Corpus-based Model Research Articles

Exploring the Integration of Corpus-Based and Multimodal Approaches in Academic English Instruction

With the comprehensive advancement of English teaching reforms, the integration of corpus-based methods and the utilization of multimodal educational approaches have emerged as a new developmental trend, aiming to further elevate the quality of academic English instruction. The combination of these two approaches offers an effective means to rectify the various shortcomings present in traditional academic English teaching, consequently enhancing classroom instruction. This paper provides an analysis of the current research landscape and practical application strategies while elucidating the corpus-based multimodal model for academic English instruction.

Problems and Reform of Business English Translation Teaching in the Background of Artificial Intelligence

Abstract In order to solve the problems of business English translation teaching, a corpus-based teaching model is proposed. The API and web crawler are used to collect data from the business English corpus, and the overall corpus structure design is completed according to the acquired data. The corpus is pre-processed with lexical processing, theme extraction, category labeling, and other operations to ensure the feasibility of the corpus. The limitations of traditional business English translation teaching are highlighted, and the structure and implementation process of the corpus-based business English translation teaching model are thoroughly examined. The research subjects are selected, and experimental comparisons are applied empirically to analyze the corpus-based business English translation teaching. The data show that the mean translation score of the experimental group increased by 8.1247. In contrast, the mean translation score of the control group not only didn’t improve but also decreased by 0.8806, and the mean interpreting score of the experimental group increased by 0.608. In contrast, the mean interpreting score of the control group increased by 0.172, which indicates that the corpus-based business English teaching mode has a facilitating effect on the improvement of the student’s abilities in all aspects of English.

Revisiting the binary view of honorifics in politeness research

Abstract In this paper, we revisit the popular assumption that politeness in languages such as Japanese and Korean with a complex honorific system is crucially different from politeness in languages with no comparably rich honorific repertoires, such as Chinese. We propose a bottom–up, contrastive and corpus-based model through which we challenge this binary view. This model combines interaction ritual and speech acts. As a case study, we compare a set of expressions representing lexico-grammatical honorifics in Japanese and Chinese, i.e., in a so-called “honorific-rich” and a “non-honorific-rich” language. Our results show that the group of honorifics studied work in an essentially comparable fashion, hence disproving the above-outlined binary view.

Unraveling lexical semantics in the brain: Comparing internal, external, and hybrid language models.

To explain how the human brain represents and organizes meaning, many theoretical and computational language models have been proposed over the years, varying in their underlying computational principles and in the language samples based on which they are built. However, how well they capture the neural encoding of lexical semantics remains elusive. We used representational similarity analysis (RSA) to evaluate to what extent three models of different types explained neural responses elicited by word stimuli: an External corpus-based word2vec model, an Internal free word association model, and a Hybrid ConceptNet model. Semantic networks were constructed using word relations computed in the three models and experimental stimuli were selected through a community detection procedure. The similarity patterns between language models and neural responses were compared at the community, exemplar, and word node levels to probe the potential hierarchical semantic structure. We found that semantic relations computed with the Internal model provided the closest approximation to the patterns of neural activation, whereas the External model did not capture neural responses as well. Compared with the exemplar and the node levels, community-level RSA demonstrated the broadest involvement of brain regions, engaging areas critical for semantic processing, including the angular gyrus, superior frontal gyrus and a large portion of the anterior temporal lobe. The findings highlight the multidimensional semantic organization in the brain which is better captured by Internal models sensitive to multiple modalities such as word association compared with External models trained on text corpora.

A content-aware corpus-based model for analysis of marine accidents

In the past decades, marine accidents brought the serious loss of life and property and environmental contamination. With the accumulation of marine accident data, especially accident investigation reports, compared with subjective reasoning based on expert experience, data-driven methods for analysis and accident prevention are more comprehensive and objective. This paper aims to develop a content-aware corpus-based model for the analysis of marine accidents to mine the accident semantic features. The general research framework is established to combine accident data, expert prior knowledge, and semi-automated natural language processing (NLP) technology. The NLP models are optimized, fused, and applied to the case study of ship collision accidents. The results show that the proposed model can accurately and quickly extract hazards, accident causes, and scenarios from the accident reports, and perform semantic analysis for the latent relationships between them to extend the accident causation theory. This study can provide a powerful and innovative analysis tool for marine accidents for maritime traffic safety management departments and relevant research institutions.

