Small-scale Corpus Research Articles

Research on the Training and Application Methods of a Lightweight Agricultural Domain-Specific Large Language Model Supporting Mandarin Chinese and Uyghur

In the field of Natural Language Processing (NLP), the lack of support for minority languages, especially Uyghur, the scarcity of Uyghur language corpora in the agricultural domain, and the lightweight nature of large language models remain prominent issues. This study proposes a method for constructing a bilingual (Uyghur and Chinese) lightweight specialized large language model for the agricultural domain. By utilizing a mixed training approach of Uyghur and Chinese, we extracted Chinese corpus text from agricultural-themed books in PDF format using OCR (Optical Character Recognition) technology, converted the Chinese text corpus into a Uyghur corpus using a rapid translation API, and constructed a bilingual mixed vocabulary. We applied the parameterized Transformer model algorithm to train the model for the agricultural domain in both Chinese and Uyghur. Furthermore, we introduced a context detection and fail-safe mechanism for the generated text. The constructed model possesses the ability to support bilingual reasoning in Uyghur and Chinese in the agricultural domain, with higher accuracy and a smaller size that requires less hardware. It (our work) addresses issues such as the scarcity of Uyghur corpora in the agricultural domain, mixed word segmentation and word vector modeling in Uyghur for widespread agricultural languages, model lightweighting and deployment, and the fragmentation of non-relevant texts during knowledge extraction from small-scale corpora. The lightweight design of the model reduces hardware requirements, facilitating deployment in resource-constrained environments. This advancement promotes agricultural intelligence, aids in the development of specific applications and minority languages (such as agriculture and Uyghur), and contributes to rural revitalization.

A case study of comparative metaphor analysis in Finnish and Hungarian news texts

Abstract Despite remarkable previous results of cognitive linguistic analysis of metaphorization in Finnish and Hungarian even in a cross-linguistic context (Huumo 2019; Máthé 2022), a systematic comparative study of the patterns of linguistic metaphors in these two languages is still to be carried out. The aim of this paper is to fill this research gap by proposing a MIPVU-inspired protocol for metaphor identification in languages with rich morphology and presenting the preliminary results of a small-scale corpus analysis. The study is organized around the following questions: (i) How can we adapt and extend the original MIPVU method for agglutinative Uralic languages like Hungarian and Finnish? (ii) How can we implement the adapted method in a collaborative parallel annotation process of sampled news texts? (iii) What are the differences in metaphorical patterns between Finnish and Hungarian? The paper details the adapted protocol of MIPVU for the Hungarian language (Simon et al. 2023), and how it can be applied to Finnish, demonstrating its applicability in a research corpus. Our findings demonstrate a relatively similar metaphorization in Hungarian and Finnish with slight differences in terms of the frequency of metaphorical expressions, their semantic relations, and the complexity of argument structure constructions.

Using interpersonal meaning making resources to build relationships and improve engagement in online teacher professional learning

AbstractAs professional learning and development (PLD) for teachers moves online, it is increasingly important to consider what constitutes effective provision. While models of effective PLD abound, online PLD faces challenges to participant engagement. In particular, the critical need to build and maintain relationships in professional learning is complicated by the geographic and temporal distribution of online participants as well as the nuances of the technology in use. We argue that online PLD occurs in collaborative text-spaces within learning management systems, social media spaces and their attendant learning objects such as forums. As such, persistent challenges to engagement in online PLD may be met by considering the language-based interpersonal affordances of these text-spaces. We employ a small-scale corpus study and appraisal analysis to demonstrate the value of Systemic Functional Linguistics in addressing the challenges of online learning, particularly how the meaning making resources of the interpersonal metafunction can be used to improve participant engagement.

Exploring the potential of data augmentation in poetry generation with small-scale corpora

Poetry generation is a complex task in the field of natural language processing, especially when working with small datasets. Data augmentation techniques have been shown to be an effective way to improve the performance of deep learning models in various tasks, including image classification and speech recognition. Therefore, this study focuses on exploring the impact of four different data augmentation methods - Synonym Replacement, Random Insertion, Random Swap, and Random Deletion - on the performance of poetry generation with a small poetry dataset. The results of the study reveal that Random Insertion performed well in terms of Bilingual Evaluation Understudy (BLEU), Recall-Oriented Understudy for Gisting Evaluation (ROUGE), and manual evaluation when compared to other data augmentation techniques. Synonym Replacement performed poorly in all three evaluations. This study confirms the potential value of data augmentation technology in poetry generation tasks and provides innovative perspectives and directions for future research in this area. Data augmentation can be employed to help address the problem of limited data in poetry generation tasks and enhance the efficiency of deep learning models. Future research could focus on exploring more advanced data augmentation techniques and their impact on poetry generation tasks.

