the most commonly used forms of GM, but these differences were small, suggesting that a generalised approach to the instruction of GM may be appropriate. The findings from this study can inform both future research into GM use and more accessible and strategic instruction of GM in ESP and EAP classes. • A research-article corpus was compiled to analyse grammatical metaphor. • A list of 3,480 grammatical metaphors, including dispersion data, is offered. • Little variation in grammatical metaphor frequency was found between disciplines. • Variation was found in grammatical metaphor types between disciplines. [ABSTRACT FROM AUTHOR]

In this study, we investigate self-denigration in Chinese by applying a strictly language-based interactional speech act approach. Previous research has often zeroed in on Chinese self-denigration by studying the pragmatic function of expressions selected beforehand by the researcher. As an alternative, we propose a bottom-up approach through which self-denigration can be captured in phases of interaction and expressions speech acts and expressions embedded in these phases. Our corpus consists of dialogues drawn from Chinese novels written between the 14th and 20th centuries. As our analysis shows, self-denigration often serves smoothening the flow of interaction and mimimising offence, along with expressing deference. Furthermore, self-denigration is more frequented in goal-oriented interactions than in ceremonial ones. Our results also contribute to previous discussions relating to Ide's (1989) renowned notion of 'discernment'. • Investigates self-denigration in Chinese by applying a strictly language-based interactional speech act approach. • Contributes to previous discussions relating to the renown notion of 'discernment'. • Studies a corpus of dialogues drawn from historical and historicising Chinese novels written between the 14th and 20th centuries. [ABSTRACT FROM AUTHOR]

Few-shot extractive question answering (EQA) is a challenging task in natural language processing, whose current methods are mainly based on pretrained language models (PLMs). Data augmentation is often employed to improve the answer predictions of EQA models in few-shot settings. However, due to the differences between pretraining objectives and the EQA task, as well as embedding space alignment bottlenecks, the performance of few-shot EQA models must be improved. We propose TPKE-QA , a few-shot extractive Q uestion A nswering approach via T ask-aware P ost-training and K nowledge E nhancement, with entity-noun-oriented span selection in post-training, which can automatically generate EQA-style examples from a large-scale unlabeled corpus. By post-training based on generated examples, the gap between PLMs and the EQA task is effectively filled. To avoid embedding space alignment issues, a knowledge-enhanced sequence generation and knowledge injection approach for the EQA task enables gapless knowledge enhancement and fine-tuning on the post-trained model. In experiments, TPKE-QA achieved state-of-the-art results in most few-shot settings on the MRQA 2019 benchmark. [ABSTRACT FROM AUTHOR]

In this era, document mining techniques play a crucial role in extracting meaningful information from the text corpus. The high dimensionality of the text corpus is one of the biggest issues in the document mining process. It degrades the efficiency of the document mining model. This work proposes a two-step feature/term selection scheme using a hybrid elitist self-adaptive cat and mouse based optimization method for document classification. The fitness function (objective function) of the proposed strategy is designed to minimize both the classification error rate and the term/feature selection rate. We utilized a collection of four text corpora, namely Reuters-R8, 10 Newsgroups, AG News Corpus, and WOS-5736, to examine the efficacy of the suggested approach. The suggested scheme outperforms other competing schemes, attaining an accuracy of 96.45%, and a F1 score of 0.9645. The comparison analysis report indicates that the suggested method performs better than the other competitor approaches. The Nemenyi post-test over the significance level of 0.05 reveals that there is a significant difference between the efficacy of the suggested scheme and other rival schemes. [ABSTRACT FROM AUTHOR]

As a new paradigm for information retrieval, generative retrieval (GR) has achieved solid performance on various retrieval tasks. Despite its promising progress, this line of research cannot generalize on a dynamic corpora, where new documents are continually added to it. There are already some continual learning-based pioneering works focusing on this issue, yet the continual learning framework requires retraining after model deployment and may suffer from catastrophic forgetting issues. Hence, we propose a new retrieval framework noted as ECI (an Efficient Corpus Indexer for dynamic corpora retrieval). The ECI is a hybrid index framework containing generative and deep hashing indexes. We design a complementary training objective noted as Prefix-Sensitive Similarity Alignment, which can further improve the performance of generative retrieval. Besides, ECI enables incremental deep hashing learning and provides a deep hashing index-based retrieval scheme for new documents, thus solving the generalization problem on dynamic corpora. Furthermore, ECI utilizes techniques like whitening and query-generated data augmentation to enhance retrieval performance. In a dynamic corpus retrieval task built on the commonly used academic benchmark Natural Question, the ECI outperforms various baselines, including the state-of-the-art GR baseline and its variants. [ABSTRACT FROM AUTHOR]

