Negative Transfer Research Articles

Teaching and Learning in a First-Year Writing Skills Transfer Course: Investigating College Professor and Student Experiences

When Seneca Polytechnic replaced EAC150, an essay-based English course, with COM101, a first-semester writing course based on writing skills transfer, we saw the opportunity to investigate both professors’ and students’ experiences of the new approach. Specifically, we wanted to know how professors conceptualized and taught COM101 and also how students connected their writing for COM101 with other writing they did at Seneca, their workplaces, and in their personal lives. From 2018–2020 we conducted qualitative surveys with professors and mixed-method surveys with students and applied inductive, thematic coding to all qualitative data. The data results were encouraging: student responses indicated that COM101 positively affected their writing and reported transferring writing skills to other areas of their lives. In addition, professor responses indicated that they actively engaged with skills transfer pedagogy, despite the fact that COM101 demanded a significant change in approach. In professor responses that indicated resistance to the new approach we found valuable lessons about the core ideas of transfer, including negative transfer, and the difficulties that anyone – professors and students alike – face in new learning situations.

FIRST LANGUAGE TRANSFER IN CHINESE UNDERGRADUATES’ USE OF CONNECTORS IN ENGLISH NARRATIVES

Background and Purpose: Connectors are vital tools for cohesive discourse, significantly influencing the writing proficiency of learners. In Chinese educational contexts, many learners grapple with the correct application of connectors, due to potential first language (L1) influence. Previous research primarily delves into lexical and syntactic aspects, overlooking discourse-level considerations. This study aims to assess the frequency and accuracy of connector usage in English narratives among Chinese learners. It also seeks to investigate the extent of language transfer concerning connectors among Chinese learners and discern how this transfer evolves as learners' second language (L2) proficiency advances. Methodology: To amass data, 150 narratives composed by 30 Chinese learners were analyzed. Initial qualitative analysis was followed by quantitative examination. Findings: The findings underscore temporal words as the most frequently employed connectors, closely trailed by additives and causal words. Notably, temporal words and adversatives exhibited higher accuracy when compared to other connector types. Positive transfer manifested more frequently than negative transfer. Moreover, positive transfer displayed a noticeable inflectional change across the five-time narratives, whereas little variation was observed in negative transfer. Contributions: This study suggests that stakeholders, including educators, learners, and policymakers, should leverage positive transfer as a strategic resource when instructing connectors to enhance L2 writing proficiency. Keywords: First language, language transfer, connector, narratives, Chinese undergraduates. Cite as: Ma, Y. Q., & Stapa, M. B. (2024). First language transfer in Chinese undergraduates’ use of connectors in English narratives. Journal of Nusantara Studies, 9(2), 131-159. http://dx.doi.org/10.24200/jonus.vol9iss2pp131-159

The Effect of L1 and L2 on L3 in Grammar Learning from the Perspective of Language Transfer ---University Students with Chinese as L1, English as L2, learning L3 Spanish Subjunctive Mood

Third language acquisition (TLA) research, as a novel area of study both domestically and internationally, has garnered significant attention from scholars worldwide in recent years. Previous studies on third language acquisition have extensively focused on interlingual transfer, but there has been limited attention given to grammatical transfer, especially related to the subjunctive mood. Additionally, Spanish has received less discussion than other languages. This present study aimed to investigate the language transfer of Chinese and English on learning Spanish subjunctive mood in a Spanish writing test conducted by 20 Chinese university students. The data were analyzed using frequency analysis and descriptive analysis in SPSS. The study revealed the following major findings: (1) Chinese transfer affects Spanish through changing verb form and the choice of tenses. (2) English transfer affects Spanish through the wrong verb form in the adverbial clause of purpose. Thus, the negative transfer from L2 to L3 is stronger than the one from L1 to L3.

The Negative Transfer of Shanghai Dialect on English Vowel Phonology Acquisition of Shanghai High School Students

With the accerleration of globalisation, the importance of English as an international common language has become more and more prominent. This study investigated the negative transfer effect of Shanghai dialect on English vowel acquisition in Shanghai high school students. Through the analysis of formant data of English vowel pronunciation in 49 Shanghai high school students, it is found that there is no significant change in tongue position when students pronounce some long and short vowels, the overall tongue position is low, and the distinction between long and short vowels is not obvious. The results show that the specific vowel phonemes in Shanghai dialect have a negative transfer effect on English pronunciation, which affects the accuracy of students' English pronunciation. Based on this, this study puts forward a variety of methods to improve English pronunciation teaching, including strengthening tongue position and oral shape training, using multimedia technology to assist teaching and providing personalized tutoring to help students overcome pronunciation barriers caused by dialects.

