Hasil untuk "Greek language and literature. Latin language and literature"

Alexander E. Walker

The article discusses why Classics is important and why its study benefits not just university students but also young children. It was runner-up in the Intermediate Category for a Classical Association competition in 2025. The article explores the value of Classics as a wonderfully diverse subject involving the study of history, archaeology, architecture, art, and literature. Classics enables students to study over a thousand years of history, to uncover cultural values, to discover how language operates, to develop critical analysis skills, and to delight in its timeless literature. The article explores how the study of Classics can benefit young students’ reading and writing proficiency and can be especially beneficial for those with special educational needs. It explores how Latin translation builds code cracking and cognitive skills, not only developing grammatical knowledge but also encouraging problem-solving suited to science, technology, engineering, and mathematics (STEM)-based subjects such as computer science and maths. The article looks at the interdisciplinary nature of the subject and the benefits of learning Latin and Greek vocabulary for language learning and science.

Muhammad Rehan

This paper challenges the traditional etymology of Greek μοιχός as a τομός-type noun derived from PIE *h₃mei̯g̑ʰ- ‘to piss’ and proposes to trace back the word to PIE *mei̯k̑- ‘to mix’. I re-evaluate the Latin and Sanskrit evidence cited in favor of a derivation for μοιχός from *h₃mei̯g̑ʰ- and find it inconclusive for the salient details of the semantic development proposed in the earlier literature. Moreover, the lateness of both the Sanskrit and Latin evidence points towards parallel innovations of the meaning ‘to ejaculate’ from ‘to piss’. Instead, the insight that reflexes of *mei̯k̑- frequently denote extramarital intercourse across Indo-European languages and especially Greek, notably in Homeric uses of μίσγω and ἐμίγην in adulterous contexts, provides semantic support for deriving μοιχός from PIE *mei̯k̑-. I formalize the derivation through a rátha-type derivational chain (*mei̯k̑- → *moi̯k̑-éh₂ → *moi̯k̑-h₂-ó-s > μοιχός) and adduce evidence from the Balto-Slavic branch for the intermediate τομή-type abstract. This analysis circumvents the difficulties posed by a lack of incontrovertible Greek evidence for the reconstruction of word-initial Saussure-Hirt effect (*#HRo > *#Ro) in PIE, especially when the resonant has the feature [–nasal]. A full re-evaluation of the evidence is, however, left for future research.

Tim Adelani

Alina Klerings, Jannik Brinkmann, Daniel Ruffinelli et al.

Large language models (LLMs) are able to generate grammatically well-formed text, but how do they encode their syntactic knowledge internally? While prior work has focused largely on binary grammatical contrasts, in this work, we study the representation and control of two multidimensional hierarchical grammar phenomena - verb tense and aspect - and for each, identify distinct, orthogonal directions in residual space using linear discriminant analysis. Next, we demonstrate causal control over both grammatical features through concept steering across three generation tasks. Then, we use these identified features in a case study to investigate factors influencing effective steering in multi-token generation. We find that steering strength, location, and duration are crucial parameters for reducing undesirable side effects such as topic shift and degeneration. Our findings suggest that models encode tense and aspect in structurally organized, human-like ways, but effective control of such features during generation is sensitive to multiple factors and requires manual tuning or automated optimization.

Richard Kimera, Dongnyeong Heo, Daniela N. Rim et al.

In this paper,we explore the application of Back translation (BT) as a semi-supervised technique to enhance Neural Machine Translation(NMT) models for the English-Luganda language pair, specifically addressing the challenges faced by low-resource languages. The purpose of our study is to demonstrate how BT can mitigate the scarcity of bilingual data by generating synthetic data from monolingual corpora. Our methodology involves developing custom NMT models using both publicly available and web-crawled data, and applying Iterative and Incremental Back translation techniques. We strategically select datasets for incremental back translation across multiple small datasets, which is a novel element of our approach. The results of our study show significant improvements, with translation performance for the English-Luganda pair exceeding previous benchmarks by more than 10 BLEU score units across all translation directions. Additionally, our evaluation incorporates comprehensive assessment metrics such as SacreBLEU, ChrF2, and TER, providing a nuanced understanding of translation quality. The conclusion drawn from our research confirms the efficacy of BT when strategically curated datasets are utilized, establishing new performance benchmarks and demonstrating the potential of BT in enhancing NMT models for low-resource languages.

