Balanced Corpus Research Articles

Usage and empirical productivity of international adjectival suffixes in Slovak based on general and specialised corpora

Abstract The paper attempts to identify the usage and productivity of five different international suffixes in Slovak by means of corpus evidence. The analysis focuses on real and potential productivity in a two-stage comparison: 1) tokens/lemmas occurring in a general balanced corpus vs general corpus of specialised and academic texts, 2) general corpus of specialised and academic texts vs specialised (sub)corpora of medical, legal, economic and religious texts. The aim of the analysis is to explore whether productivity varies across registers by means of statistical measures.

Can a corpus-driven lexical analysis of human and machine translation unveil discourse features that set them apart?

Abstract There is still much to learn about the ways in which human and machine translation differ with regard to the contexts that regulate the production and interpretation of discourse. The present study explores whether a corpus-driven lexical analysis of human and machine translation can unveil discourse features that set the two apart. A balanced corpus of source texts aligned with authentic, professional translations and neural machine translations was compiled for the study. Lexical discrepancies in the two translation corpora were then extracted via a corpus-driven keyword analysis, and examined qualitatively through parallel concordances of source texts aligned with human and machine translation. The study shows that keyword analysis not only reiterates known problems of discourse in machine translation such as lexical inconsistency and pronoun resolution, but can also provide valuable insights regarding contextual aspects of translated discourse deserving further research.

Overcoming issues in frequency-based extraction and lexicographic lnclusion of Korean neologisms

This paper discusses issues regarding frequency as a criterion for Korean neologism extraction from the perspective of corpus linguistics and lexicography. Most studies agree that frequency plays a central role in the inclusion of neologisms in the dictionary; however, frequency entails a number of complex factors such as the time span of a word’s use as well as the variety of registers. The use of web data to extract neologisms – instead of a balanced corpus – has brought about a new range of issues that call for new ways to address them. Section 2 reviews previous research trends related to neologism frequency from the point of view of linguistics and neologism studies. Section 3 examines and discusses issues in the detection of phrasal and semantic neologisms, and in the use of Web corpora. Section 4 suggests the use of triangulation in order to cope with such shortcomings, combining use-based methodology and used-based approach.

Metaphors of mental illness: a corpus-based approach analysing first-person accounts of patients and mental health professionals

In this paper we describe the building, manual annotation and analysis of a balanced corpus to assess conceptual metaphors on mental illness as used in Spanish blogger writing by patients and mental health professionals. The corpus was structured as eight subgroups: four patient subgroups (composed of persons who declared having been diagnosed with major depression, schizophrenia, bipolar disorder, or obsessive-compulsive disorder) and four mental health professional subgroups (psychiatrists, psychologists, social educators, nurses). The quantitative analysis identified similarities and differences between groups regarding the volume of metaphors produced and the topics linguistically expressed through metaphors. The most frequent metaphors used by each major group, patients and professionals, were qualitatively analysed, with the principal findings showing a set of source domains used to conceptualize all four severe mental disorders, thus pointing to a common conceptualization of mental suffering irrespective of the specific diagnosis, and two major types of metaphors, WAR and JOURNEY, used by all subgroups of patients and professionals to talk about their first-hand experiences.

Balanced Arabic corpus design for speech synthesis

This paper aims to design and validate a phonetically balanced speech corpus for Arabic language. Designing and developing a rich and phonetically balanced corpus in optimal context is one of the key issues in building high quality of text-to-speech synthesis systems. The rich characteristic is in the sense that it must contain all the possible phonemes on the right and left context, whereas the balanced characteristic is in the sense that it respects the phonetic distribution in the language. We propose a new methodology for designing and implementing such corpus for speech synthesis purposes. The paper explains the whole creation process of this corpus, beginning with the design stage, corpus creation, recording phases, and finally the segmentation of the speech corpus. The speech corpus contains 202 sentences with 6174 phonemes. In order to validate the speech corpus, an Arabic speech synthesis system using Hidden Markov Models has been developed.

