Abstract Communication comprises a wealth of multimodal signals (e.g., gestures, eye gaze, intonation) in addition to speech and there is a growing interest in the study of multimodal language by psychologists, linguists, neuroscientists and computer scientists. The ECOLANG corpus provides audiovisual recordings and ELAN annotations of multimodal behaviours (speech transcription, gesture, object manipulation, and eye gaze) by British and American English-speaking adults engaged in semi-naturalistic conversation with their child (N = 38, children 3-4 years old, face-blurred) or a familiar adult (N = 31). Speakers were asked to talk about objects to their interlocutors. We further manipulated whether the objects were familiar or novel to the interlocutor and whether the objects could be seen and manipulated (present or absent) during the conversation. These conditions reflect common interaction scenarios in real-world communication. Thus, ECOLANG provides ecologically-valid data about the distribution and co-occurrence of multimodal signals across these conditions for cognitive scientists and neuroscientists interested in addressing questions concerning real-world language acquisition, production and comprehension, and for computer scientists to develop multimodal language models and more human-like artificial agents.
Early language development is crucial for children’s cognitive and social growth. This study aims to investigate the relationship between parental linguistic stimulation, emotional support, and children’s speaking ability. The research involved 30 preschool and kindergarten children at KB-TK Amanah Qurani in the 2024/2025 academic year. Data were collected using questionnaires completed by teachers and parents. Due to non-normal distribution in two variables, Spearman Rank Correlation was used for analysis. Results show significant positive correlations among all variables. Parental linguistic stimulation and emotional support are strongly correlated (? = 0.735, p < 0.001). Children’s speaking ability has a moderate positive correlation with linguistic stimulation (? = 0.573, p = 0.001) and a strong positive correlation with emotional support (? = 0.640, p < 0.001). These findings indicate that higher parental involvement in language and emotional development is associated with better speaking skills in early childhood. This study emphasizes the vital role of both cognitive and emotional support in language acquisition and recommends greater focus on parent-child interaction in early education programs.
Language. Linguistic theory. Comparative grammar, English language
This study highlights the role of idioms and expressions in language acquisition and teaching as essential and basic tools for communicating in any language. Learning idioms, and their complexities and simplicity, creates a significant barrier for language learners because it involves interpretation and comprehension, as there are multi-layered meanings attached to cultural dimensions. This study investigated learners’ and teachers’ opinions on the importance of idioms and expressions in enhancing their linguistic competence and language skills, as well as the challenges faced in learning and teaching them. A qualitative approach was employed to dig into the participants’ experiences. To this end, eight teachers and twenty-seven English language learners from a private language institute were interviewed using semi-structured interviews. It was found that learners realized the importance of idioms in their journey toward fluency in the target language culture, though it was a long journey. This gives particular focus on context-enriched learning, dynamic learning, and exploring cultural studies to boost learners’ knowledge and use of idiomatic expressions. It's training meaningful images into a context for which learners construct a linguistic and cultural understanding of how to operate those images across their entire life span. This study suggests that idiomatic language be given due recognition and that processes be developed to adequately render idiomatic language relevant for improved language learning and acquisition.
Language and Literature, Indo-Iranian languages and literature
Pre-trained language models have achieved remarkable success across diverse applications but remain susceptible to spurious, concept-driven correlations that impair robustness and fairness. In this work, we introduce CURE, a novel and lightweight framework that systematically disentangles and suppresses conceptual shortcuts while preserving essential content information. Our method first extracts concept-irrelevant representations via a dedicated content extractor reinforced by a reversal network, ensuring minimal loss of task-relevant information. A subsequent controllable debiasing module employs contrastive learning to finely adjust the influence of residual conceptual cues, enabling the model to either diminish harmful biases or harness beneficial correlations as appropriate for the target task. Evaluated on the IMDB and Yelp datasets using three pre-trained architectures, CURE achieves an absolute improvement of +10 points in F1 score on IMDB and +2 points on Yelp, while introducing minimal computational overhead. Our approach establishes a flexible, unsupervised blueprint for combating conceptual biases, paving the way for more reliable and fair language understanding systems.
