Conversational Language Research Articles

AUDIO TO SIGN LANGUAGE CONVERSION

Due to high demand for inclusive technologies and the rapid electrons communication age, there is growing interest in automatic audio to sign language conversion systems. This article has given the relevant way to solve this problem by converting spoken language into sign languages in real time using machine translation that can shape communication path for deaf and dumb community. In our method, we use Natural Language Processing (NLP) to transcribe audio into text and then deep learning for sign language generation. The system recognizes keywords and important phrases in the transcribed text, associates them with corresponding sign language gestures, and presents such signs animations using an avatar-based visualization. The recent accomplishment is different from previous systems that work directly and only translate a sentence to its corresponding sign, rather focusing on the contextual analysis for better conveying of meaning. Key Words: Audio to Sign Language Translation, Speech- to- Text Conversion, Natural Language Processing (NLP), Machine Translation, Deep Learning, Gesture Recognition, Avatar-based Sign Language Visualization, Accessibility, Multilingual Sign Language

Are large language models qualified reviewers in originality evaluation?

Large language models (LLMs) are a new generation of conversational language model with impressive semantic comprehension, text generation, and knowledge inference capabilities. LLMs are significantly influencing the development of science by assisting researchers in analyzing, understanding, and grasping original knowledge in scientific papers. This study investigates LLMs’ potential as qualified reviewers in originality evaluation in zero-shot learning, utilizing a unique, manually crafted prompt. Using biomedical papers as the data source, we constructed two evaluation datasets based on Nobel Prize papers and disruptive index. The evaluation performance of multiple LLMs of different types and scales on the datasets was scrutinized through the analysis of originality score (OS), originality type (OT), and originality description (OD), all of which were generated by the LLM. Our results show that LLMs can to some extent discern papers with distinct originality level via OS; however, they appear to be overly lenient reviewers. In LLMs’ evaluation mechanism, five distinct OTs reflecting varied research contributions do not manifest independently, but together they positively influence OS. Of all the LLMs analyzed, GPT-4 stood out as able to produce the most readable ODs, effectively explaining the inference process for both OS and OT from the perspectives of completeness, logicality, and regularity.

Generative AI and future education: a review, theoretical validation, and authors' perspective on challenges and solutions.

Generative AI (Gen AI), exemplified by ChatGPT, has witnessed a remarkable surge in popularity recently. This cutting-edge technology demonstrates an exceptional ability to produce human-like responses and engage in natural language conversations guided by context-appropriate prompts. However, its integration into education has become a subject of ongoing debate. This review examines the challenges of using Gen AI like ChatGPT in education and offers effective strategies. To retrieve relevant literature, a search of reputable databases was conducted, resulting in the inclusion of twenty-two publications. Using Atlas.ti, the analysis reflected six primary challenges with plagiarism as the most prevalent issue, closely followed by responsibility and accountability challenges. Concerns were also raised about privacy, data protection, safety, and security risks, as well as discrimination and bias. Additionally, there were challenges about the loss of soft skills and the risks of the digital divide. To address these challenges, a number of strategies were identified and subjected to critical evaluation to assess their practicality. Most of them were practical and align with the ethical and pedagogical theories. Within the prevalent concepts, "ChatGPT" emerged as the most frequent one, followed by "AI," "student," "research," and "education," highlighting a growing trend in educational discourse. Moreover, close collaboration was evident among the leading countries, all forming a single cluster, led by the United States. This comprehensive review provides implications, recommendations, and future prospects concerning the use of generative AI in education.

PEERS® for School-Based Professionals via Zoom as a Pre-Employment Transition Service

Previous research has shown that conversational language predicts vocational independence and friendship development. This study investigated the efficacy of the Program for the Education and Enrichment of Relational Skills (PEERS®) for School-Based Professionals delivered via Zoom as a pre-employment transition service on the social competence of teens and young adults with autism spectrum disorder (ASD). The program was administered weekly by vocational rehabilitation counselors via telehealth. A Quasi-experimental pilot design using pre/post-test analysis was used because all students in the vocational rehabilitation case management telehealth group indicated a desire to participate in the PEERS® training, leaving none available to form a comparison group. Results indicate a decrease in the severity of ASD symptomatology regarding social communication and an increase in conversational skills associated with vocational independence. They were invited to and hosted more get-togethers with peers who shared their interests, a measure of future collegiality. Implications for individuals on the autism spectrum and employment transition service providers are discussed.