An Analysis of the Translation Output and Value Dissemination of ChatGPT

ChatGPT, as a generative artificial intelligence system, possesses the ability to engage in interactive conversations with the users. This research focuses on evaluating the translation capabilities of ChatGPT3.5 by utilizing a small-scale corpus with bilingual news reports, renowned poetry translations, and government legal law documents. The translation by ChatGPT 3.5 are compared against those generated by Google Translate and DeepL Translate using the BLEU score as an evaluation metric. The study employs the Political Compass as a theoretical framework to investigate the potential ideological stance and values of ChatGPT by statements. The research reveals that ChatGPT performs greatly in Chinese-to-English translation, particularly demonstrating proficiency in technical, economic, and medical texts, compared to influential machine translation software. It should also be noted that the texts generated by ChatGPT is not completely rational and objective, as it exhibits inclinations towards left-wing economic ideologies. This highlights the role of ChatGPT in disseminating values within the realm of artificial intelligence and deepens our understanding of generative AI systems. Moreover, possibilities for exploring opportunities and providing guidance for shaping AI values open up.

How much Romanian does Google Translate know?

To compensate, even if on a small scale, for the scarcity of investigations of English-into-Romanian machine translations from a corpus-based genre-specific perspective, this case study concerns the quality of such translation outcomes in the particular case of two different, but closely related, text genres – everyday and newspaper/ news releases language, that have been translated with the help of Google Translate. The paper starts from Keshavarz’s (1999) very general model of error analysis to create a linguistic error profile in the two genres taken into consideration. The errors identified and discussed are illustrated with small-scale corpus examples. Since they reflect the capabilities of the Google translation platform, the findings of this paper may be relevant for its developers to get a clearer picture of its strengths and limitations and suggest ways of improving it so that it can ultimately provide higher quality translations. It may also contribute to raising translators’ attention to which areas are potentially problematic when they work, in the post-editing stage, with the English-Romanian language pair, with the particular text genres considered here.

A Corpus-based Study on Translation Norms of Prose: Taking Zhu Ziqing's The View of My Father's Back as an Example

This paper builds a small-scale corpus consisting of Zhu Ziqing's original work The View of My Father's Back and its English translations by Zhang Peiji, Yang Xianyi and Gladys Yang. Through the corpus, the lexical and syntactic features of prose are studied to explore the translation norms of prose. The quantitative and qualitative analysis of Zhang's and Yangs' translations shows a tendency towards conventionalized expectations at the levels of words and sentences. This study hopes to provide translators with helpful insights on translating Chinese prose and promote overseas dissemination of Chinese prose works and cultural communication.

Productivity from a Metapragmatic Perspective: Measuring the Diachronic Coverage of the Low Level Lexico-Grammatical Construction Have the N (Body Part/Attitude) to ↔<Metapragmatic Comment> Using the COHA

This paper seeks to address the relation between semantics, pragmatics and the productivity of a low level lexico-grammatical construction, Have the N (body part/attitude) to ↔metapragmatic comment. The question posed is how semantics affects productivity, in the generative sense of extensibility of a construction (a form meaning pairing). The method identifies the specificity and variations of the Have the N (body part/attitude) to ↔metapragmatic comment construction within the pragmeme of politeness using the COHA. Hereafter, we consider how to measure the extensibility within the onomasiological frame based on the available pool of forms expressing an attitude/emotion, i.e., the coverage or attractivity of the Have the N to construction. The paper discusses the findings, namely, how to overcome methodological issues relating to a qualitative rather than quantitative approach to the constructional architecture and the relative productivity of constructions. The experimental small scale corpus study of Have the N to in the COHA suggests that a global view of constructional architecture at multiple levels should be pertinent to identifying the extensibility potential of the construction.