Hitomi Yanaka, Xinqi He, Jie Lu et al.

An increasing number of studies have examined the social bias of rapidly developed large language models (LLMs). Although most of these studies have focused on bias occurring in a single social attribute, research in social science has shown that social bias often occurs in the form of intersectionality -- the constitutive and contextualized perspective on bias aroused by social attributes. In this study, we construct the Japanese benchmark inter-JBBQ, designed to evaluate the intersectional bias in LLMs on the question-answering setting. Using inter-JBBQ to analyze GPT-4o and Swallow, we find that biased output varies according to its contexts even with the equal combination of social attributes.

Samuel Ebimobowei Johnny, Blessed Guda, Andrew Blayama Stephen et al.

Continuously recognizing sign gestures and converting them to glosses plays a key role in bridging the gap between the hearing and hearing-impaired communities. This involves recognizing and interpreting the hands, face, and body gestures of the signer, which pose a challenge as it involves a combination of all these features. Continuous Sign Language Recognition (CSLR) methods rely on multi-stage pipelines that first extract visual features, then align variable-length sequences with target glosses using CTC or HMM-based approaches. However, these alignment-based methods suffer from error propagation across stages, overfitting, and struggle with vocabulary scalability due to the intermediate gloss representation bottleneck. To address these limitations, we propose AutoSign, an autoregressive decoder-only transformer that directly translates pose sequences to natural language text, bypassing traditional alignment mechanisms entirely. The use of this decoder-only approach allows the model to directly map between the features and the glosses without the need for CTC loss while also directly learning the textual dependencies in the glosses. Our approach incorporates a temporal compression module using 1D CNNs to efficiently process pose sequences, followed by AraGPT2, a pre-trained Arabic decoder, to generate text (glosses). Through comprehensive ablation studies, we demonstrate that hand and body gestures provide the most discriminative features for signer-independent CSLR. By eliminating the multi-stage pipeline, AutoSign achieves substantial improvements on the Isharah-1000 dataset, achieving an improvement of up to 6.1\% in WER score compared to the best existing method.

Wenhong Zhu, Weinan Zhang, Rui Wang

Post-training alignment has increasingly become a crucial factor in enhancing the usability of language models (LMs). However, the strength of alignment varies depending on individual preferences. This paper proposes a method to incorporate alignment control into a single model, referred to as CLM. This approach adds one identity layer preceding the initial layers and performs preference learning only on this layer to map unaligned input token embeddings into the aligned space. Experimental results demonstrate that this efficient fine-tuning method performs comparable to full fine-tuning. During inference, the input embeddings are processed through the aligned and unaligned layers, which are then merged through the interpolation coefficient. By controlling this parameter, the alignment exhibits a clear interpolation and extrapolation phenomenon.

Sabiya Nuridinova

The translation of dental terminology from English into Uzbek presents a range of linguistic and conceptual challenges due to structural differences between the two languages and the unequal development of specialized medical lexicons. English dental terms are largely derived from Latin and Greek, whereas Uzbek medical terminology combines native lexical elements with international borrowings. This article investigates the major difficulties encountered in translating English dental terms into Uzbek, including semantic non-equivalence, morphological adaptation and terminological inconsistency.

M. Cristea

The first half of the twentieth century was crucial for the formation of modern Romanian education and culture, including in the field of classical studies. The year 1919 was one of transformation, marked by the reorganization of the former Hungarian Royal University “Francisc Iosef” as a Romanian university, where the study of the ancient Greek and Latin languages, literatures, and cultures was con-sidered essential not only to the advancement of Romanian educa-tion in Transylvania but also to the identity of the newly united na-tion, integral to its social and cultural progress. Together with his professors and colleagues, Ștefan Bezdechi contributed through his teaching and philological work to inspiring students and training au-thentic classicists, in order to promote excellence in Romanian liter-ary culture. His intellectual curiosity encompassed all periods and genres of ancient Greek literature, Byzantine hagiographic literature, and—no less—Neo-Greek lyric poetry from the nineteenth century to the early twentieth century. The themes he approached are ar-borescent in conception: the philological and historiographic threads are braided with philosophical, ethical, and aesthetic issues, all inter-twined within a historical and social perspective that approximates the life and thought of a world that, although it reaches us only through its literary documents, endures in posterity.