An analysis of Rem Koolhaas's discourses on architecture and urban design using a corpus-based model

This paper explores the networks of thoughts behind words by reading architectural texts in the context of linguistics. Using a corpus-based model, our main interest is what the collocational data contribute at an architectural level. The notion of “collocation identification in architecture” was drawn from an analysis of the use of the data for Rem Koolhaas. The data for this study were collected from Rem Koolhaas's articles from 1977 to 2014. An analysis of Koolhaas's ideas about architecture and urban planning was conducted using the AntConc 3.4.4.0 corpus analysis toolkit. The corpus-based model reduced selected articles to words and lexical bundles using the corpus analysis toolkit. This paper explores Koolhaas's thoughts on understanding architecture and the metaphors he uses by analyzing articles using the corpus-based analysis model. The discourse codes in the texts of Koolhaas are examined. A corpus-based model proposal tested on architectural texts has the potential to create new areas of knowledge for architectural discourse. Deciphering the texts and analyzing the codes contribute to the development of new models.

Stochastic Data-to-Text Generation Using Syntactic Dependency Information

Data-to-Text Generation (D2T) is one of the most important sub-fields of Natural Language Generation where structured data is transcribed into natural language text. Several solutions have been proposed for D2T so far with relative success, including template-based, phrase structure grammar-based, and neural attention models. However, these methods also have problems such as grammatical flaws, limited naturalness, and semantic deficiencies. In this work, we propose a stochastic corpus-based model for the data-to-text generation that produces a tree-form structure for sentences based on dependency information. This information includes the dependency relations between words and meaning labels extracted from the aligned training sentences parsed with a dependency parser. By combining the dependency relations and meaning labels to construct a tree structure in an up-down manner, each word is placed into the output sentence based on its preceding and succeeding words. This results in fluent sentences with correct grammatical structures. This approach also ensures that all required semantic information are present in the output sentences while irrelevant or redundant labels are avoided. In addition, by using beam search in producing the structure of sentences, the proposed model can generate highly diverse sentences. We test our model on eight domains in tabular, dialogue act, and RDF formats. Our model improves the BLEU by 30% compared to the corpus-based state-of-the-art methods trained on the tabular datasets and also achieves comparable results with the neural network-based approaches trained on dialogue act, E2E, and WebNLG datasets in the BLEU evaluation metric. Furthermore, the value of ERR metric for our results is always zero; that means our model generates sentences without losing any information. Human evaluations show that our model produces high-quality utterances in aspects of informativeness and naturalness as well as quality.

A Corpus-based empirical study on energy enterprises digital transformation

[Objective/Significance]This study aims to explore the role and mechanism of dynamic KM for the digital transformation of energy enterprises. This paper conducts an empirical study on mechanism of energy enterprises’ digital transformation from the perspective of KM, which is likely to contribute to relevant academic literatures. [Method/Process]On the basis of the KM theory, the SECIE Model was constructed, together with a self-built Corpus-based empirical model; Corpus’ AHP-DEMATEL evaluation system applies to verify both reliability and consistency of the model by Concordance and Matrix. [Results/Conclusions]The research results show that KM during the energy firm digitalization is evidently a dynamic process, rather than static, in which there are 5 key elements(with corpus word hits), namely, learning (942 hits), embed (771 hits), internalize (685 hits), adapt (614 hits), integration (315 hits), that are identified and worth paying attention to. Other than that , those five factors are essential for influencing the digital transformation of enterprises; empirical study confirms that they all play a positive and significant role. Knowledge integration possesses the greatest weight(0.28), followed by internalization(0.19), learning(0.18) and adaptation(0.18), with embedding(0.16) being much lower. Based on the results, this paper draws a conclusion: the mechanism of KM for promoting digital transformation of energy enterprises lies in the integration and optimization of digital resources. “Self-transcendence knowledge” is very pivotal in the role of knowledge management in promoting energy firms to swiftly respond to future digital opportunities.