"Treatable not curable": trade-offs in the use of treatment-oriented language with patients who have incurable cancer.

Treatment-oriented language is used by physicians to convey to patients that treatment is available for their cancer (eg, "our usual treatment for this is…," "we can treat this," "your cancer is still treatable"). For patients who have incurable cancer, especially for patients with a poor prognosis or who are at the end of life, it is important to understand how physicians conceptualize and use this "everyday" clinical language. We conducted a qualitative interview study with a multidisciplinary group of physicians (n = 30) who may care for patients with cancer at different points in their clinical course, from diagnosis to end of life. Physicians report a wide range of reasons for using treatment-oriented language in conversations with patients who have incurable cancer. However, physicians also reported concerns that this language can be ambiguous, can convey unintended positive prognostic information, and can shift attention away from important matters such as the non-curative nature of treatment or the inevitability of death. On the basis of these concerns, physicians should (1) consider whether their aims in using treatment-oriented language can be better achieved using other evidence-based communication strategies, and (2) recognize and proactively mitigate potential adverse effects of treatment-oriented language, which may manifest much later in the patient's clinical course.

Exploring the silence of Japanese EFL learners: Its relationship with the degree of willingness to communicate (WTC)

The concept of willingness to communicate (WTC) is pivotal in understanding student engagement in English as a foreign language (EFL) classrooms. As a dynamic and multifaceted construct, WTC is subject to continual fluctuations throughout the communication process, often changing situationally. Traditionally, learner silence in class or during second language (L2) conversations is perceived negatively, as it can signal a lack of motivation to speak, thereby impeding optimal language acquisition. However, silence can also function as a communicative strategy or reflect learners’ essential feelings and emotions. This study explored the relationship between Japanese EFL learners’ situational WTC and their instances of silence during L2 (English) conversations. Participants were recruited for the study and were asked to deliver short speeches and engage in conversations with a researcher, during which their periods of silence were observed and recorded. All participant utterances were meticulously transcribed and analysed. Additionally, participants self-assessed their WTC for each utterance, and a stimulated recall interview was conducted to gather in-depth qualitative data. The results generally indicated that participants’ WTC was somewhat decreased when silence occurred between and during utterances. However, the findings also revealed the intricate and multifaceted nature of silence in relation to WTC, underscoring the need for a deeper understanding and nuanced consideration of learners’ silence. This study highlights the complexity of the interplay between WTC and silence, suggesting that silence should not be solely viewed as a negative phenomenon but rather as an integral part of the communicative process that warrants careful analysis and understanding.

A Social Group Chatbot System by Multiple Topics Tracking and Atkinson‐Shiffrin Memory Model Using AI Agents Collaboration

ABSTRACTThe widespread use of Internet has accelerated the explosive growth of data, which in turn leads to information overload and information confusion. This makes it difficult for us to communicate effectively in social groups, thereby intensifying the demands for emotional companionship. Therefore, we propose a novel social group chatting framework based on Large Language Model (LLM) powered multiple autonomous agents collaboration in this article. Specifically, BERTopic is used to extract topics from history chatting content for each social group everyday, and then multiple topics tracking is realised through multi‐level association by adaptive time sliding‐window mechanism and optimal matching. Furthermore, we use topic tracking architecture and prompts to design and implement an AI Chatbot system with different characters that can conduct natural language conversations with users in online social group. LLM, as the controller and coordinator of the whole AI Chatbot for sub‐tasks, allows different AI Agents to autonomously decide whether to participate in current topic, how to generate response, and whether to propose a new topic. Each AI Agent has their own multi‐store memory system based on the Atkinson‐Shiffrin model. Finally, we construct a verification environment based on online game that is consistent with real society. Subjective and objective evaluation methods were deployed to perform qualitative and quantitative analyses to demonstrate the performance of our AI Chatbot system.

