Small Corpus Research Articles

Media, politics, and scandal: public scandal as preparation for war

ABSTRACT This contribution is based on the thesis of a connection between media scandals, political scandals, including the scandalization of society, and the language used in everyday settings. Different kinds of scandals in politics, the media, and society have been common in the Russian public sphere. This could be seen as the preparation of an atmosphere in which overaggressive, demagogical communication is valued, where victims are portrayed as self-inflicted or even become the “real perpetrators.” This article aims to explore the extent to which this kind of scandalization and the provocation associated with it diffuse into social discourses, for example, everyday language. To do so, the characteristics of scandalous language are first elaborated based on the literature and a small corpus of media coverage of scandals. Then, data from YouTube with political as well as non-political content is examined to see if the same linguistic mode is used in these different communicative areas. The comments and discussions are analyzed in terms of their use of language, with a focus on the lexical and other indicators of a language of scandal. Its existence, dissemination, and linguistic integration are key to finding hints of a more general preference for provocative language.

Distributional hypothesis as isomorphism between word-word co-occurrence and analogical parallelograms.

Most of the modern natural language processing (NLP) techniques are based on the vector space models of language, in which each word is represented by a vector in a high dimensional space. One of the earliest successes was demonstrated by the four-term analogical reasoning task: what is to C as B is to A? The trained word vectors form "parallelograms" representing the quadruple of words in analogy. This discovery in NLP offers us insight into our understanding of human semantic representation of words via analogical reasoning. Despite successful applications of the large-scale language models, it has not been fully understood why such parallelograms emerge by learning through natural language data. As the vector space model is not optimized to form parallelograms, the key structure related to geometric shapes of word vectors is expected to be in the data, rather than the models. In the present article, we test our hypothesis that such parallelogram arrangement of word vectors readily exists in the co-occurrence statistics of language. Our approach focuses more on the data itself, and it is different from the existing theoretical approach trying to find the mechanism of parallelogram formation in the algorithms and/or vector arithmetic operations on word vectors. First, our analysis suggested that analogical reasoning is possible by decomposition of the bigram co-occurrence matrix. Second, we demonstrated the formation of a parallelepiped, a more structured geometric object than a parallelogram, by creating a small artificial corpus and its word vectors. With these results, we propose a refined form of distributional hypothesis pointing out an isomorphism between a sort of symmetry or exchangeability and word co-occurrence statistics.

Small data in der linguistischen Diskursanalyse?

Quantitative approaches and big data analyses are widespread in current discourse linguistic studies and have influenced the field considerably. However, building on Fix’ ONE-text-discourse-analysis (2015) and using the example of discourses around multilingualism, this paper will try and attempt an exemplary small data analysis, using a very small corpus of just three very different data samples in order to see what kind of insights can be gained with a strictly qualitative approach. The texts that were chosen deal with discourses around multilingualism in very different ways. They include an online newspaper article, language portraits by multilingual school children and a language learning video on methods how to study a language such as German. After analysing all three artefacts and their particular value and insights when it comes to research on discursive attitudes towards multilingualism in Germany, it shall be discussed whether and how this small data approach could generally be useful in combination with other discourse linguistic methods.

Produktverpackungen als alltägliche Manifestationen des Umweltdiskurses

The article outlines the technical, semiotic and discourse-linguistic challenges that arise in the data collection, the data processing and the data analysis of a quite small corpus of current product packaging texts (food/beverages, cleaning agents and hygiene products) of a running DFG-project. These three types of problems are due to the materiality, multimodality and multithematicity of the material analysed. The project which is focused on the German environmental discourse 1990-2020 and which is investigating the tension between individual freedom and social norms expressed in it, thus deliberately leaves the big data level of the originally project context within a bigger research group and switches to the level of hermeneutic-qualitative analyses of selected key texts. The hypothesis underlying this decision is that the environmental discourse is reflected in a condensed form in texts that are committed to corporate sustainability communication and are influencing the discourse space of everyday life in everyone’s household. Initial insights into the material and multimodal nature of the corpus are provided and the first results of the discourse linguistic analysis are offered, whereby the material is always discussed in its special significance "beyond big data". The preliminary results show that the everyday products analysed here are discursive mediators between companies, consumers and the environment, in that responsibility is shifted and displaced several times so that in the end it can to be concluded on all sides: Good for you and good for the planet.