A Corpus-Based Research on Ditransitive Constructions with Tell and Say

Learners’ errors give insights to teachers, textbook writers, curriculum designers and many applied linguists about the learning difficulty in the acquisition of a target language. Studying systematically these errors is therefore considered indispensable in learning teaching process. Basing on the corpus-based model, this paper investigates the constructions of two ditransitive verbs: TELL and SAY which cause much trouble to second language learners. After analysing the exam copies of 200 second-year students in the English department of Cheikh Anta DIOP University of Dakar, the study comes up with two general observations. First, the analysis of the overall data shows that learners of the corpus largely prefer using TELL (62, 5%) to SAY (14, 5%). The second observation is, these students use more correctly the ditransitive form involving TELL than SAY. For instance, 79, 2% of them employ TELL in the double object construction corresponding to the basic structure TELL + someone + something, while only 5,1% correctly use the dative construction of SAY which is SAY + something + to + someone. These findings conform to several studies which claim that the dative form is the most complicated construction, and is consequently the rarely used one. All of this indicates, that even if these students are English majors, they are still in their basic level in the acquisition of ditransitive constructions. This suggests that special strategies and mechanisms are required in teaching and learning ditransitive verbs. More efforts are also needed in teaching and learning constructions in grammar (e.g. alternating pairs like Passive /Active, Will /Be going to, Verb-particle constructions etc.). This will help students become more accurate in using English, the target language.

From ‘Ratcatraz Prison’ to ‘Penjara Pudutikus’: Lexical Creativity in Children’s Literature and its Translation into Malay

Lexical creativity is one of the ways through which authors of children’s literature shape their stories, entertain children and build their identity and style. In spite of their uniqueness, creative lexical items are often replaced with more common words in translation, a phenomenon described through the law of growing standardisation hypothesis. In view of the importance of creative lexical items in children’s literature, there is a need to explore how creativity is transferred in the process of translation. This study is carried out to examine the translation of lexical creativity in children’s literature from English to Malay. More specifically, the aims of the study are to identify the types of creative lexical items in children’s literature in English and to determine how translators transfer creativity from English into Malay. The study also aims to determine whether the hypothesis of the law of growing standardisation applies in the context of the translation of children’s literature from English into Malay. The study employs a corpus-based model for research on the translation of creative lexical items. Corpus analysis tools are used to identify different forms of creative lexical items and to identify the translations for these items. Based on the analysis, five types of creative lexical items are identified; many, however, are replaced with common words when translated into Malay, resulting in the loss of unique features of the original works. It is, however, also found that translators compensate for some of these losses by introducing other elements in the Malay translations.

The Anatomy of Consonance/Dissonance: Evaluating Acoustic and Cultural Predictors Across Multiple Datasets with Chords

Acoustic and musical components of consonance and dissonance perception have been recently identified. This study expands the range of predictors of consonance and dissonance by three analytical operations. In Experiment 1, we identify the underlying structure of a number of central predictors of consonance and dissonance extracted from an extensive dataset of chords using a hierarchical cluster analysis. Four feature categories are identified largely confirming the existing three categories (roughness, harmonicity, familiarity), including spectral envelope as an additional category separate from these. In Experiment 2, we evaluate the current model of consonance/dissonance by Harrison and Pearce by an analysis of three previously published datasets. We use linear mixed models to optimize the choice of predictors and offer a revised model. We also propose and assess a number of new predictors representing familiarity. In Experiment 3, the model by Harrison and Pearce and our revised model are evaluated with nine datasets that provide empirical mean ratings of consonance and dissonance. The results show good prediction rates for the Harrison and Pearce model (62%) and a still significantly better rate for the revised model (73%). In the revised model, the harmonicity predictor of Harrison and Pearce’s model is replaced by Stolzenburg’s model, and a familiarity predictor coded through a simplified classification of chords replaces the original corpus-based model. The inclusion of spectral envelope as a new category is a minor addition to account for the consonance/dissonance ratings. With respect to the anatomy of consonance/dissonance, we analyze the collinearity of the predictors, which is addressed by principal component analysis of all predictors in Experiment 3. This captures the harmonicity and roughness predictors into one component; overall, the three components account for 66% of the consonance/dissonance ratings, where the dominant variance explained comes from familiarity (46.2%), followed by roughness/harmonicity (19.3%).