Most companies struggle to find and extract relevant information from their technical documents. In particular, the Oil and Gas (O&G) industry faces the challenge of dealing with large amounts of data hidden within old and new geoscientific reports collected over decades of operation. Making this information available in a structured format can unlock valuable information among these mountains of data, which is crucial to support a wide range of industrial and academic applications. However, most natural language processing resources were built from general domain corpora extracted from the Internet and primarily written in English. This paper presents Petro NLP, a comprehensive set of natural language processing and information extraction resources for the oil and gas industry in Portuguese. We connected an interdisciplinary team of geoscientists, linguists, computer scientists, petroleum engineers, librarians, and ontologists to build a knowledge graph and several annotated corpora. The Petro NLP resources comprise: (i) Petro KGraph– a knowledge graph populated with entities and relations commonly found on technical reports; and (ii) Petrolês, PetroGold, PetroNER, and PetroRE– sets of corpora containing raw text and documents annotated with morphosyntactic labels, named entities, and relations. These resources are fundamental infrastructure for future research in natural language processing and information extraction in the oil industry. Our ongoing research uses these datasets to train and enhance pre-trained machine learning models that automatically extract information from geoscientific technical documents. [Display omitted] • Petro NLP is a set of natural language processing and information extraction resources. • It is specialized in the oil and gas industry in Portuguese. • We formally defined a geoscientific ontology and populated a knowledge graph. • We present corpora with morphosyntactic, named entities, and relations annotations. • These are key resources to train machine learning models for information extraction. [ABSTRACT FROM AUTHOR]

In the field of human olfactory perception (OP) and odorant chemistry (OC), a substantial corpus of data has been amassed, with efforts directed towards constructing empirically valid and applicable models. Machine learning (ML) is highly adept at processing vast quantities of data and generating sophisticated models. Recently, there has been a surge of interest in the potential of ML to bridge the gap between OC and OP. This review presents a brief discussion on the application of ML to OC scenarios, accompanied by an overview of the most commonly used models. It highlights the selected input information, outlining the application of ML in OC in terms of data variability. While predictive models based on standardized data are valuable in odor quantification, their applicability is limited in mixed compounds. The incorporation of diagrams and spatial attributes expands the odor space, and models generated from instrumentally acquired data are primarily utilized for the differentiation of odor types. Odor descriptions, which are often directly used to influence model generation, subjective and voluminous, necessitating a process of interpretation. Meanwhile, objective data do not provide a comprehensive description of odor. This paper illustrates the future of odor prediction, offering insights into the evolving landscape of this field. It is anticipated that ML will be employed in the future to facilitate a more profound comprehension of odorants and the human olfactory sensory mechanisms, thereby offering a valuable contribution to the field. • Recent odor prediction combined with ML has been summarized. • The use of ML in odor prediction is described from a data perspective. • Highlights of linguistic descriptions as output information are presented. • The evolution of odor prediction and its applications are discussed. [ABSTRACT FROM AUTHOR]

Recently, exciting progress has been made in the research of supervised image captioning. However, manually annotated image-annotation pair data is difficult and expensive to obtain. Therefore, unpaired image captioning becomes an emerging challenge. This paper proposes a method called Mining Core Information by Evaluating Semantic Importance (MCIESI) for Unpaired Image Captioning, which is a method for image captioning using unpaired images and sentences. The main difference from the existing methods is that MCIESI focuses on mining the information that should be described in the image and embodies them in the generated natural language that conforms to human thinking. To achieve this goal, we use scene graphs to represent the semantics of images and evaluates the importance of objects and interaction relationships to mine core information in images, which are then encouraged to be embodied in generated sentences through semantic constraint. Combined with grammatical constraint using adversarial training with real sentence corpus and relative constraint using a triplet loss, the generator is trained to generate semantically plausible and grammatically correct sentences. Extensive experiments verify the effectiveness of MCIESI. [Display omitted] • We explore image captioning without paired image-annotation data. • We evaluate the importance of semantics with a reasonable and stable strategy. • We rationally use the combination of the three constraints to optimize the model. • Experiments on real datasets demonstrate the effectiveness of our method. [ABSTRACT FROM AUTHOR]