Improving the Classification of Defect Levels in Main Electrical Equipment Through Advanced Prompting on a Large Language Model

The stable operation of the power system is closely related to the national economy and people’s livelihoods. Therefore, the timely detection, qualitative assessment, and handling of major equipment defects are crucial. The classification of defect levels in main electrical equipment is a fundamental task in this process, which is often manually completed, supplemented by knowledge bases or expert systems. However, this approach is time-consuming, labor-intensive, involves challenging human–machine interaction, and relies on expert experience. Conversational large language models, such as ChatGPT, ERNIE Bot, ChatGLM, have garnered widespread recognition in various domains. However, these models may have errors in the reasoning process, resulting in biased or even erroneous outputs, which is referred to as the “hallucinations”. The hallucinations’ problem of large language models poses challenges in specific fields. To mitigate the hallucinations in large language models, researchers often seek to incorporate domain-specific knowledge into these models through methods like fine-tuning or prompt learning. In order to enhance model performance while minimizing computational costs, this study adopts the prompt learning approach. Specifically, we propose a large language model prompt learning framework based on knowledge graphs, aiming to provide the large language model with reasoning support by leveraging specific information stored within the knowledge graph and receive explainable reasoning result. Experimental results demonstrate that our module achieved superior results on the power defect dataset compared to the non-prompt method.

A study on an efficient OSS inspection scheme based on encrypted GML

The importance of Open Source Software (OSS) has increased in recent years. OSS is software that is jointly developed and maintained globally through open collaboration and knowledge sharing. OSS plays an important role, especially in the Information Technology (IT) field, by increasing the efficiency of software development and reducing costs. However, licensing issues, security issues, etc., may arise when using OSS. Some services analyze source code and provide OSS-related data to solve these problems, a representative example being Blackduck. Blackduck inspects the entiresource code within the project and provides OSS information and related data included in the whole project. Therefore, there are problems such as inefficiency due to full inspection of the source code and difficulty in determining the exact location where OSS is identified. This paper proposes a scheme to intuitively analyze source code through Graph Modelling Language (GML) conversion to solve these problems. Additionally, encryption is applied to GML to performsecure GML-based OSS inspection. The study explains the process of converting source code to GML and performing OSS inspection. Afterward, we compare the capacity and accuracy of text-based OSS inspection and GML-based OSS inspection. Signcryption is applied to performsafe, GML-based, efficient OSS inspection.

Preoperative subjective impairments in language and memory in brain tumor patients.

Subjective reports can reveal relevant information regarding the nature of the impairment of brain tumor patients, unveiling potential gaps in current assessment practices. The co-occurrence of language and memory impairments has been previously reported, albeit scarcely. The aim of this study is therefore to understand the co-occurrence of subjective language and memory complaints in the preoperative state of brain tumor patients and its impact on Quality of Life (QoL). 31 brain tumor patients (12 LGG, 19 HGG) underwent a semi-structured interview to assess subjective complaints of language deficits, co-occurrences between language and memory dysfunction, and changes in QoL. Group and subgroup analyses were conducted to provide general and tumor grade specific data. 48.4% of patients mentioned co-occurrence of language and memory impairments in reading, writing, and conversation. The HGG group reported co-occurrences in all three of these (reading: 31.6%; writing: 21.1%; conversation: 26.3%), while the LGG only described co-occurrences in reading (25%) and conversation (8.3%), although these were not statistically significant. All patients with co-occurring language and memory deficits reported these to be linked to reduced QoL (48.4%). In patients with an HGG, this number was slightly higher (52.6%) than in patients with an LGG (41.7%). Language impairments co-occur with memory dysfunction as perceived in patients' daily life. Patients see these impairments as affecting their quality of life. Further attention to dedicated language and memory tasks seems necessary.

Chatbots in medicine: certification process and applied use case.

Chatbots are computer programs designed to engage in natural language conversations in an easy and understandable way. Their use has been accelerated recently with the advent of large language models. However, their application in medicine and healthcare has been limited due to concerns over data privacy, the risk of providing medical diagnoses, and ensuring regulatory and legal compliance. Medicine and healthcare could benefit from chatbots if their scope is carefully defined and if they are used appropriately and monitored long-term. The confIAnce chatbot, developed at the Geneva University Hospitals and the University of Geneva, is an informational tool aimed at providing simplified information to the general public about primary care and chronic diseases. In this paper, we describe the certification and regulatory aspects applicable to chatbots in healthcare, particularly in primary care medicine. We use the confIAnce chatbot as a case study to explore the definition and classification of a medical device and its application to chatbots, considering the applicable Swiss regulations and the European Union AI Act. Chatbots can be classified anywhere from non-medical devices (informational tools that do not handle patient data or provide recommendations for treatment or diagnosis) to Class III medical devices (high-risk tools capable of predicting potentially fatal events and enabling a pre-emptive medical intervention). Key considerations in the definition and certification process include defining the chatbot's scope, ensuring compliance with regulations, maintaining security and safety, and continuously evaluating performance, risks, and utility. A lexicon of relevant terms related to artificial intelligence in healthcare, medical devices, and regulatory frameworks is also presented in this paper. Chatbots hold potential for both patients and healthcare professionals, provided that their scope of practice is clearly defined, and that they comply with regulatory requirements. This review aims to provide transparency by outlining the steps required for certification and regulatory compliance, making it valuable for healthcare professionals, scientists, developers, and patients.