‘Politeness Markers’ Revisited - A Contrastive Pragmatic Perspective

AbstractThis paper revisits the concept of ‘politeness marker’, by proposing the bottom-up and corpus-based model of ‘ritual frame indicating expressions’ (RFIEs). Our central argument is that, in certain linguacultures, the relationship between ‘politeness markers’ and politeness itself is significantly stronger than in others. Therefore, any theory which argues that there is a definite relationship between form and politeness - or totally rejects this relationship - is potentially problematic if it does not take a contrastive pragmatic perspective, simply because this relationship is subject to significant linguacultural variation. The contrastive pragmatic study of RFIEs also helps us to determine the relationship between forms and speech acts and, indirectly, politeness. As a case study, we examine in this paper one-word and more complex expressions which are commonly associated with the speech acts of request and apology, drawn from the typologically distant Chinese and English linguacultures.

Agreement syncretization and the loss of null subjects: quantificational models for Medieval French

AbstractThis paper examines the nature of the dependency between the availability of null subjects and the “richness” of verbal subject agreement, known as Taraldsen's Generalisation (Adams, 1987; Rizzi, 1986; Roberts, 2014; Taraldsen, 1980). We present a corpus-based quantitative model of the syncretization of verbal subject agreement spanning the Medieval French period and evaluate two hypotheses relating agreement and null subjects: one relating the two as reflexes of the same grammatical property and a variational learning-based hypothesis whereby phonology-driven syncretization of agreement marking creates a learning bias against the null subject grammar. We show that only the latter approach has the potential to reconcile the intuition behind Taraldsen's Generalisation with the fact that it has proven nontrivial to formulate the notion of agreement richness in a way that would unequivocally predict whether a language has null subjects.

CLASENTI

Arabic text sentiment analysis suffers from low accuracy due to Arabic-specific challenges (e.g., limited resources, morphological complexity, and dialects) and general linguistic issues (e.g., fuzziness, implicit sentiment, sarcasm, and spam). The limited resources problem requires efforts to build new and improved Arabic corpora and lexica. We propose a class-specific sentiment analysis (CLASENTI) framework. The framework includes a new annotation approach to build multi-faceted Arabic corpus and lexicon allowing for simultaneous annotation of different facets, including domains, dialects, linguistic issues, and polarity strengths. Each of these facets has multiple classes (e.g., the nine classes representing dialects found in the Arab world). The new corpus and lexicon annotations facilitate the development of new class-specific classification models and polarity strength calculation. For the new sentiment classification models, we propose a hybrid model combining corpus-based and lexicon-based models. The corpus-based model has two interrelated phases to build; (1) full-corpus classification models for all facets; and (2) class-specific models trained on filtered subsets of the corpus according to the performances of the full-corpus models. To calculate polarity strengths, the lexicon-based model filters the annotated lexicon based on the specific classes of the domain and dialect. As a case study, we collect and annotate 15274 reviews from various sources, including surveys, Facebook comments, and Twitter posts, pertaining to governmental services. In addition, we develop a new web-based application to apply the proposed framework on the case study. CLASENTI framework reaches up to 95% accuracy and 93% F1-Score surpassing the best-known sentiment classifiers implemented in Scikit-learn library that achieve 82% accuracy and 81% F1-Score for Arabic when tested on the same dataset.

A Corpus-based Discourse Information Analysis of Chinese EFL Learners’ Autonomy in Legal Case Brief Writing

Legal case brief writing is pedagogically important yet insufficiently discussed for Chinese EFL learners majoring in law. Based on process genre approach and discourse information theory (DIT), the present study designs a corpus-based analytical model for Chinese EFL learners’ autonomy in legal case brief writing and explores the process of case brief formation. The results show that in the stages of writing and self-assessment, learners can make full use of the corpus to obtain information, complete macro-information and micro-information output, and self-assess the specific legal register. The findings indicate that the realization of global coherence and local coherence in case briefs depends on discourse information output with appropriate language use. It is hoped that the study will help Chinese EFL learners build information awareness in legal writing and promote their language output and autonomous learning.

A Corpus-based Toy Model for DisCoCat

The categorical compositional distributional (DisCoCat) model of meaning rigorously connects distributional semantics and pregroup grammars, and has found a variety of applications in computational linguistics. From a more abstract standpoint, the DisCoCat paradigm predicates the construction of a mapping from syntax to categorical semantics. In this work we present a concrete construction of one such mapping, from a toy model of syntax for corpora annotated with constituent structure trees, to categorical semantics taking place in a category of free R-semimodules over an involutive commutative semiring R.