The advent of Large Language Models (LLMs) has raised concerns about their enormous carbon footprint, starting with energy-intensive training and continuing through repeated inference. This study investigates the potential of using fine-tuned Small Language Models (SLMs) as a sustainable alternative for predefined tasks. Here, we present a comparative analysis of the performance-emissions trade-off between LLMs and fine-tuned SLMs across selected tasks under Natural Language Processing, Reasoning and Programming. Our results show that in four out of the six selected tasks, SLMs maintained comparable performances for a significant reduction in carbon emissions during inference. Our findings demonstrate the viability of smaller models in mitigating the environmental impact of resource-heavy LLMs, thus advancing towards sustainable, green AI.
A prominent achievement of natural language processing (NLP) is its ability to understand and generate meaningful human language. This capability relies on complex feedforward transformer block architectures pre-trained on large language models (LLMs). However, LLM pre-training is currently feasible only for a few dominant companies due to the immense computational resources required, limiting broader research participation. This creates a critical need for more accessible alternatives. In this study, we explore whether tiny language models (TLMs) exhibit the same key qualitative features of LLMs. We demonstrate that TLMs exhibit a clear performance gap between pre-trained and non-pre-trained models across classification tasks, indicating the effectiveness of pre-training, even at a tiny scale. The performance gap increases with the size of the pre-training dataset and with greater overlap between tokens in the pre-training and classification datasets. Furthermore, the classification accuracy achieved by a pre-trained deep TLM architecture can be replicated through a soft committee of multiple, independently pre-trained shallow architectures, enabling low-latency TLMs without affecting classification accuracy. Our results are based on pre-training BERT-6 and variants of BERT-1 on subsets of the Wikipedia dataset and evaluating their performance on FewRel, AGNews, and DBPedia classification tasks. Future research on TLM is expected to further illuminate the mechanisms underlying NLP, especially given that its biologically inspired models suggest that TLMs may be sufficient for children or adolescents to develop language. The data and code that support the findings of this study are openly available on https://github.com/Rg32601/Tiny-Language-Models .
GPT-4V has attracted considerable attention due to its extraordinary capacity for integrating and processing multimodal information. At the same time, its ability of face recognition raises new safety concerns of privacy leakage. Despite researchers' efforts in safety alignment through RLHF or preprocessing filters, vulnerabilities might still be exploited. In our study, we introduce AutoJailbreak, an innovative automatic jailbreak technique inspired by prompt optimization. We leverage Large Language Models (LLMs) for red-teaming to refine the jailbreak prompt and employ weak-to-strong in-context learning prompts to boost efficiency. Furthermore, we present an effective search method that incorporates early stopping to minimize optimization time and token expenditure. Our experiments demonstrate that AutoJailbreak significantly surpasses conventional methods, achieving an Attack Success Rate (ASR) exceeding 95.3\%. This research sheds light on strengthening GPT-4V security, underscoring the potential for LLMs to be exploited in compromising GPT-4V integrity.
Despite large language models' (LLMs) recent advancements, their bias and hallucination issues persist, and their ability to offer consistent preferential rankings remains underexplored. This study investigates the capacity of LLMs to provide consistent ordinal preferences, a crucial aspect in scenarios with dense decision space or lacking absolute answers. We introduce a formalization of consistency based on order theory, outlining criteria such as transitivity, asymmetry, reversibility, and independence from irrelevant alternatives. Our diagnostic experiments on selected state-of-the-art LLMs reveal their inability to meet these criteria, indicating a strong positional bias and poor transitivity, with preferences easily swayed by irrelevant alternatives. These findings highlight a significant inconsistency in LLM-generated preferential rankings, underscoring the need for further research to address these limitations.
Jannik Peters, Constantin Waubert de Puiseau, Hasan Tercan
et al.
The field of emergent language represents a novel area of research within the domain of artificial intelligence, particularly within the context of multi-agent reinforcement learning. Although the concept of studying language emergence is not new, early approaches were primarily concerned with explaining human language formation, with little consideration given to its potential utility for artificial agents. In contrast, studies based on reinforcement learning aim to develop communicative capabilities in agents that are comparable to or even superior to human language. Thus, they extend beyond the learned statistical representations that are common in natural language processing research. This gives rise to a number of fundamental questions, from the prerequisites for language emergence to the criteria for measuring its success. This paper addresses these questions by providing a comprehensive review of 181 scientific publications on emergent language in artificial intelligence. Its objective is to serve as a reference for researchers interested in or proficient in the field. Consequently, the main contributions are the definition and overview of the prevailing terminology, the analysis of existing evaluation methods and metrics, and the description of the identified research gaps.