ANALISIS KESALAHAN FONOLOGI DALAM KETERAMPILAN BERBICARA BAHASA ARAB DI PONDOK PESANTREN PUTRI DAARUL FIKRI MALANG

This research focuses on the analysis of phonological errors in Arabic speaking abilities at the Daarul Fikri Malang Islamic Boarding School for Girls. The research method applied is qualitative descriptive, meaning that the search for facts is conducted through the interpretation of the data found, with the approach being a language error analysis approach to identify and elaborate on the phonological errors that arise in the mastery of speaking Arabic at the Daarul Fikri Malang Islamic Boarding School for Girls. In this research, the researcher used data collection techniques such as observation, documentation, and interviews. The analysis results indicate that 1.) Some types of phonological errors such as a. letter substitution b. letter addition c. letter deletion. 2.) The causes of phonological errors are a. language interference b. differences in phonological systems c. differences in educational backgrounds d. lack of practice in speaking Arabic. 3.) Proposed solutions include a. ishlahul lughoh for language correction b. and the provision of special intensive classes for learning ashwat using phonetic methods and mim-mem methods to improve speaking skills.

A Research Review on the Impact of English Language Transfer on the Spanish Learning of Chinese Students

In the diverse context of multilingual learning, language transfer has garnered significant attention. Spanish, the third most widely spoken language globally, after Chinese and English, is being increasingly explored by learners from various cultural and linguistic backgrounds. The stark differences between Chinese and Spanish create an inevitable influence on the learning process of Chinese students. Their existing knowledge of English as a second language plays a crucial role in shaping their learning of Spanish. As a result, English language transfer has become a crucial topic in the study of Chinese students' learning of Spanish. The purpose of this paper is to analyze previous studies and explore the effects of English language transfer on Spanish grammar, vocabulary and pronunciation Exploring the impact of English language transfer on Chinese students' learning of Spanish can offer valuable insights into the complexities of the language learning process. Moreover, it can provide essential theoretical guidance and practical recommendations for students' Spanish learning. Ultimately, the exploration results can drive the development and advancement of Spanish teaching practices, benefiting both students and educators alike.

A Review of Research on Mother Tongue Transfer During English as a Second Language Acquisition

In the process of second language acquisition, learners are often influenced by native language transfer. The effect can be both positive and negative. That is, the impact of positive and negative migrations. This paper is aimed at learners who speak English as a second language. This paper mainly summarizes the influence of mother tongue transfer on grammar, pronunciation, and vocabulary. Mother tongue transfer may be challenging for English as a second language learners in terms of grammatical morphological changes, word order, virtual word functions, and syntactic component correspondence. The sentence structure in English is more complex than in Chinese. This makes it easy for learners to have difficulties in mastering English tenses, voice, singular and plural, etc. The second is the impact of native language transfer on English pronunciation. Native Chinese speakers are influenced by Chinese tones when learning English pronunciation. This results in a lack of fluency and accuracy in pronunciation. Finally, there is the impact of mother tongue transfer on English vocabulary. Since each language will have similarities, it indirectly enriches the vocabulary of the second language, and it is faster for us to learn and memorize. Learners should be aware to avoid the adverse effects of negative transfer of mother tongue.

Phonetic Transfer of Teochew Dialect to English Pronunciation Acquisition

The transfer of phonological features in the first language to second language pronunciation during the process of second language acquisition has long been attested. A great majority of studies analyze the negative transfer from Chinese to English, yet few focus on Chinese dialects transfer to English. This paper aims to summarize the previous studies on Teochew phonological transfer to English acquisition, including consonant transfer, vowel transfer, and tonal transfer. Under the framework of Language Transfer Theory and Contrastive Analysis Hypotheses, research gaps and possible research directions for future studies will be put forward. This paper shows that phonetic transfer occurs in Teochew-speaking groups learning English, influenced by the dialects consonants, vowels and likely tones. Further research should explore transfer causes across different age groups and test classroom phonetic correction methods for Teochew-speaking English learners.