Chains-BERT: A High-Performance Semi-Supervised and Contrastive Learning-Based Automatic Question-and-Answering Model for Agricultural Scenarios

Achieving automatic question-and-answering for agricultural scenarios based on machine reading comprehension can facilitate production staff to query information and process data efficiently. Nevertheless, when studying agricultural question-and-answer classification, there are barriers, such as small-scale corpus, narrow content range of corpus, or the need for manual annotation. In the context of such production needs, this paper proposed a text classification model based on text-relational chains and applied it to machine reading comprehension and open-ended question-and-answer tasks in agricultural scenarios. This paper modified the BERT network based on semi-supervised and contrastive learning to enhance the model’s performance. By incorporating the text-relational chains with the BERT network, the Chains-BERT model is constructed. Our efficient mode method outperformed other methods on the CAIL2018 dataset. Ultimately, we developed an automatic question-and-answering application to embed the contrastive-learning information aggregation model in this paper. The accuracy of the proposed model exceeded that of several contrasting mainstream models in many open-source datasets. In agricultural scenarios, the model has achieved state-of-the-art levels and is the best in efficiency.

A Multigranularity Text Driven Named Entity Recognition CGAN Model for Traditional Chinese Medicine Literatures.

Recognition of Traditional Chinese Medicine (TCM) entities from different types of literature is challenging research, which is the foundation for extracting a large amount of TCM knowledge existing in unstructured texts into structured formats. The lack of large-scale annotated data makes unsatisfactory application of conventional deep learning models in TCM text knowledge extraction. Some other unsupervised methods rely on other auxiliary data, such as domain dictionaries. We propose a multigranularity text-driven NER model based on Conditional Generation Adversarial Network (MT-CGAN) to implement TCM NER with small-scale annotated corpus. In the model, a multigranularity text features encoder (MTFE) is designed to extract rich semantic and grammatical information from multiple dimensions of TCM texts. By differentiating the conditional constraints of the generator and discriminator of MT-CGAN, the synchronization between the generated tag labs and the named entities is guaranteed. Furthermore, seeds of different TCM text types are introduced into our model to improve the precision of NER. We compare our method with other baseline methods to illustrate the effectiveness of our method on 4 kinds of gold-standard datasets. The experiment results show that the standard precision, recall, and F1 score of our method are higher than the state-of-the-art methods by 0.24∼8.97%, 0.89∼12.74%, and 0.01∼10.84%. MT-CGAN is able to extract entities from different types of TCM literature effectively. Our experimental results indicate that the proposed approach has a clear advantage in processing TCM texts with more entity types, higher sparsity, less regular features, and a small-scale corpus.

A hybrid method based on semi-supervised learning for relation extraction in Chinese EMRs

BackgroundBuilding a large-scale medical knowledge graphs needs to automatically extract the relations between entities from electronic medical records (EMRs) . The main challenges are the scarcity of available labeled corpus and the identification of complexity semantic relations in text of Chinese EMRs. A hybrid method based on semi-supervised learning is proposed to extract the medical entity relations from small-scale complex Chinese EMRs.MethodsThe semantic features of sentences are extracted by a residual network and the long dependent information is captured by bidirectional gated recurrent unit. Then the attention mechanism is used to assign weights for the extracted features respectively, and the output of two attention mechanisms is integrated for relation prediction. We adjusted the training process with manually annotated small-scale relational corpus and bootstrapping semi-supervised learning algorithm, and continuously expanded the datasets during the training process.ResultsWe constructed a small corpus of Chinese EMRs relation extraction based on the EMR datasets released at the China Conference on Knowledge Graph and Semantic Computing. The experimental results show that the best F1-score of the proposed method on the overall relation categories reaches 89.78%, which is 13.07% higher than the baseline CNN.

Personification in EFL learners’ academic writing: A cognitive linguistic stance

The paper offers a report of a small-scale corpus investigation into some advanced EFL learners’ use of personification in academic writing within a higher education context (Mouloud Mammeri University). Its main objective is to shed light on the extent of the occurrence of this phenomenon in their writings. The question raised is whether the examination of their dissertations would reveal extensive use of personification. Conceptual Metaphor Theory constitutes the theoretical framework featuring in this research. Relying on the Metaphor Identification Procedure (MIP) (Pragglejaz Group 2007), six master’s dissertations are selected for examination, and a quantitative analysis of the identified metaphorically used words is conducted. The results of the study reveal an overwhelming manifestation of personification. These findings may constitute a small contribution to the field of education, as offering useful data to educational practitioners and researchers.