Urjitkumar Patel, Fang-Chun Yeh, Chinmay Gondhalekar
et al.
In the rapidly evolving financial sector, the accurate and timely interpretation of market news is essential for stakeholders needing to navigate unpredictable events. This paper introduces FANAL (Financial Activity News Alerting Language Modeling Framework), a specialized BERT-based framework engineered for real-time financial event detection and analysis, categorizing news into twelve distinct financial categories. FANAL leverages silver-labeled data processed through XGBoost and employs advanced fine-tuning techniques, alongside ORBERT (Odds Ratio BERT), a novel variant of BERT fine-tuned with ORPO (Odds Ratio Preference Optimization) for superior class-wise probability calibration and alignment with financial event relevance. We evaluate FANAL's performance against leading large language models, including GPT-4o, Llama-3.1 8B, and Phi-3, demonstrating its superior accuracy and cost efficiency. This framework sets a new standard for financial intelligence and responsiveness, significantly outstripping existing models in both performance and affordability.
Runsheng "Anson" Huang, Lara J. Martin, Chris Callison-Burch
WHAT-IF -- Writing a Hero's Alternate Timeline through Interactive Fiction -- is a system that uses zero-shot meta-prompting to create branching narratives from a prewritten story. Played as an interactive fiction (IF) game, WHAT-IF lets the player choose between decisions that the large language model (LLM) GPT-4 generates as possible branches in the story. Starting with an existing linear plot as input, a branch is created at each key decision taken by the main character. By meta-prompting the LLM to consider the major plot points from the story, the system produces coherent and well-structured alternate storylines. WHAT-IF stores the branching plot tree in a graph which helps it to both keep track of the story for prompting and maintain the structure for the final IF system. A demo of WHAT-IF can be found at https://what-if-game.github.io/.
Neurocognitive studies of the translation (including interpreting) process have developed quickly for decades. They not only shed new light on the black box of the translating brain but also spark discussions in neurolinguistic and neurocognitive issues in the translation process, which are firstly explicated in the paper, underlying a more in-depth explanation of a dynamic view of the neurocognition of translation. This dynamic view is further expounded by a comprehensive analysis of the research findings of ten representative neurocognitive studies in terms of cognitive components and relevant brain areas involved in the translation process. The above analysis reveals the connectivity and complexity of the neural basis of translation and demonstrates the neurocognitive variety of translating under the influence of different factors, such as language proficiency, the translator's age of second language acquisition, translation directionality, and the specific tasks of translation or interpreting, among others. The explication of the dynamic view offers a neurocognitive lens to observe the neural basis of translation psychology more comprehensively on the one hand and raises further problems to explore on the other.
Jianghong Zhou, Bo Liu, Jhalak Nilesh Acharya Yao Hong
et al.
In the dynamic field of eCommerce, the quality and comprehensiveness of product descriptions are pivotal for enhancing search visibility and customer engagement. Effective product descriptions can address the 'cold start' problem, align with market trends, and ultimately lead to increased click-through rates. Traditional methods for crafting these descriptions often involve significant human effort and may lack both consistency and scalability. This paper introduces a novel methodology for automating product description generation using the LLAMA 2.0 7B language model. We train the model on a dataset of authentic product descriptions from Walmart, one of the largest eCommerce platforms. The model is then fine-tuned for domain-specific language features and eCommerce nuances to enhance its utility in sales and user engagement. We employ multiple evaluation metrics, including NDCG, customer click-through rates, and human assessments, to validate the effectiveness of our approach. Our findings reveal that the system is not only scalable but also significantly reduces the human workload involved in creating product descriptions. This study underscores the considerable potential of large language models like LLAMA 2.0 7B in automating and optimizing various facets of eCommerce platforms, offering significant business impact, including improved search functionality and increased sales.
Shengyun Gu, Deborah Chen Pichler, L. Viola Kozak
et al.