Efficient incremental training using a novel NMT-SMT hybrid framework for translation of low-resource languages.

The data-hungry statistical machine translation (SMT) and neural machine translation (NMT) models offer state-of-the-art results for languages with abundant data resources. However, extensive research is imperative to make these models perform equally well for low-resource languages. This paper proposes a novel approach to integrate the best features of the NMT and SMT systems for improved translation performance of low-resource English-Tamil language pair. The suboptimal NMT model trained with the small parallel corpus translates the monolingual corpus and selects only the best translations, to retrain itself in the next iteration. The proposed method employs the SMT phrase-pair table to determine the best translations, based on the maximum match between the words of the phrase-pair dictionary and each of the individual translations. This repeating cycle of translation and retraining generates a large quasi-parallel corpus, thus making the NMT model more powerful. SMT-integrated incremental training demonstrates a substantial difference in translation performance as compared to the existing approaches for incremental training. The model is strengthened further by adopting a beam search decoding strategy to produce k best possible translations for each input sentence. Empirical findings prove that the proposed model with BLEU scores of 19.56 and 23.49 outperforms the baseline NMT with scores 11.06 and 17.06 for Eng-to-Tam and Tam-to-Eng translations, respectively. METEOR score evaluation further corroborates these results, proving the supremacy of the proposed model.

Frequency and Universal Principles in the Acquisition of Word-Initial Consonant Clusters in Palestinian Arabic.

In this article, we present the results of our study on the acquisition of word-initial consonant clusters in Palestinian Arabic. The remarkable property of word-initial clusters in Palestinian Arabic is that they often violate the Sonority Sequencing Principle, a universal constraint that disfavors sonority fall (e.g., lba:n "gum") and, to a lesser degree, sonority plateau (e.g., kta:b "book"). Given the violation of the Sonority Sequencing Principle, the acquisition of word-initial consonant clusters serves as a fruitful ground for studying the interaction of universal principles and language-specific effects. The goal of the study was to shed light on the role of universal principles in language acquisition and how they interact with frequency. We carried out picture naming and sentence completion tasks with 60 monolingual and monodialectal children aged 2;00-5;00 (years;months), equally divided into six age groups. In addition, we built a small corpus of child-directed speech (CDS) to evaluate the distribution of the three sonority profiles of word-initial consonant clusters-rise, plateau, and fall. We found a major language-specific effect in the cluster simplification strategies, with prothesis being the most common strategy (e.g., lbá:n → ilbá:n "gum"). However, this effect was not found in the youngest group (2;00-2;05), where the cross-linguistically favorite strategy of C-deletion was most common (e.g., lbá:n → bá:n "gum"). We also found a major effect of the Sonority Sequencing Principle in the children's productions, contrary to insignificant differences in CDS. Our study supports the claim that language acquisition is affected by both language-specific effects and universal principles. As for the interaction of these factors, we show that the effect of universal principles emerges under two conditions: (a) in early speech (youngest age group), before children accumulate sufficient data that allow language-specific properties to override, and/or (b) when frequency does not play a role.

Dense and Disconnected: Analyzing the Sedimented Style of ChatGPT-Generated Text at Scale

ChatGPT and other LLMs are at the forefront of pedagogical considerations in classrooms across the academy. Many studies have spoken to the technology’s capacity to generate one-off texts in a variety of genres. This study complements those by inquiring into its capacity to generate compelling texts at scale. In this study, we quantitatively and qualitatively analyze a small corpus of generated texts in two genres and gauge it against novice and published academic writers along known dimensions of linguistic variation. Theoretically, we position and historicize ChatGPT as a writing technology and consider the ways in which generated text may not be congruent with established trajectories of writing development in higher education. Our study found that generated texts are more informationally dense than authored texts and often read as dialogically closed, “empty,” and “fluffy.” We close with a discussion of potentially explanatory linguistic features, as well as relevant pedagogical implications.