Customer service chatbot enhancement with attention-based transfer learning

Customer service is an important and expensive aspect of business, often being the largest department in most companies. With growing societal acceptance and increasing cost efficiency due to mass production, service robots are beginning to cross from the industrial domain to the social domain. Currently, customer service robots tend to be digital and emulate social interactions through on-screen text, but state-of-the-art research points towards physical robots soon providing customer service in person. This article explores the feasibility of Transfer Learning different customer service domains to improve chatbot models. In our proposed approach, transfer learning-based chatbot models are initially assigned to learn one domain from an initial random weight distribution. Each model is then tasked with learning another domain by transferring knowledge from the previous domains. To evaluate our approach, a range of 19 companies from domains such as e-Commerce, telecommunications, and technology are selected through social interaction with X (formerly Twitter) customer support accounts. The results show that the majority of models are improved when transferring knowledge from at least one other domain, particularly those more data-scarce than others. General language transfer learning is observed, as well as higher-level transfer of similar domain knowledge. For each of the 19 domains, the Wilcoxon signed-rank test suggests that 16 have statistically significant distributions between transfer and non-transfer learning. Finally, feasibility is explored for the deployment of chatbot models to physical robot platforms including “Pepper”, a semi-humanoid robot manufactured by SoftBank Robotics, and “Temi”, a personal assistant robot.

Assessing language models’ task and language transfer capabilities for sentiment analysis in dialog data

Our work explores the differences between GRU-based and transformer-based approaches in the context of sentiment analysis on text dialog. In addition to the overall performance on the downstream task, we assess the knowledge transfer capabilities of the models by applying a thorough zero-shot analysis at task level, and on the cross-lingual performance between five European languages. The ability to generalize over different tasks and languages is of high importance, as the data needed for a particular application may be scarce or non existent. We perform evaluations on both known benchmark datasets and a novel synthetic dataset for dialog data, containing Romanian call-center conversations. We study the most appropriate combination of synthetic and real data for fine-tuning on the downstream task, enabling our models to perform in low-resource environments. We leverage the informative power of the conversational context, showing that appending the previous four utterances of the same speaker to the input sequence has the greatest benefit on the inference performance. The cross-lingual and cross-task evaluations have shown that the transformer-based models possess superior transfer abilities to the GRU model, especially in the zero-shot setting. Considering its prior intensive fine-tuning on multiple labeled datasets for various tasks, FLAN-T5 excels in the zero-shot task experiments, obtaining a zero-shot accuracy of 51.27% on the IEMOCAP dataset, alongside the classical BERT that obtained the highest zero-shot accuracy on the MELD dataset with 55.08%.

Enhancing Hydrological Variable Prediction through Multitask LSTM Models

Deep learning models possess the capacity to accurately forecast various hydrological variables, encompassing flow, temperature, and runoff, notably leveraging Long Short-Term Memory (LSTM) networks to exhibit exceptional performance in capturing long-term dynamics. Nonetheless, these deep learning models often fixate solely on singular predictive tasks, thus overlooking the interdependencies among variables within the hydrological cycle. To address this gap, our study introduces a model that amalgamates Multitask Learning (MTL) and LSTM, harnessing inter-variable information to achieve high-precision forecasting across multiple tasks. We evaluate our proposed model on the global ERA5-Land dataset and juxtapose the results against those of a single-task model predicting a sole variable. Furthermore, experiments explore the impact of task weight allocation on the performance of multitask learning. The results indicate that when there is positive transfer among variables, multitask learning aids in enhancing predictive performance. When jointly forecasting first-layer soil moisture (SM1) and evapotranspiration (ET), the Nash–Sutcliffe Efficiency (NSE) increases by 19.6% and 4.1%, respectively, compared to the single-task baseline model; Kling–Gupta Efficiency (KGE) improves by 8.4% and 6.1%. Additionally, the model exhibits greater forecast stability when confronted with extreme data variations in tropical monsoon regions (AM). In conclusion, our study substantiates the applicability of multitask learning in the realm of hydrological variable prediction.