Nikolay Aretov

This study examines the formation and durability of the Bulgarian literary canon. Bulgarian literature is primarily written in Bulgarian, and it includes texts in other languages, such as Latin, Greek, and Russian, as well as works by minorities and emigrants. Bulgarian culture simultaneously falls into two communities: one linguistic (Slavic) and the other territorial (Balkan). The tension between these communities is influenced by a sense of belonging to different cultural identities, such as the Slavic, Balkan, and European communities. The territorial answer to Bulgarian literature became satisfactory only after the establishment of the new Bulgarian state in 1878. Before that, Bulgarian literature developed within a multinational empire, and much of it was created and published outside the territories predominantly inhabited by Bulgarians. In the 20th and 21st centuries, Bulgarian culture has continued to attract texts from traditional migrant groups outside Bulgarian territories, though literary history has been more cautious about including them in the corpus of Bulgarian literature. In the early 20th century, the core of the Bulgarian literary canon was established—Hristo Botev, Ivan Vazov, Peyo Yavorov, Elin Pelin, and Yordan Yovkov. This canon remains largely unchanged to this day. Adjustments began during that period and even earlier, becoming particularly relevant after the fall of the Berlin Wall.

Hüseyin Yıldız, Nargiza Jamalova, Abdulkadir Öztürk

In the 19th century, Armenians living within the borders of the Ottoman Empire who spoke Turkish produced numerous works on religion, language, history, literature, and other subjects using their own alphabet.One notable aspect of missionary activity aimed at spreading Christianity among the Turks was the translation of the Bible into Turkish. The work examined in this study is a translation of the Bible into late Ottoman Turkish, narrating the life, teachings, and miracles of Jesus Christ –from his birth to his final days. The full title of the work is Yeni Ahit - İncil-i Şerif Arakâlların Amelleri, Boğosun ve Sayip Arakâlların Mektüpleri. It was originally translated from Greek into Turkish, and published in Istanbul in 1858. The text includes selected sections and epistles from the New Testament, one of the foundational scriptures of Christianity. The publication and dissemination of such works in the Ottoman Empire were intended to increase the accessibility of religious texts to local communities in various languages and scripts. During this period, Ottoman Turkish was frequently used for translating religious texts. These texts were often written not only in Arabic script, but also in Latin, Greek, and Armenian scripts. The use of Ottoman Turkish in this particular work is significant, as it reflects the linguistic characteristics of the era. From a cultural and linguistic perspective, Armenians established close relations with both Kipchaks and Anatolian Turks throughout history. As a result of these interactions, distinct written traditions emerged, such as “Kipchak written in Armenian script” and “Turkish written in Armenian script”, accompanied by a rich corpus of literary works. In recent years, scholarly interest in these two linguistic traditions has grown significantly in Turkey.This study analyzes the phonological and orthographic features of a Bible translation written in late Ottoman Turkish using the Armenian script. By presenting detailed phonetic and orthographic observations, this research aims to contribute to broader research on the phonological characteristics of the Turkish language during the late Ottoman period.

The term “purism” (from the Latin purus “pure”) refers to an oversensitivity to purity. Such an oversensitivity might manifest itself in sports, in a certain doctrine, or in literature, art, or language. A sporting purist, for instance, might refuse to recognize any disciplines other than those once practiced by the ancient Greeks. A doctrinal purist will doggedly defend the first principles of his or her particular doctrine, no matter how they actually function in today’s world. A purist critic will rip to shreds any work whose creator dares to cross different styles or genres. And a linguistic purist will cry: “Down with all borrowings, neologisms, colloquialisms!” – railing against anything he or she feels to be undesirable, polluting the language’s purity.