Optimizing word embeddings for small datasets: a case study on patient portal messages from breast cancer patients

Patient portal messages often relate to specific clinical phenomena (e.g., patients undergoing treatment for breast cancer) and, as a result, have received increasing attention in biomedical research. These messages require natural language processing and, while word embedding models, such as word2vec, have the potential to extract meaningful signals from text, they are not readily applicable to patient portal messages. This is because embedding models typically require millions of training samples to sufficiently represent semantics, while the volume of patient portal messages associated with a particular clinical phenomenon is often relatively small. We introduce a novel adaptation of the word2vec model, PK-word2vec (where PK stands for prior knowledge), for small-scale messages. PK-word2vec incorporates the most similar terms for medical words (including problems, treatments, and tests) and non-medical words from two pre-trained embedding models as prior knowledge to improve the training process. We applied PK-word2vec in a case study of patient portal messages in the Vanderbilt University Medical Center electric health record system sent by patients diagnosed with breast cancer from December 2004 to November 2017. We evaluated the model through a set of 1000 tasks, each of which compared the relevance of a given word to a group of the five most similar words generated by PK-word2vec and a group of the five most similar words generated by the standard word2vec model. We recruited 200 Amazon Mechanical Turk (AMT) workers and 7 medical students to perform the tasks. The dataset was composed of 1389 patient records and included 137,554 messages with 10,683 unique words. Prior knowledge was available for 7981 non-medical and 1116 medical words. In over 90% of the tasks, both reviewers indicated PK-word2vec generated more similar words than standard word2vec (p = 0.01).The difference in the evaluation by AMT workers versus medical students was negligible for all comparisons of tasks’ choices between the two groups of reviewers (p=0.774\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\ ext{p}} = 0.774$$\\end{document} under a paired t-test). PK-word2vec can effectively learn word representations from a small message corpus, marking a significant advancement in processing patient portal messages.

A Small Brazilian Portuguese Speech Corpus for Speaker Recognition Study

A small Brazilian speech corpus was created for educational purposes to study a state-of-the-art speaker recognition system. The system uses the Gaussian Mixture Model (GMM) as a statistical model for speakers and employs the Mel-frequency cepstral coefficients (MFCC) as acoustic features. The results using clean and noisy speech are compatible with the expected results, showing that the bigger the mismatch between training and test conditions, the worse the results. The results also improve with the increase in the utterance length. Finally, the obtained results can be used as baselines to compare with other speaker statistical models created with different acoustic features in different acoustic conditions.

Smuggling in the Grand Duchy of Lithuania–kingdom of Prussia borderland at the end of the eighteenth century: the case of Palanga customs

ABSTRACT Using a surviving set of monthly confiscation reports covering the period between August 1793 and March 1794 (23 cases in total), this article examines smuggling in the territory subordinated to the Palanga customs on the Lithuanian-Prussian border. The article aims to reveal the patterns of smugglers’ behavior, the characteristics of individuals involved in contraband, as well as the state’s response to these illegal economic activities. In addition to other more context-specific conclusions, the research reveals how even a small corpus of data can allow us to analyze the phenomenon of smuggling from a qualitative perspective.

Eirenias of Miletus’ Career between the Attalids and the Seleucids

The role played by Eirenias of Miletus in the mid-2nd century B.C. between his city and the Attalids and Seleucids demonstrates the vitality of his polis, which had de facto the status of a free city after the Treaty of Apamea of 188. A small corpus of Milesian inscriptions shows that Eirenias, known only from epigraphy, dedicated most of his political activity to relations with external powers, playing numerous times the role of ambassador in favour of his city, for which he was able to obtain many privileges from the Attalids and, to a lesser extent, from the Seleucids. Notables such as Eirenias, who used their external relations for the benefit of their own cites, constituted the connecting element between the euergetism of kings and powerful outsiders and that of private citizens.