In the education sector, there is a growing need for efficient Automatic Short Answer Grading (ASAG) systems that require less training time, have faster inference, and provide higher accuracy. Over the past five years, researchers have made progress using deep learning for ASAG, but acquiring domain-specific labeled data remains a significant challenge. This study introduces an ASAG framework designed to learn from unlabeled domain-specific contexts while minimizing training and inference times. Our framework involves three main phases: 1) Building an unlabeled domain-specific corpus. 2) Training a pre-trained language model with this corpus. 3) Fine-tuning the pre-trained model with limited labeled domain-specific data. We explored various domain adaptations using two types of vocabulary during the pretraining stage. Our approach achieved an accuracy of 88.77% and an F1-score of 91.59% on the SPRAG dataset, improving state-of-the-art methods by 4%–5.7%. The models, including one trained from scratch with domain-specific data and vocabulary, showed robust performance, enhancing accuracy by 5.7% and F1-score by 4.84%. However, the framework's effectiveness in handling different languages or multilingual datasets remains to be explored. This framework's adaptability and efficiency enable educators to use AI for automated grading in various educational domains. [ABSTRACT FROM AUTHOR]

In this study, we examine how family members expressed sympathy when visiting patients in mid-eighteenth-century China, by studying a corpus of patient visits drawn from the renowned Chinese novel Hongloumeng 红楼梦 (Dream of the Red Chamber). The study of such visits is methodologically challenging because these visits were less scripted than many other ritual interactions, operating with many seemingly ad hoc elements. We approach patient visits in our data as interaction rituals which – similar to any ritual – operate with a frame and a related cluster of conventionalized patterns of language use. We argue that the pragmatic dynamics of many interactional ritual phenomena such as patient visits can be reliably captured by bringing together ritual, speech acts and interaction. • Captures the pragmatic dynamics of historical Chinese patient visits. • Presents an integrative model for the study of interactional ritual phenomena with seemingly ad hoc elements. • Brings together ritual, speech acts and discourse. [ABSTRACT FROM AUTHOR]

Inspired by propositions founded on the Communication as Constitutive of Organization literature, this paper employed a mixed method combining Cultural Discourse Analysis (CuDA) with Corpus Linguistics to investigate how companies manage aspirational talk-action tensions in corporate social responsibility (CSR) discourses for social license to operate. Ten sustainability reports of five companies in Singapore and Hong Kong were accessed to examine the emergence of CuDA discursive hubs (identity, relation, emotion, action, and dwelling), the articulation of linguistic elements via concordance analysis, and the adoption of socio-cultural dimensions in addressing tension mechanisms. Findings revealed intricate communication dynamics resulting in four strategies that companies undertook to address and manage discrepancies, thereby enabling the construction of effective narrative CSR communication. These strategies provide valuable insights for public relations practitioners who engage in sustainability reporting and seek to adequately handle aspirational talk-action tensions that can possibly cause negative impacts. • Provided empirical discourse to illustrate the communication dynamics of CSR talk-action discrepancies. • Identified, exemplified and interpreted interactions of cultural and linguistic practices in specific CSR contexts. • Offered a holistic interpretation of analyzing CSR discourse in communicative interactions. [ABSTRACT FROM AUTHOR]