Exploring ethical issues in the Chinese translation of medical terms: The convergence of translation ethics and medical ethics

This article explores the ethical aspects of translating medical terminology into English, with a particular emphasis on the joint influence of medical ethics and translation ethics. This study identified four key issues in current translation practice through the analysis of various cases from the perspective of Chestermans five kinds of translation ethics, including representation ethics, service ethics, communication ethics, norm-based ethics, and commitment ethics. The four issues involve discrimination, superstition, contempt, and overgeneralization. These issues may hinder the provision of appropriate medical services and social acceptance by fueling social stigma, public misunderstandings, and negative perceptions of diseases. This study suggests that medical translation is not only about language conversion, but also involves considerations of cultural, ethical, and social factors. This article attempts to provide practical advice for medical translators and healthcare workers by combining medical ethics with translation ethics. It suggests that future research should prioritize interdisciplinary collaboration to improve ethical standards in medical translation and promote more accurate and respectful descriptions of diseases in different cultures.

"Look Who's Talking": Language Patterns in Autistic and Non-Autistic Youth Across Different Conversation Partners.

Autistic individuals have varying levels of verbal fluency which can impact social outcomes. Although 70-75% of autistic individuals have functional language, findings regarding language patterns (syntax and semantics) in autistic adolescents remain inconclusive. Additionally, previous studies of language complexity use narrative samples, which do not capture autistic language in conversation. The current study examined language patterns in autistic (n = 20) and non-autistic (n = 17) youth aged 9-16 years during a conversation with a familiar versus unfamiliar adult. The study aimed to address gaps in the literature regarding autistic youth's language patterns, particularly in conversation, and the impact of speaking partners. Recordings of the conversation task were transcribed using SALT software conventions to yield measures of language production. Average length of communication units was higher among autistic compared to non-autistic youth, and among all youth when talking with familiar compared to unfamiliar partners. Youth speech also reflected greater linguistic diversity with familiar interlocutors, with no differences between autistic and non-autistic youth. Additionally, familiar interlocutors used more speech elicitation strategies (i.e., questions, prompts) than unfamiliar interlocutors across groups and interlocutors speaking with autistic youth used more speech elicitation strategies. These findings identify important similarities and differences between autistic and non-autistic youth and interlocutor speech that provide a better understanding of language patterns in autism. Importantly, this study can increase understanding and enhance support of autistic youth by highlighting that some aspects of autistic youth's language patterns in the context of conversation may be currently underestimated.

Can AI Lie? Chabot Technologies, the Subject, and the Importance of Lying

This article poses a simple question: can AI lie? In response to this question, the article examines, as its point of inquiry, popular AI chatbots, such as, ChatGPT. In doing so, an examination of the psychoanalytic, philosophical, and technological significance of AI and its complexities are located in relation to the dynamics of truth, falsity, and deception. That is, by critically considering the chatbot’s ability to engage in natural language conversations and provide contextually relevant responses, it is argued that what separates the AI chatbot from anthropocentric debates, which allude to some form of conscious recognition on behalf of AI, is the importance of the lie – an importance which a psychoanalytic approach can reveal. Indeed, while AI technologies can undoubtedly blur the line between lies and truth-speaking, in the case of the AI chatbot, it is detailed how such technology remains unable to lie authentically or, in other words, is unable to lie like a human. For psychoanalysis, the capacity to lie bears witness to the unconscious and, thus, plays an important role in determining the subject. It is for this reason that rather than uncritically accepting the chatbot’s authority – an authority that is easily reflected in its honest responses and frank admissions – a psychoanalytic (Lacanian) perspective can highlight the significance of the unconscious as a distorting factor in determining the subject. To help elucidate this argument, specific attention is given to introducing and applying Lacan’s subject of enunciation and subject of the enunciated. This is used to assert that what continues (for now) to set us apart from AI technology is not necessarily our ‘better knowledge’ but our capability to consciously engage in acts of falsehood that function to reveal the social nuances and significances of the lie.