IMPOLITENESS IN ONLINE DISCOURSE: ANALYSIS OF READER COMMENTS ON SERBIAN NEWS SITE

Online discourse represents an important part of human communication. When ex- pressing their opinions in an online environment, people are empowered to be more direct, often without considering other people’s face. In certain situations, they may even be offensive or impolite. The focus of this study is the manifestation of impoliteness in reader comments on Serbian news sites. The study adopts Bousfield’s (2008) extended framework of Culpeper’s model of impoliteness strategies (1996, 2005) in order to examine and classify different manifestations of impoliteness. It relies on the analysis of a small specialized corpus which consists of the data collected from reader comments on Serbian online news articles related to Novak Đoković being denied entry into Australia in January 2022 due to the men’s number one tennis player not being vaccinated, which was one of the most vigorously debated topics in both mass media and social media in Serbia amid the coronavirus pandemic. The data were analyzed qualitatively. This study reflects how Serbian language users engage in impolite communicative behavior by identifying the most common impoliteness strategies, as well as their functions, and indicating distinctive features of Serbian impolite online discourse.

Optimizing Word Embeddings for Patient Portal Message Datasets with a Small Number of Samples.

Patient portal messages often relate to specific clinical phenomena (e.g., patients undergoing treatment for breast cancer) and, as a result, have received increasing attention in biomedical research. These messages require natural language processing and, while word embedding models, such as word2vec, have the potential to extract meaningful signals from text, they are not readily applicable to patient portal messages. This is because embedding models typically require millions of training samples to sufficiently represent semantics, while the volume of patient portal messages associated with a particular clinical phenomenon is often relatively small. We introduce a novel adaptation of the word2vec model, PK-word2vec, for small-scale messages. PK-word2vec incorporates the most similar terms for medical words (including problems, treatments, and tests) and non-medical words from two pre-trained embedding models as prior knowledge to improve the training process. We applied PK-word2vec on patient portal messages in the Vanderbilt University Medical Center electric health record system sent by patients diagnosed with breast cancer from December 2004 to November 2017. We evaluated the model through a set of 1000 tasks, each of which compared the relevance of a given word to a group of the five most similar words generated by PK-word2vec and a group of the five most similar words generated by the standard word2vec model. We recruited 200 Amazon Mechanical Turk (AMT) workers and 7 medical students to perform the tasks. The dataset was composed of 1,389 patient records and included 137,554 messages with 10,683 unique words. Prior knowledge was available for 7,981 non-medical and 1,116 medical words. In over 90% of the tasks, both reviewers indicated PK-word2vec generated more similar words than standard word2vec (p=0.01).The difference in the evaluation by AMT workers versus medical students was negligible for all comparisons of tasks' choices between the two groups of reviewers (p = 0.774 under a paired t-test). PK-word2vec can effectively learn word representations from a small message corpus, marking a significant advancement in processing patient portal messages.

Oversampling effect in pretraining for bidirectional encoder representations from transformers (BERT) to localize medical BERT and enhance biomedical BERT

BackgroundPretraining large-scale neural language models on raw texts has made a significant contribution to improving transfer learning in natural language processing. With the introduction of transformer-based language models, such as bidirectional encoder representations from transformers (BERT), the performance of information extraction from free text has improved significantly in both the general and medical domains. However, it is difficult to train specific BERT models to perform well in domains for which few databases of a high quality and large size are publicly available. ObjectiveWe hypothesized that this problem could be addressed by oversampling a domain-specific corpus and using it for pretraining with a larger corpus in a balanced manner. In the present study, we verified our hypothesis by developing pretraining models using our method and evaluating their performance. MethodsOur proposed method was based on the simultaneous pretraining of models with knowledge from distinct domains after oversampling. We conducted three experiments in which we generated (1) English biomedical BERT from a small biomedical corpus, (2) Japanese medical BERT from a small medical corpus, and (3) enhanced biomedical BERT pretrained with complete PubMed abstracts in a balanced manner. We then compared their performance with those of conventional models. ResultsOur English BERT pretrained using both general and small medical domain corpora performed sufficiently well for practical use on the biomedical language understanding evaluation (BLUE) benchmark. Moreover, our proposed method was more effective than the conventional methods for each biomedical corpus of the same corpus size in the general domain. Our Japanese medical BERT outperformed the other BERT models built using a conventional method for almost all the medical tasks. The model demonstrated the same trend as that of the first experiment in English. Further, our enhanced biomedical BERT model, which was not pretrained on clinical notes, achieved superior clinical and biomedical scores on the BLUE benchmark with an increase of 0.3 points in the clinical score and 0.5 points in the biomedical score. These scores were above those of the models trained without our proposed method. ConclusionsWell-balanced pretraining using oversampling instances derived from a corpus appropriate for the target task allowed us to construct a high-performance BERT model.