The language of stories: Modelling East African fiction and oral narratives

Few corpus-based studies have looked at the linguistic features that co-create narrative focus in a text. In this article, a corpus-based narrativity model with four feature groups is proposed that reflects the underlying structure of narratives: Agency, contextualisation, evaluation and causation. Agents affect events or are in turn affected by them; events are contextualised in space and time; the affective impact is reflected by evaluative language; and cause and effect give the narrative momentum. The aim was to illustrate how the narrativity model could be used to identify and map linguistic features associated with narratives in a corpus of East African English. A quantitative approach was followed to analyse and compare East African English fiction and oral narratives. The four feature groups included 16 linguistic features and WordSmith Tools 4.0 was used for the analyses. The results show that eight linguistic features frequently co-occur in fiction and oral narratives. However, some of the features are more frequent in fiction, whereas others are typical of oral narratives. First, fiction and oral narratives are discussed independently and then these registers are compared. Overall, the model is sufficiently robust to identify texts with a narrative focus.

The cognitive plausibility of statistical classification models: Comparing textual and behavioral evidence

AbstractUsage-based linguistics abounds with studies that use statistical classification models to analyze either textual corpus data or behavioral experimental data. Yet, before we can draw conclusions from statistical models of empirical data that we can feed back into cognitive linguistic theory, we need to assess whether the text-based models are cognitively plausible and whether the behavior-based models are linguistically accurate. In this paper, we review four case studies that evaluate statistical classification models of richly annotated linguistic data by explicitly comparing the performance of a corpus-based model to the behavior of native speakers. The data come from four different languages (Arabic, English, Estonian, and Russian) and pertain to both lexical as well as syntactic near-synonymy. We show that behavioral evidence is needed in order to fine tune and improve statistical models built on data from a corpus. We argue that methodological pluralism is the key for a cognitively realistic linguistic theory.

A new family of information content models with an experimental survey on WordNet

This paper introduces a new family of intrinsic and corpus-based Information Content (IC) models for ontology-based similarity measures based on the IC theory, a detailed state of the art, an experimental survey of IC models and IC-based similarity measures on WordNet, and a comparison between intrinsic and corpus-based IC models. The family of IC models is made up of five intrinsic IC models, called CondProbHypo, CondProbUniform, CondProbLeaves, CondProbLogistic, and CondProbCosine, and one corpus-based IC model called CondProbCorpus which completes the family. The proposed IC models rely on two previously unconsidered notions: (1) the preservation of the probabilistic structure of the taxonomy associated to the conditional probabilities between child and parent concepts, and (2) the explicit consideration of a cognitive similarity notion in the definition of the IC model. The family of IC models defines a new method for the proposal of new intrinsic IC models based on the exploration of other alternatives for the intrinsic estimation of the conditional probabilities between child and parent concepts. Our work is inspired by an unexplored relationship between the Jiang–Conrath distance and a shortest path on an IC-based weighted graph, derived from the conditional probabilities between concepts, as well as certain cognitive evidence about the perception distance between concepts. The new IC models obtain results comparable to the state of the art and satisfy a set of well-founded structure axioms. In addition, we prove that most of intrinsic IC models and IC-based similarity measures do not show a significant statistical difference as regards a baseline corpus-based IC model and the Jiang–Conrath similarity, with the exception of the overall outperformance shown by the Sánchez et al. (2012) IC model and the cosJ&C similarity measure, which has recently been introduced by the authors.

Unsupervised Phrasal Near-Synonym Generation from Text Corpora

Unsupervised discovery of synonymous phrases is useful in a variety of tasks ranging from text mining and search engines to semantic analysis and machine translation. This paper presents an unsupervised corpus-based conditional model: Near-Synonym System (NeSS) for finding phrasal synonyms and near synonyms that requires only a large monolingual corpus. The method is based on maximizing information-theoretic combinations of shared contexts and is parallelizable for large-scale processing. An evaluation framework with crowd-sourced judgments is proposed and results are compared with alternate methods, demonstrating considerably superior results to the literature and to thesaurus look up for multi-word phrases. Moreover, the results show that the statistical scoring functions and overall scalability of the system are more important than language specific NLP tools. The method is language-independent and practically useable due to accuracy and real-time performance via parallel decomposition.