Age Group Differences in Social Media Language Use: Identity Representation via Emojis and Abbreviations

This paper aims to analyse how age groups impact the use of social media language (SL) in online textual conversations, focusing on the type, polarity, and subjectivity of SL to explore identity representation across different age groups. During data collection in 2023, there were 4.76 billion social media users worldwide; hence, understanding the use of SL is essential for analysing how individuals express their emotions and identities online. SL consists of emoticons/emojis, abbreviations, and mixed language in textual conversations. For this study, a qualitative approach was used, with 46 participants from three age groups: 25-34, 35-44, and 45-54 engaging in mock group conversations for one hour using WhatsApp. The participants were separated into six group conversations. The findings show that the older age groups (35-44 and 45-54) prefer to discuss technical or professional topics, whereas the younger age groups (25-34) are more interested in games, food, entertainment, and fun. Furthermore, for the age group 25-34 there is a high usage of emoticons/emojis, whereas the age group 45-54 uses them the least. These findings indicate that age groups have a significant influence on the use of SL and topic preferences in textual conversations, providing insights that will be used to develop a Natural Language Processing (NLP) tool for online identity classification. This study enhances human-computer interaction by investigating how different age groups use SL in digital environments.

The Role of Language in Remote Healthcare Triage: AMeta-Aggregative Review.

To provide an overview of evidence on the role of language in remote healthcare services prioritisation, from now on termed triage. This study synthesises literature, to better understand how language affects triage interactions, aiming to improve these processes. We conducted a meta-aggregative review. A systematic search in Scopus from inception to September 2023 identified 437 studies on language in remote healthcare triage, of which 23 were included. Information was selected using both inductive and deductive coding, focusing on six linguistic features of interaction that have been described in the literature on studies using conversation analysis: turn-taking organisation, overall structure, sequence organisation, turn design, lexical choice, and epistemological and other forms of asymmetry. The process followed the RAMESES Publication Standards for Meta-narrative Reviews. Two main findings emerged. First, all six linguistic features are present in triage conversations, indicating that language involves more than just what is said. It also matters, for example, how and when a question is asked. Second, computerised decision support systems (CDSS) significantly affect conversation flow and dynamics. Language in triage involves more than literal speech and is heavily influenced by CDSS. Our study suggests that quality assessments of triage conversations should consider not only what is said but cover all relevant aspects of language. The influence of computerised decision support systems (CDSS) on linguistic features highlights the need for systems to be adaptable, to improve conversation quality and better addressing caller needs rather than focusing on one-size-fits-all questions. This review highlights the complex role of language in triage conversations and its impact on interaction. It calls for a broader view of language in quality assessments, recognising that both call-takers and callers contribute to call quality. Insights from this review can help developers enhance question types, sequence, and delivery methods of computerised decision support systems. Finally, education for call-takers in healthcare sectors may be improved based on our findings. No patient or public contribution.

Teaching Daily Conversational Language to Toddlers at PAUD Tunas Mekar in Melatiwangi Village, Bandung

Teaching Daily Conversational Language to Toddlers at PAUD Tunas Mekar in Melatiwangi Village, Bandung

ISETECH-AI-Powered Dental Consultation Chatbot: Enhancing Public Healthcare Accessibility and Awareness