유학생 에세이에 나타난 대학 생활에 대한 인식 연구

This study aimed to compile a small corpus of essays written by international students studying at South Korean universities, extract high-frequency vocabulary from the corpus, and analyze the content of the essays to research the perceptions of international students about university life. To achieve this, the researchers compiled a small corpus of 68 essays written by foreign international students about their university experiences. The researcher manually typed and saved the materials written by the learners in Word files, preserving any errors made by the learners in the raw data. Subse quently, the raw data were refined by correcting the errors. The processed data were subjected to morphological analysis using the Kkma morphological analyzer, and a word frequency list was generated using the AntConc word module (version 4.2.0). The extracted words were then searched in the essays to understand the context before and after their usage and to analyze the learners’ perceptions. The learners' experiences revealed that the difficulty level of university life could be influenced by their friends and that they attempted to overcome the challenges of their studies with a positive attitude. They also faced challenges in various fields, especially when engaging in team projects or presentations in Korean. Furthermore, university life was evidently not just about academic learning; it also involved ac quiring knowledge related to life’s principles and various aspects of life in general.

On the use of statistical machine translation for suggesting variable names for decompiled code: The Pharo case

Adequately selecting variable names is a difficult activity for practitioners. In 2018, Jaffe et al. proposed the use of statistical machine translation (SMT) to suggest descriptive variable names for decompiled code. A large corpus of decompiled C code was used to train the SMT model. Our paper presents the results of a partial replication of Jaffe’s experiment. We apply the same technique and methodology to a dataset made of code written in the Pharo programming language. We selected Pharo since its syntax is simple – it fits on half of a postcard – and because the optimizations performed by the compiler are limited to method scope. Our results indicate that SMT may recover between 8.9% and 69.88% of the variable names depending on the training set. Our replication concludes that: (i) the accuracy depends on the code similarity between the training and testing sets; (ii) the simplicity of the Pharo syntax and the satisfactory decompiled code alignment have a positive impact on predicting variable names; and (iii) a relatively small code corpus is sufficient to train the SMT model, which shows the applicability of the approach to less popular programming languages. Additionally, to assess SMT’s potential in improving original variable names, ten Pharo developers reviewed 400 SMT name suggestions, with four reviews per variable. Only 15 suggestions (3.75%) were unanimously viewed as improvements, while 45 (11.25%) were perceived as improvements by at least two reviewers, highlighting SMT’s limitations in providing suitable alternatives.

Kulinaryzmy i kulinaria w języku biznesowym w glottodydaktyce italianistycznej na podstawie podręczników „L’italiano in azienda” (2002) i „Un vero affare!” (2015)

The article is devoted to analyzing the cultural context typical for food and its associated etiquetteand rituals from 2 coursebooks on the Italian Business language, and aims at a) investigating to what extent coursebooks in Business Italian include both theoretical and practical content information on the Italian dining culture; b) reconstructing guidelines on etiquette for dining and socializing in professional settings in companies and with regard to the role of language in culture (Hall, 1976, Bańczerowski, et al., 1982). These objectives were fulfilled by performing a semantic analysis of the lexical content of 2 coursebooks by Giovanna Pelizza and Marco Mezzadri (2002 and 2015). For this purpose, a lexical corpus was formed by extracting single-word culinary-related lexical items from the coursebooks. The remaining sources related to food in business culture (Meyer, 2023) was used as a point of reference for validation of the small corpus.