Étude des interférences intralinguistiques en FLE: l'impact des homographes et des pronoms sur l'apprentissage des compléments et articles définis à Osun State University-Nigeria

Study of Intralinguistic Interference in FLE: The Impact of Homographs and Pronouns on the Learning of Complements and Definite Articles at Osun State University-Nigeria Linguistic interference is a psycholinguistic phenomenon which slows down the learning process of the target language due to the existing similarities between it (L2) and the one already acquired (L1). Many research works had been carried out on linguistic transfer and interference. However, among those works, only a significant few have underscored the intra-linguistic interference as a trap associated with understanding the homographs, personal pronoun complements and definite articles in the class of French as a foreign language FLE. Therefore, this paper evaluates the grammatical implications that the distinctions between le, la, la les as personal pronoun complements and definite articles in Allah n’est pas obligé by Ahmadou Kourouma have on Anglophone French learners. The theoretical framework adopted to foreground our claims is linguistic transfer and interference. Percentage (%) is the scale used to interpret their performance while data are analysed based on textual analysis. The text is purposively selected to situate our analysis in the context of a postcolonial Francophone African literary text whose learning is always complicated among Anglophone and Yorubaphone French learners. Five questions were asked to evaluate the language skills of the seventy-five (75) third-year students at Osun State University across twenty corpora drawn from the work. The morpho-phonetic similarities between the definite articles and the personal pronoun complement inhibit the comprehension of instructional materials. In conclusion, contextual interpretation and more rigorous study of the grammatical subject matters are suggested to both teachers and learners, respectively, to eliminate the confusion associated with the teaching and learning of the homographs. To significantly improve academic performance among non-French-speaking learners, authorities should implement the pragmatic interventions outlined in this article, involving teaching professionals and learners alike.

Weighted unsupervised domain adaptation considering geometry features and engineering performance of 3D design data

The product design process in manufacturing involves iterative design modeling and analysis to achieve the target engineering performance, but such an iterative process is time consuming and computationally expensive. Recently, deep learning-based engineering performance prediction models have been proposed to accelerate design optimization. However, they only guarantee predictions on training data and may be inaccurate when applied to new domain data. In particular, 3D design data have complex features, which means domains with various distributions exist. Thus, the utilization of deep learning has limitations due to the heavy data collection and training burdens. We propose a bi-weighted unsupervised domain adaptation approach that considers the geometry features and engineering performance of 3D design data. The proposed method is the first to apply domain adaptation to deep learning-based engineering performance prediction to overcome the problem of insufficient industrial data. In addition, we present a source instance weighting method called the bi-weighting strategy to consider the geometry features and engineering performance of 3D design data and avoid negative transfer for accurate prediction. Domain-invariant features can be extracted through an adversarial training strategy by using hypothesis discrepancy, and a multi-output regression task can be performed with the extracted features to predict the engineering performance. The proposed model is tested on a wheel impact analysis problem to predict the magnitude of the maximum von Mises stress and the corresponding location of 3D road wheels. This mechanism can reduce the target risk for unlabeled target domains on the basis of weighted multi-source domain knowledge and can efficiently replace conventional finite element analysis.

Phacoemulsification to MSICS: A transfer of skills study in a simulated environment.: Transfer of phacoemulsification surgical skills to MSICS.

To investigate if phacoemulsification experience impacts and transfers to the skill acquisition of novices in manual small incision cataract surgery (MSICS) within a simulation environment. Copenhagen Academy for Medical Education and Simulation, Denmark. Prospective controlled experimental study. The study included 29 residents or specialist in ophthalmology with no prior MSICS experience. Participants were required to either have 1) proficiency on the EyeSi surgical simulator (PG; phaco group) or 2) no prior extensive exposure to the EyeSi simulator or any phacoemulsification surgery experience as a primary surgeon (CG; control group). Possible skill transfer was assessed using a test on the HelpMeSee virtual-reality simulator, including nine modules and 30 steps (points). A pass/fail score was determined at 20 points (out of 30 points). Performance scores were analyzed using independent samples t-tests, pass rates using Fisher's exact test, and individual modules using Fisher-Freeman-Halton analysis. The PG performed significantly better in terms of overall performance score (mean SD, PG; 21.8 points 2.3 versus CG; 18.9 points 2.2, p = 0.002, t-value = -3.39) and pass rates (PG; 72 % versus CG; 22%, p = 0.018). The sub-analysis for individual modules demonstrated significant differences exclusively in the capsulorhexis and cortex removal modules. The results suggest a positive inter-procedural transfer of skills between phacoemulsification and MSICS. Pretraining and experience in phacoemulsification demonstrated a significant effect on the performance of MSICS within a simulation environment. These findings suggest that phacoemulsification experience provides an advantage before MSICS training in surgical residency programs.