THE IMPACT OF CULTURE AND CULTURAL DIMENSIONS ON STUDENTSʼ E-MAIL CORRESPONDENCE – A PILOT STUDY

E-mail correspondence between teachers and students is common. Online communication provides students with possibilities to write to teachers directly, using a diverse level of knowledge, level of ignorance and personal beliefs. The aim of this paper is to answer whether culture and cultural dimensions (defined by Hofstede’s high and low Power Distance dimension) influence the professional correspondence between teachers and students. The small-scale corpus consists of 100 e-mails, 50 written by Slovene students in English or Slovene, and 50 by Serbian students in Serbian or English. The research investigates the choice of e-mail template, the choice of language (native tongue or language of instructions), and the norms related to politeness and power distance, with the focus on salutations, formality, polite expressions, and directness. Usage of lexical modifiers, such as downtoners, upstaters and hedges will also be investigated. The results will demonstrate that e-mails by Slovene students follow new cultural standards and have become more indirect and informal, while Serbian students write e-mails with formal salutations and direct requests following the inherited hierarchy and still unmodified cultural dimensions.

Snack Names In China

Previous studies of snack names have focused on their psychological impact on consumers in different cultures but have tended to ignore their onomastic features. This study helps to address this gap based on a corpus of 121 snack names extracted from the book Chinese Famous Local Delicious Food and Special Products. This study explores the patterns of syllables, sounds, and name types of snack names compiled in this small-scale corpus. In this investigation, it was found that descriptive names were the dominant type in the corpus and the most frequently described type feature was the food ingredient. Interestingly, metaphorical names in the corpus were in general found to be related to shape similarities. Contrary to previous findings on dish names and drinking brand names in China, the snack names examined in this corpus showed a preference for three-syllable patterns and “light” or “flat” tones. After discussing these and other findings of this research, this paper discusses what insights this study may provide for other name investigations that utilize corpus linguistic approaches.

Corpus Routes and Experiments in Language Teaching

Corpus quantitative approach in teaching, which is of growing interest, entails some revision of the L2 vocabulary selection procedures and provides solutions for a wide range of practical problems. The focus throughout is on the discussion of research on the quantitative and qualitative characteristics of language which both teaching content and language acquisition practice could draw on. This research regards human language as a rank distribution, which has serious implications on quantitative aspect of learner’s vocabulary. We also looked into the ways to factor in the data from a small professional discourse corpus in order to target the units that have the greatest statistical prominence. Both BNC and our own collection of texts are explored. The study also elaborates on academic writing cohesive devices and grammar patterns introduction, which is approached through concordancing corpus strings to (i) provide frequency evidence and (ii) introduce a contrast in usage in various corpus genres. Striking differences that are evidenced by the frequency lists could be related not only to register, but also to the choice and instances of academic cohesive clusters which are favoured by the apprentice writers and the expert writers. With the aim of capitalizing on corpus approach a number of small-scale corpus research tasks were developed. This study also uses corpus tools and data to give a seemingly subjective phenomenon of hedging some quantitative measurement. While experimenting on corpus in the classroom, the attention of learners was drawn to various means of hedging, such as lexical bundles or down-toners that manifest themselves as important communicative strategies. Thus, corpus was used to inform both the language instructor and the student in the classroom to look in detail at differences in the use of lexical and grammatical units in different varieties of language, address contrasting register variations, and readily provide contemporary professionally relevant examples of actual language usage. It has to be noted that university students have a tendency not to perceive register violations as language errors on a par with those of grammar, lexis or punctuation. Hence, corpus investigation as raising awareness tool also proved to be an effective teaching material generator. Nowadays syllabi have the opportunity to be rather sensitive to the quantitative evidence that corpora offer us; what is more, as a result of this study, we would conclude that university students are responsive to the small-scale investigation of register differences, lexico-grammatical frequency and patterning, which have been brought directly into the L2 classroom.