Jodie Reynolds

Marco Carrozza

Il presente contributo si propone di approfondire, sulla scia di studi già personalmente condotti, ulteriori sfaccettature di un epillio tardo-bizantino poco noto (Michele Coniata, Theanó), eppure estremamente interessante sia sotto il profilo letterario che sotto quello più latamente culturale. In particolar modo ci soffermeremo su due sezioni del poema, ovvero il proemio e la rivelazione ispirata della (neo)pitagorica Theanó, annunciata, non a caso, fin dal principio del poemetto. Per esigenze di accessibilità al testo e alla traduzione da noi fornita, i versi presi in esame sono riportati come appendice al presente contributo.

Julia Fröhlich

Tim Schopf, Florian Matthes

Scientific literature searches are often exploratory, whereby users are not yet familiar with a particular field or concept but are interested in learning more about it. However, existing systems for scientific literature search are typically tailored to keyword-based lookup searches, limiting the possibilities for exploration. We propose NLP-KG, a feature-rich system designed to support the exploration of research literature in unfamiliar natural language processing (NLP) fields. In addition to a semantic search, NLP-KG allows users to easily find survey papers that provide a quick introduction to a field of interest. Further, a Fields of Study hierarchy graph enables users to familiarize themselves with a field and its related areas. Finally, a chat interface allows users to ask questions about unfamiliar concepts or specific articles in NLP and obtain answers grounded in knowledge retrieved from scientific publications. Our system provides users with comprehensive exploration possibilities, supporting them in investigating the relationships between different fields, understanding unfamiliar concepts in NLP, and finding relevant research literature. Demo, video, and code are available at: https://github.com/NLP-Knowledge-Graph/NLP-KG-WebApp.

Lance Calvin Lim Gamboa, Mark Lee

Bias studies on multilingual models confirm the presence of gender-related stereotypes in masked models processing languages with high NLP resources. We expand on this line of research by introducing Filipino CrowS-Pairs and Filipino WinoQueer: benchmarks that assess both sexist and anti-queer biases in pretrained language models (PLMs) handling texts in Filipino, a low-resource language from the Philippines. The benchmarks consist of 7,074 new challenge pairs resulting from our cultural adaptation of English bias evaluation datasets, a process that we document in detail to guide similar forthcoming efforts. We apply the Filipino benchmarks on masked and causal multilingual models, including those pretrained on Southeast Asian data, and find that they contain considerable amounts of bias. We also find that for multilingual models, the extent of bias learned for a particular language is influenced by how much pretraining data in that language a model was exposed to. Our benchmarks and insights can serve as a foundation for future work analyzing and mitigating bias in multilingual models.

Lars Klöser, Mika Beele, Jan-Niklas Schagen et al.

This study pioneers the use of synthetically generated data for training generative models in document-level text simplification of German texts. We demonstrate the effectiveness of our approach with real-world online texts. Addressing the challenge of data scarcity in language simplification, we crawled professionally simplified German texts and synthesized a corpus using GPT-4. We finetune Large Language Models with up to 13 billion parameters on this data and evaluate their performance. This paper employs various methodologies for evaluation and demonstrates the limitations of currently used rule-based metrics. Both automatic and manual evaluations reveal that our models can significantly simplify real-world online texts, indicating the potential of synthetic data in improving text simplification.

David Wadden, Kejian Shi, Jacob Morrison et al.

We present SciRIFF (Scientific Resource for Instruction-Following and Finetuning), a dataset of 137K instruction-following instances for training and evaluation, covering 54 tasks. These tasks span five core scientific literature understanding capabilities: information extraction, summarization, question answering, claim verification, and classification. SciRIFF is unique in being entirely expert-written, high-quality instruction-following dataset for extracting and synthesizing information from research literature across diverse scientific fields. It features complex instructions with long input contexts, detailed task descriptions, and structured outputs. To demonstrate its utility, we finetune a series of large language models (LLMs) using a mix of general-domain and SciRIFF instructions. On nine out-of-distribution held-out tasks (referred to as SciRIFF-Eval), LLMs finetuned on SciRIFF achieve 70.6% average improvement over baselines trained only on general-domain instructions. SciRIFF facilitates the development and evaluation of LLMs to help researchers navigate the rapidly growing body of scientific literature.