A Review of the Influence of Wu Dialect on English Pronunciation Acquisition

The Wu dialect is widely spoken in Wu-speaking regions of China, which covers a considerable number of English learners. However, the phonetic system of the Wu dialect differs significantly from that of English, which greatly affects the English learning process of Wu-speaking learners, thus attracting numerous researchers to conduct related studies. This paper provides a comprehensive overview of research on the influence of Wu dialect on English phonological learning over the past two decades and presents a selection of twenty-seven papers to explore the theoretical framework, the research methodology, and the specific impacts encountered by English language learners who speak Wu dialect. Among them, the theoretical discussion centered on language transfer theory, Fleges Speech Learning Model and Bests Perceptual Assimilation Model; the research methodology consisted of qualitative analysis and empirical study; and the focus was placed on positive and negative transfer effects, i.e., the challenges posed by the absence or presence of phonological differences in the Wu dialect to the learners of English. The findings reveal significant segmental and suprasegmental influences in English pronunciation acquisition, with segmental challenges including differentiation of consonants and incomplete articulation of vowels, and suprasegmental challenges including stress placement, intonation patterns, and phonological rhythm. In addition, the discussion highlights the need for more empirical studies to supplement qualitative analyses, the consideration of regional variations within Wu dialect, and the exploration of teaching strategies informed by research findings. Therefore, the paper advocates for future research focusing on regional differences, deeper exploration of suprasegmental influences, integration of research into teaching practices, and diversified research methods.

Adapting differential molecular representation with hierarchical prompts for multi-label property prediction.

Accurate prediction of molecular properties is crucial in drug discovery. Traditional methods often overlook that real-world molecules typically exhibit multiple property labels with complex correlations. To this end, we propose a novel framework, HiPM, which stands for Hierarchical Prompted Molecular representation learning framework. HiPM leverages task-aware prompts to enhance the differential expression of tasks in molecular representations and mitigate negative transfer caused by conflicts in individual task information. Our framework comprises two core components: the Molecular Representation Encoder (MRE) and the Task-Aware Prompter (TAP). MRE employs a hierarchical message-passing network architecture to capture molecular features at both the atom and motif levels. Meanwhile, TAP utilizes agglomerative hierarchical clustering algorithm to construct a prompt tree that reflects task affinity and distinctiveness, enabling the model to consider multi-granular correlation information among tasks, thereby effectively handling the complexity of multi-label property prediction. Extensive experiments demonstrate that HiPM achieves state-of-the-art performance across various multi-label datasets, offering a novel perspective on multi-label molecular representation learning.

A Fundamental Correlative Spectroscopic Study on Li1‐xNiO2 and NaNiO2

AbstractThe intricate relationship between local atomic arrangements and electronic states significantly influences the electrochemical properties of Li‐ion battery cathode materials. Despite decades of investigation, a consensus regarding the local atomic and electronic structure of LiNiO2 remains elusive. This ambiguity stems from the potential distortion of Ni sites, either via Jahn‐Teller (JT) distortion or bond disproportionation (BD), complicating the understanding of the charge compensation mechanism involving Ni and O. This study compares the structures of LiNiO2 and NaNiO2, a JT system, using an innovative approach that integrates bulk spectroscopy techniques on standardized interoperable samples for enhanced reliability. While X‐r and theoretical calculations fail to differentiate between the proposed scenarios, Raman spectroscopy highlights local structural distinctions between monoclinic NaNiO2 and rhombohedral LiNiO2. HAXPES confirms various formal oxidation states for Ni, supported by RIXS data indicating 3d8 states, emphasizing negative charge transfer from Ni and some bond disproportionation in LiNiO2. Regarding charge compensation, XRS and RIXS suggest oxygen hole involvement in redox activity, whereas Raman spectroscopy does not detect molecular oxygen. This comprehensive spectroscopic analysis highlights the importance of correlative characterization workflows in elucidating complex structural‐electrochemical relationships.