Text reuse is the process of creating new texts from pre-existing ones. In recent years, Urdu Text Reuse Detection (U-TRD) has garnered the attention of researchers due to the ready availability of digital text all over the internet, which can be copied or paraphrased from other sources without proper attribution, making it easier to reuse but challenging to detect. Previous studies have explored the issue of U-TRD at the phrasal, sentence/passage, and document levels, using benchmark corpora and methods. However, the problem of U-TRD has not been investigated at the lexical level in terms of corpora and methods. To address this research gap, our study has developed a large benchmark corpus manually annotated at the lexical level. This corpus consists of 22,184 text pairs categorized into two levels of rewrite: (1) Derived (8,660) and (2) Non-Derived (13,524). Additionally, our research has involved the development, application, evaluation, and comparison of a range of methods, including baseline methods (uni-gram overlap and word embedding-based methods), along with state-of-the-art transformer-based methods and feature-fusion-based methods, using the proposed UTRD-Lex-23 corpus. Our study concludes that one of our proposed feature-fusion methods outperforms all other methods. The model we propose, which combines seven different Sentence Transformers (ST) (each producing 768 dimension vectors) with one uni-gram (at word level) and sixteen different features extracted from four different Word Embedding (WE) based models (yielding 300 dimension vectors), achieves an F 1 score of 0.70601 using 10-fold cross validation. To foster and promote research in Urdu (a low-resourced language) proposed corpus will be freely and publicly available for research purposes. [ABSTRACT FROM AUTHOR]

Video corpus moment retrieval aims to locate video segments that semantically correspond to natural language queries in a large collection. Effective representation learning across the video, subtitle, and language modalities is crucial for success. However, existing methods prematurely combine visual and subtitle features, leading to redundant information that hinders language query matching. Moreover, these methods lack query guidance during visual encoding, overlooking fine-grained semantics. To address these limitations, we propose a simple yet effective approach called Semantic-guided Late Fusion Transformer (SgLFT) for video corpus moment retrieval. Our method leverages a Semantic-guided Locality-aware Transformer to capture fine-grained visual embeddings using distinguishable query semantics, independent of the presence of subtitles. Furthermore, we introduce a late fusion approach that merges subtitle and visual features under a Cross-modality Global-aware Context Fusion unit, enhancing global contextual details. Finally, a Query-aware Feature Learning module aligns the query and video into a unified representation for localization. Our framework effectively models fine-grained semantic interactions between modalities with query guidance, advancing cross-modal representation learning. Extensive experiments on the TVR and DiDeMo benchmarks demonstrate that SgLFT significantly outperforms previous state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Keyword extraction involves the application of Natural Language Processing (NLP) algorithms or models developed in the realm of text mining. Keyword extraction is a common technique used to explore linguistic patterns in the corpus linguistic field, and Dunning's Log-Likelihood Test (LLT) has long been integrated into corpus software as a statistic-based NLP model. While prior research has confirmed the widespread applicability of keyword extraction in corpus-based research, LLT has certain limitations that may impact the accuracy of keyword extraction in such research. This paper summarized the limitations of LLT, which include benchmark corpus interference, elimination of grammatical and generic words, consideration of sub-corpus relevance, flexibility in feature selection, and adaptability to different research goals. To address these limitations, this paper proposed an extended Term Frequency-Inverse Document Frequency (TF-IDF) method. To verify the applicability of the proposed method, 20 highly cited research articles on climate change from the Web of Science (WOS) database were used as the target corpus, and a comparison was conducted with the traditional method. The experimental results indicated that the proposed method could effectively overcome the limitations of the traditional method and demonstrated the feasibility and practicality of incorporating the TF-IDF algorithm into relevant corpus-based research. [ABSTRACT FROM AUTHOR]

• The exact phonetic implementation of laryngeal contrast can modulate sound change. • This is true even within 'aspirating' varieties. • Voice onset time and voicing probability systematically covary in Danish dialects. • VOT and voicing in turn covary with historical lenition patterns. This paper provides evidence for the assumption that the precise phonetic implementation of laryngeal contrast in obstruents can have an influence on higher order linguistic structure. Traditional varieties of Jutland Danish – which are all broadly 'aspirating' varieties – are used as a case study. The paper shows that the precise implementation of the aspirated–unaspirated contrast in stops varied systematically in these varieties, and that this covaries with the morphophonological process of stop gradation. Stop gradation is a lenition process which is historically found in the entire Danish-speaking area, but with quite varying outcomes, which were mapped extensively by dialectologists more than a century ago. Using a large legacy corpus of sociolinguistic interviews from the 1970s, this study shows that more sonorous outcomes of stop gradation covary with higher rates of continuous closure voicing in /b d g/ and shorter aspiration in /p t k/, and vice versa for less sonorous outcomes of stop gradation. [ABSTRACT FROM AUTHOR]