An Automatic Question Answering Method for Small-Scale Corpus

It is a growing trend for automatic question answering system to be prominent in the development process of society. There are many methods trying to address this problem, but with deficiencies—relatively developed methods based on template matching need a lot of manual work writing templates, and machine learning based methods need plenty of work collecting a large number of corpus, bring huge burden on small-scale scenery. Facing these problems, we propose an automatic question answering methods meeting the needs of small-scale corpus. This method consists of combining an improved text similarity calculation algorithm and an intention recognition method based on slot filling. We conduct experiments on the problem sets of related fields, and it shows a good performance of the proposed automatic question answering method. Our methods make small scene applications for dialog systems more practical.

A study of phatic emoji use in WhatsApp communication

AbstractMobile messaging is considered as a prominent site for phatic communication, where interpersonal connection is often foregrounded over information transaction. Though frequently overlooked, a large amount of this interpersonal work is done nonverbally through regular and meaningful emoji use. This exploratory study deals with emoji use withinLaver’s (1975)phatic token framework, showing that different relationship structures (e.g., status-differential vs. solidary) correspond to distinct phatic token norms. The article analyzes phatic emoji use in a small-scale corpus of WhatsApp interactions between (a) a teacher and her L2-English students and (b) a teacher and her friends/family. Qualitative and quantitative analyses reveal patterns which widely corroborate Laver’s account of socially marked and unmarked token options: the teacher, the students, and the friends/family members tend towards addressee-specific use of neutral, other-oriented, and self-oriented phatic emojis.

A Hybrid Network Model for Tibetan Question Answering

Currently, research on question answering (QA) with deep learning methods is a hotspot in natural language processing. In addition, most of the research mainly focused on English or Chinese since there are large-scale open corpora, such as WikiQA or DoubanQA. However, how to use deep learning methods to QA of the low resource languages, like Tibetan becomes a challenge. In this paper, we propose a hybrid network model for the Tibetan QA, which combines the convolutional neural network and long short memory network (LSTM) to extract effective features from small-scale corpora. Meanwhile, since the strong grammar rules of Tibetan, we use the language model to decode the output of the LSTM layer which makes the answer more accurate and smoother. In addition, we add the batch normalization to accelerate deep network training and prevent overfitting. Finally, the experiments show that the ACC@1 value of the proposed model in Tibetan QA is 126.2% higher than the baseline model.

"We are Proud to be a Leading Company with Global Reach and Worldwide Impact": Positively Evaluative Lexis in the Language of Recruitment Advertising

Companies’ positive self-presentation has become an essential part of job advertising and helps employers attract potential applicants for the vacancies they are offering. This paper focuses on the various forms of positively evaluative lexis that appears in the company profiles cited on the employment website Monster.com. Evaluative or loaded words can be defined as words that evaluate reality either in a positive or a negative way, thus evoking in people a positive or a negative attitude towards a particular subject, in this case the potential employer. The analysis is both qualitative and quantitative and is based on the three-pronged approach, i.e. the combination of manual analysis of individual texts, small-scale corpus analysis and large-scale corpus analysis. It is based primarily on Appraisal Theory, focusing predominantly on the category of attitude expressed in the vocabulary used. However, the analysis has shown that the persuasive techniques of employers trying to attract applicants are rather complex and it has proved necessary to take into account notions from other theories, namely Orientational Metaphor and the concept of the so-called Purr Words, i.e. words whose conceptual meaning is backgrounded by its positive connotations. This concept has recently been developed mostly under the name of Schlagwort within the field of Politolinguistik in Germany, but, as the analysis has shown, can usefully be transferred from political discourse to corporate discourse, including the discourse of recruitment. DOI: http://dx.doi.org/10.5755/j01.sal.33.0.20081

Uyghur morphological analysis using joint conditional random fields: Based on small scaled corpus

As a fundamental research in the field of natural language processing, the Uyghur morphological analysis is used mainly to determine the part of speech (POS) and segmental morphemes (stem and affix) of a word in a given sentence, as well as to automatically annotate the grammatical function of the morphemes based on the context. It is necessary to provide various information for other tasks of natural language processing including syntactic analysis, machine translation, automatic summarization, and semantic analysis, etc. In order to increase the morphological analysis efficiency, this paper puts forward a hybrid approach to create a statistical model for Uyghur morphological tagging through a small-scale corpus. Experimental results show that this plan can obtain an overall accuracy of 92.58 % with a limited training corpus.