Polyester Microfibers Exposure Modulates Mytilus galloprovincialis Hemolymph Microbiome.

Microplastic (MP) contamination in the aquatic environment is a cause of concern worldwide since MP can be taken up by different organisms, altering different biological functions. In particular, evidence is accumulating that MP can affect the relationship between the host and its associated microbial communities (the microbiome), with potentially negative health consequences. Synthetic microfibers (MFs) represent one of the main MPs in the marine environment, which can be accumulated by filter-feeding invertebrates, such as bivalves, with consequent negative effects and transfer through the food chain. In the mussel Mytilus galloprovincialis, polyethylene terephthalate (PET) MFs, with a size distribution resembling that of an MF released from textile washing, have been previously shown to induce multiple stress responses. In this work, in the same experimental conditions, the effects of exposure to PET-MF (96 h, 10, and 100 μg/L) on mussel hemolymph microbiome were evaluated by 16S rRNA gene amplification and sequencing. The results show that PET-MF affects the composition of bacterial communities at the phylum, family and genus level, with stronger effects at the lowest concentration tested. The relationship between MF-induced changes in hemolymph microbial communities and responses observed at the whole organism level are discussed.

Pronoun interpretation in intermediate-advanced L2 English speakers: L2 to L1 cross-linguistic effects

Aims and objectives: L2 learners who speak a null subject (L1 Spanish) and a non-null subject language (L2 English) may experience cross-linguistic interference from the L2 on L1 pronoun interpretation. In this study, we test pronoun interpretation in the L1 and L2 of adult learners, in comparison with two groups of monolingual speakers, to assess if L1 pronoun interpretation can change as a result of L2 acquisition at intermediate-advanced levels of L2 proficiency and in the absence of long L2 immersion. Methodology: A group of L2 English speakers (L1 Spanish) participated in two offline sentence comprehension tasks where they interpreted pronouns in the L1 and L2. The results are compared with English and Spanish monolingual speakers. Data and analysis: We find that adult English learners comprehend pronouns in their L1 (Spanish) differently than Spanish monolingual speakers, demonstrating a strong subject bias for interpreting null and overt pronouns. In addition, we show that pronoun interpretation patterns acquired in the L2 explain the changes to L1 interpretation biases. Conclusions: The results of this study significantly advance the understanding of the factors that contribute to bilingual language comprehension, showing the permeability of the L1 comprehension system at the discourse level as soon as the L2 sets in.

Enhancing molecular property prediction with auxiliary learning and task-specific adaptation

Pretrained Graph Neural Networks have been widely adopted for various molecular property prediction tasks. Despite their ability to encode structural and relational features of molecules, traditional fine-tuning of such pretrained GNNs on the target task can lead to poor generalization. To address this, we explore the adaptation of pretrained GNNs to the target task by jointly training them with multiple auxiliary tasks. This could enable the GNNs to learn both general and task-specific features, which may benefit the target task. However, a major challenge is to determine the relatedness of auxiliary tasks with the target task. To address this, we investigate multiple strategies to measure the relevance of auxiliary tasks and integrate such tasks by adaptively combining task gradients or by learning task weights via bi-level optimization. Additionally, we propose a novel gradient surgery-based approach, Rotation of Conflicting Gradients (RCGrad\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\mathop {\ exttt{RCGrad}}\\limits$$\\end{document}), that learns to align conflicting auxiliary task gradients through rotation. Our experiments with state-of-the-art pretrained GNNs demonstrate the efficacy of our proposed methods, with improvements of up to 7.7% over fine-tuning. This suggests that incorporating auxiliary tasks along with target task fine-tuning can be an effective way to improve the generalizability of pretrained GNNs for molecular property prediction.Scientific contributionWe introduce a novel framework for adapting pretrained GNNs to molecular tasks using auxiliary learning to address the critical issue of negative transfer. Leveraging novel gradient surgery techniques such as RCGrad\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\mathop {\ exttt{RCGrad}}\\limits$$\\end{document}, the proposed adaptation framework represents a significant departure from the dominant pretraining fine-tuning approach for molecular GNNs. Our contributions are significant for drug discovery research, especially for tasks with limited data, filling a notable gap in the efficient adaptation of pretrained models for molecular GNNs.