Although the grant proposal is a high-stakes genre for researchers, there are few analyses available for consultation by learners and most studies investigate only a limited number of successful proposals written by experienced academics. This study reports on a genre analysis of a learner corpus of grant proposal summaries (abstracts), written by researchers who operate at the periphery of academia. The proposals were written by exiled Syrian academics and submitted to the Council for At-Risk Academics (Cara) grant awarding body for research funding. A corpus of 102 proposal summaries was compiled consisting of 27 successful and 75 unsuccessful summaries, and a genre framework of three moves and ten steps was developed. Successful summaries were contrasted with unsuccessful summaries; this comparison reveals that unsuccessful summaries underuse the move Indicating the value of the research. Specifically, they tend to omit two steps: Importance (of the research) and Research Outcomes. All Cara summaries were also compared with Matzler's (2021) prototype; results show that both successful and unsuccessful summaries underuse the Methods step. These findings provide pointers to the genre functions likely to be most problematic for learners, and have immediate practical applications in pedagogic materials for proposal writing. [ABSTRACT FROM AUTHOR]

Precise logistic support is essential after a disaster occurs. It must be timely, accurate, targeted, and based on existing needs. However, obtaining sufficient and accurate information related to logistic distribution locations remains a key problem. Therefore, implementing Named Entity Recognition (NER) can address this issue. In recent years, news coverage through Indonesian digital news media and social media accounts has emerged as a promising source for building a disaster data corpus. This study implemented NER to extract and identify named entities from text-based information, particularly from Indonesian digital news media. In addition to using regular entities from the NER standard, this study introduced new entities specialized for disaster-related information, including DISASTER, SCALE, SUPPLIES, CASUALTIES, and OUTSIDE. The new disaster corpus in the Indonesian language for the NER model was obtained with an imbalanced dataset composition. To overcome this problem, random oversampling was applied. This study also utilized the BiLSTM model to recognize each entity in new textual information, evaluating its performance when the proposed Indonesian disaster corpus was used as a training reference in the deep learning model. Several optimization algorithms applied in BiLSTM were evaluated. The results showed improved BiLSTM performance using Adam optimization and a balanced corpus. Performance indicators achieved were 93.4 %, 82.4 %, and 87.5 % for precision, recall, and F1-score, respectively. The BiLSTM network captured long-range dependencies in sequential data provided by NER. Oversampling ensured that the proposed NER model could precisely recognize all entities and reduce biased results. Thus, the BiLSTM method can better identify entities in the textual corpus of Indonesian disaster-related online news. [ABSTRACT FROM AUTHOR]

This study examines diachronic changes in number use in written American English from 1923 to 2008 in TIME magazine, analyzing a 170-million-token corpus. We investigate the effects of magnitudes, roundness, and representational formats on number frequency, along with the evolution of culturally salient numbers reflecting societal shifts. Utilizing Bayesian negative binomial regression for in-depth corpus analysis, our findings demonstrate a consistent influence of magnitudes and roundness, with smaller magnitudes and rounder numbers appearing more frequently. We observe a significant standardization in portraying large numbers, marked by a shift from numerical to mixed forms (e.g., "6,000,000,000" to "6 billion") around 1940s. This reflects changes in both formal writing conventions and editorial practices of numerical representation. Our research further identifies distinct culturally significant numbers for each decade, linked to social, economic, and technological trends, underscoring the role of numerical analysis in media to decode complex cultural and societal patterns. This study contributes significantly to understanding the dynamic interplay between language, culture, and media in the context of numerical representations. • Culturally significant numbers in TIME magazine across decades are determined. • Numerical number in TIME magazine shifted to mixed forms around 1950s. • Number use is consistently affected by magnitudes and roundness across decades. • Numerical representations show a dynamic interplay of language, culture, and media. [ABSTRACT FROM AUTHOR]