Automatically Testing Functional Properties of Code Translation Models

Large language models are becoming increasingly practical for translating code across programming languages, a process known as transpiling. Even though automated transpilation significantly boosts developer productivity, a key concern is whether the generated code is correct. Existing work initially used manually crafted test suites to test the translations of a small corpus of programs; these test suites were later automated. In contrast, we devise the first approach for automated, functional, property-based testing of code translation models. Our general, user-provided specifications about the transpiled code capture a range of properties, from purely syntactic to purely semantic ones. As shown by our experiments, this approach is very effective in detecting property violations in popular code translation models, and therefore, in evaluating model quality with respect to given properties. We also go a step further and explore the usage scenario where a user simply aims to obtain a correct translation of some code with respect to certain properties without necessarily being concerned about the overall quality of the model. To this purpose, we develop the first property-guided search procedure for code translation models, where a model is repeatedly queried with slightly different parameters to produce alternative and potentially more correct translations. Our results show that this search procedure helps to obtain significantly better code translations.

Imagining women working stone: the ancient sculptor Marcia in Boccaccio’s De mulieribus and responses in fifteenth- and sixteenth-century art literature

This article confronts the widespread stereotype of the supposedly natural inability of women to work stone through an analysis of the iconographic tradition of the ancient sculptor Marcia. A small corpus of fifteenth-century manuscript illuminations shows the ancient artist at work. This case study allows us to trace the memory of a woman artist working in stone that was already circulating at that time. The textual basis for the iconography under discussion was introduced by Giovanni Boccaccio in his biography of Marcia, part of his De mulieribus claris. Boccaccio’s text was received by subsequent generations of authors and reprinted in an Italian version four years before Giorgio Vasari published his account of the life of Properzia de Rossi in his Lives. The article introduces a fresh perspective on the relationship between Boccaccio’s Marcia and Vasari’s Properzia and provides new evidence for connections between the querelles des femmes and sixteenth-century art literature.

ExKidneyBERT: a language model for kidney transplant pathology reports and the crucial role of extended vocabularies.

Pathology reports contain key information about the patient's diagnosis as well as important gross and microscopic findings. These information-rich clinical reports offer an invaluable resource for clinical studies, but data extraction and analysis from such unstructured texts is often manual and tedious. While neural information retrieval systems (typically implemented as deep learning methods for natural language processing) are automatic and flexible, they typically require a large domain-specific text corpus for training, making them infeasible for many medical subdomains. Thus, an automated data extraction method for pathology reports that does not require a large training corpus would be of significant value and utility. To develop a language model-based neural information retrieval system that can be trained on small datasets and validate it by training it on renal transplant-pathology reports to extract relevant information for two predefined questions: (1) "What kind of rejection does the patient show?"; (2) "What is the grade of interstitial fibrosis and tubular atrophy (IFTA)?" Kidney BERT was developed by pre-training Clinical BERT on 3.4K renal transplant pathology reports and 1.5M words. Then, exKidneyBERT was developed by extending Clinical BERT's tokenizer with six technical keywords and repeating the pre-training procedure. This extended the model's vocabulary. All three models were fine-tuned with information retrieval heads. The model with extended vocabulary, exKidneyBERT, outperformed Clinical BERT and Kidney BERT in both questions. For rejection, exKidneyBERT achieved an 83.3% overlap ratio for antibody-mediated rejection (ABMR) and 79.2% for T-cell mediated rejection (TCMR). For IFTA, exKidneyBERT had a 95.8% exact match rate. ExKidneyBERT is a high-performing model for extracting information from renal pathology reports. Additional pre-training of BERT language models on specialized small domains does not necessarily improve performance. Extending the BERT tokenizer's vocabulary library is essential for specialized domains to improve performance, especially when pre-training on small corpora.