The evolving landscape of healthcare has witnessed remarkable transformations owing to the integration of Artificial Intelligence (AI) and chatbot technology. This paper presents the development of an AI-powered dental consultation chatbot aimed at revolutionizing dental healthcare accessibility and awareness, particularly in the context of Malaysia. The chatbot serves as a valuable resource for dental self-diagnosis, information dissemination, and remote consultations, reducing the need for in-person visits to dental clinics, especially considering the COVID-19 pandemic. The adoption of chatbots in healthcare has gained momentum, driven by their ability to engage in natural language conversations, disseminate reliable information, and provide personalized assistance. Chatbots have emerged as essential tools in public health management, offering services ranging from disease understanding and information provision to counselling and facilitating doctor-patient communication. In the specific domain of dental healthcare, chatbots have the potential to alleviate challenges associated with iatrophobia (fear of disclosing dental problems) and the need for professional dental consultations. The objective of this work is to develop a user-friendly, AI-powered dental consultation chatbot accessible online. This chatbot allows users to engage in real-life conversations with the machine, conducting short self-diagnosis of their oral health conditions. Users can seek information about their dental issues, understand symptoms, and gain insights into their dental health before or after visiting a dentist. Furthermore, the chatbot system presented in this study holds the potential to ease the workload of dental practitioners by offering an alternative method of consultation. Modern digital dental technologies are increasingly focused on diagnosis and classification tasks, but few have explored the realm of dental consultation chatbots for public use. This project bridges this gap and extends the utility of AI-powered chatbots to the field of dental healthcare. This AI-powered dental consultation chatbot not only addresses the evolving healthcare needs of the post-pandemic era but also contributes to advancing dental healthcare accessibility and awareness among the population.

Oscillatory features of German–English sentence processing: Evidence from an EEG study

Aims and objectives: In this study, we investigated the neural oscillatory patterns occurring during sentence-level overt bilingual processing tasks on 46 German–English bilinguals with the intention of finding out if there were task-specific oscillations for this language pair. Methodology: Four tasks, including Language 2 (L2) listening, Language 1 (L1) speaking, L2 shadowing (i.e., repeating a segment of an utterance), and backward interpreting (i.e., orally translating L2 sentences to L1), were implemented with the technique of electroencephalography (EEG). Data and analysis: Data included demographic indices, neural oscillation metrics, and oral production scores from participants. Static spectral analysis, time–frequency analysis, principal components analysis, cross-frequency coupling, and inferential statistics were applied to the data. Findings: The sensory task and articulatory tasks differed greatly in the delta–theta band and the beta–gamma band in terms of time–frequency dynamics. Analyses also revealed variations in a frontal/prefrontal theta component and a distinct gamma component across the four tasks. Moreover, we found a right fronto-temporal theta–gamma coupling unique to backward interpreting. Originality: The task-wise comparison experiment paradigm has rarely been implemented in neuroimaging research on language, especially so in EEG research on bilingualism. Significance: The findings provide novel insights for the understanding of language conversion mechanism and may benefit the training of professional interpreters.

The Effect of Age of First Exposure on Vocabulary, Mean Length of Utterance, Morphosyntactic Accuracy, and Semantic and Sentence-Level Patterns in the First 2 Years of French Second-Language Learning by Preschool- to Adolescent-Age Mandarin Speakers.

Bilingual assessment is particularly difficult in the very first period of children's second language (L2) exposure. This exploratory, longitudinal study examined L2 learning after 1 and 2 years of L2 exposure by young immigrants and how it is affected by their age at first exposure to the L2 (AoE). Participants were 18 immigrants ranging in age from 2;11 to 14;2 (years;months), all within their first year in Montreal at Time 1, enrolled in a French school or day care, and from a Mandarin first language background. Participants were tested again a year later. Measures included receptive and expressive French vocabulary tests and conversational language samples analyzed using traditional measures of mean length of utterance (MLU) and morphological accuracy as well as novel measures of semantic and sentence-level patterns. Performance was relatively high already at Time 1 and increased significantly at Time 2 in both vocabulary and MLU. At Time 2, vocabulary scores were below normative values, whereas MLU was within expected values relative to monolingual and simultaneous bilinguals for the majority of the participants. However, higher MLUs were accompanied by more instances of both semantic errors and creative semantic strategies. French performance was strongly related to AoE; with amount of exposure equivalent, older participants outperformed the younger ones on MLU and vocabulary. Semantic errors and creative uses were strongly predicted by AoE; however, morphological accuracy and number of agrammatical utterances were not. This initial period of French learning involved a rapid growth spurt for most of the participants. We argue that the pattern observed, particularly among the older children, constitutes an early stage of L2 learning characterized by long utterances that are also frequently hard to understand as speakers encounter challenges and use creative strategies in their attempt to convey meaning. Comparison with normative reference bases for monolinguals and bilinguals with greater cumulative L2 exposure who have similar MLUs should be done with much caution during this early period.