In this study, we conducted a pseudosign (nonce sign) repetition task with 22 children (mean age: 6;04) acquiring American Sign Language (ASL) as a first language (L1) from deaf parents. Thirty-nine pseudosigns with varying complexity were developed and organized into eight categories depending on number of hands, number of simultaneous movement types, and number of movement sequences. Pseudosigns also varied in handshape complexity. The children’s performance on the ASL pseudosign task improved with age, displaying relatively accurate (re)production of location and orientation, but much less accurate handshape and movement, a finding in line with real sign productions for both L1 and L2 signers. Handshapes with higher complexity were correlated with lower accuracy in the handshape parameter. We found main effects of sequential and simultaneous movement combinations on overall performance. Items with no movement sequence were produced with higher overall accuracy than those with a movement sequence. Items with two simultaneous movement types or a single movement type were produced with higher overall accuracy than those with three simultaneous movement types. Finally, number of hands did not affect the overall accuracy. Remarkably, movement sequences impose processing constraints on signing children whereas complex hands (two hands) and two simultaneous movement types do not significantly lower accuracy, indicating a capacity for processing multiple simultaneous components in signs. Spoken languages, in contrast, manifest greater complexity in temporal length. Hearing children’s pseudoword repetition still displays high levels of accuracy on disyllabic words, with complexity effects affecting only longer multisyllabic words. We conclude that the pseudosign repetition task is an informative tool for studies of signing children’s phonological development and that sheds light on potential modality effects for phonological development.
This study examines several issues concerning the acquisition of evidential markers in Chinese with Taiwan Mandarin-speaking children, including evidential type, acquisition order, and age effect. A production task (i.e., picture-description task) and a comprehension task (i.e., multiple-choice task) were completed by forty children who were divided into two age groups, 3-year-olds and 5-year-olds. Twenty adult native speakers of Taiwan Mandarin were recruited as controls.
The results showed that direct evidential markers were comprehended and produced prior to indirect. For the order of acquisition, the child participants found visual markers and non-visual markers equally easy, while they understood reported markers better than inferring markers. Finally, age was identified as a crucial factor in children’s acquisition of Chinese evidential markers. The 3-year-olds had some success with the use of direct evidential markers, but they still had difficulties with indirect evidential markers. The 5-year-olds significantly outperformed the 3-year-olds but they did not reach an adult-like level.
Special aspects of education, Language acquisition
Salih Furkan Akkurt, Büşra Marşan, Susan Uskudarli
The value of quality treebanks is steadily increasing due to the crucial role they play in the development of natural language processing tools. The creation of such treebanks is enormously labor-intensive and time-consuming. Especially when the size of treebanks is considered, tools that support the annotation process are essential. Various annotation tools have been proposed, however, they are often not suitable for agglutinative languages such as Turkish. BoAT v1 was developed for annotating dependency relations and was subsequently used to create the manually annotated BOUN Treebank (UD_Turkish-BOUN). In this work, we report on the design and implementation of a dependency annotation tool BoAT v2 based on the experiences gained from the use of BoAT v1, which revealed several opportunities for improvement. BoAT v2 is a multi-user and web-based dependency annotation tool that is designed with a focus on the annotator user experience to yield valid annotations. The main objectives of the tool are to: (1) support creating valid and consistent annotations with increased speed, (2) significantly improve the user experience of the annotator, (3) support collaboration among annotators, and (4) provide an open-source and easily deployable web-based annotation tool with a flexible application programming interface (API) to benefit the scientific community. This paper discusses the requirements elicitation, design, and implementation of BoAT v2 along with examples.
We explore the idea of compressing the prompts used to condition language models, and show that compressed prompts can retain a substantive amount of information about the original prompt. For severely compressed prompts, while fine-grained information is lost, abstract information and general sentiments can be retained with surprisingly few parameters, which can be useful in the context of decode-time algorithms for controllability and toxicity reduction. We explore contrastive conditioning to steer language model generation towards desirable text and away from undesirable text, and find that some complex prompts can be effectively compressed into a single token to guide generation. We also show that compressed prompts are largely compositional, and can be constructed such that they can be used to control independent aspects of generated text.