Insubordinate Conditional Clauses Formed by Japanese Conjunction -ba

Insubordination describes the phenomenon that a subordinate clause is used independently as a main clause. Insubordinate clauses endorse discourse functions alone without connecting with any apodosis. Many researchers studied this phenomenon from a typology perspective, clarifying the discourse functions and forming mechanism. However, those studies pursuit the cross-language features too much, investigating large numbers of languages and source constructions. As a result, their conclusion may not fit a specific language or construction. This paper aims to verify and complement the classification of one of the typology studies: Evans (2007), based on the investigation of the insubordinate conditional clauses formed by Japanese conjunction -ba, through the Balanced Corpus of Contemporary Written Japanese (BCCWJ). The investigation result showed that Evans’ (2007) classification fit the insubordinate -ba-clauses to some extent with a slight adjustment. Specifically, the types of “indirection and interpersonal control” and “modal insubordination” were combined as Type A, which was named “modal insubordination”. Meanwhile, a new discourse function, “requirement of further information” was added to the other type, “signaling presupposed material” to compose Type B, which was named “signaling presupposed material”. In Type A, the insubordinate -ba-clauses function as two subtypes, which are “epistemic and evidential meanings” and “deontic meanings”. Those two subtypes relate to each other by expressing or requiring evaluation towards the conditional clauses. Meanwhile, Type B has two subtype functions as well, which are “disagreement with assertions by the previous speaker” and “requirement of further information”. Both subtypes are formed by the process of providing a contrastive condition and requesting a response. Two further related issues were also discussed. The insubordinate -ba-clauses endorsing Type B functions appear to have some similarities with another type of insubordinate clause that was pointed out by other typology studies. However, they indeed differ from each other in their discourse functions and forming mechanisms. Moreover, as to the -ba-clauses, the distinction between Type A and Type B was found to be more apparent when analyzed from a functional perspective rather than a structural perspective. Those arguments complement Evans’ (2007) classification. This study shows its significance in verifying Evans’(2007) viewpoint by the Japanese language as a unique Asian language. Such verification is necessary because Evans’ research was carried out mostly based on Indo‐European languages. It is also implicated that a theory of general linguistics can contribute to studying an individual language in observing it from an outside perspective.

Collocations and contextual semantics of the ethnonyms leiši and lietuvieši ‘Lithuanians’ in the text corpus of modern Latvian language

The aim of this paper is, by using Computational linguistics method to analyse collocations of ethnonyms leiši and lietuvieši ‘Lithuanians’ in “Balanced Corpus of Modern Latvian” (Līdzsvarots mūsdienu latviešu valodas tekstu korpuss), compare contextual semantics of both ethnonyms. Comparing the frequency of use of the lexemes lietuvieši and leiši ‘Lithuanians’ in the modern Latvian language text corpus, the prevalence of the ethnonym lietuvieši is evident, therefore, the ethnonym leiši, regardless of its use as synonymic designation (sometimes in one and the same text), can be considered as an obsolete word. The desemantisation cases of the ethnonym leiši, detected in the corpus, show its oldness and varied functionality in Latvian language. The evidences to prove the statement of the “Modern Latvian Language Dictionary” (Mūsdienu latviešu valodas vārdnīca) that the lexeme leiši nowadays “carries slightly pejorative stylistically expressive colouring” were not found in the text corpus. In general, the image of a Lithuanian reflected in the modern Latvian language text corpus is rather positive – mostly in historical, language, culture and sports contexts, yet in the context of economics and emigration rather negative impression of Lithuanians is expressed.

A Corpus-based Study of the Use of Adjectives in Spoken and Written Registers

The present study is designed to investigate the use of adjectives in spoken and written registers. The word “strong” is taken as an example and its synonyms are retrieved from TV/Movies Corpus and Academic Journals Corpus for analysis in terms of word frequency and semantic preferences. The results indicate that there are significant differences in the word frequency and semantic features of the adjectives between oral and written registers. 1) Adjectives in written register generally occur with higher frequency than those in spoken register. 2) The polysemy of adjectives is not determined by the word frequency, but consistent with linguistic registers. The adjectives used in written English carry a wider range of extended and metaphorical meanings than those in spoken English. 3) Nouns in adjective-noun collocations tend to be abstract and indirect in written register, but concrete and direct in spoken register. 4) Nouns influence the negative, neutral or positive semantic prosody of the collocations. As a corpus-based empirical research, the study also offers a few constructive suggestions for improving the efficiency of vocabulary acquisition and teaching by using authentic data from balanced corpora.

SQLSketch: Generating SQL Queries using a sketch-based approach

In the last decade, many intelligent interfaces and layers have been suggested to allow the use of relational databases and extraction of the content using only the natural language. However most of them struggle when exposed to new databases. In this article, we present SQLSketch, a sketch-based network for generating SQL queries to address the problem of automatically translate Natural Languages questions to SQL using the related databases schemas. We argue that the previous models that use full or partial sequence-to-sequence structure in the decoding phase can, in fact, have counter-effect on the generation operation and came up with more loss of the context or the meaning of the user question. In this regard, we use a full sketch-based structure that decouples the generation process into many small prediction modules. The SQLSketch is evaluated against GreatSQL, a new cross-domain, large-scale and balanced dataset for the Natural Language to SQL translation task. For a long-term aim of making better models and contributing in adding more improvements to the semantic parsing tasks, we propose the GreatSQL dataset as the first balanced cross-domain corpus that includes 45,969 pairs of natural language questions and their corresponding SQL queries in addition to simplified and well structured ground-truth annotations. We establish results for SQLSketch using GreatSQL dataset and compare the performance against two popular types of models that represent the sequential and partial-sketch based approaches. Experimental result shows that SQLSketch outperforms the baseline models by 13% in exact matching accuracy and achieve a score of 23.9% to be the new state-of-the-art model on GreatSQL.

ВОСПРИЯТИЕ СВОБОДЫ В КИТАЙСКОЙ ЯЗЫКОВОЙ КАРТИНЕ МИРА: ТРАДИЦИЯ И СОВРЕМЕННОСТЬ

This article examines the changes in the linguistic and cultural concepts of freedom (自由 zìyóu) in the Chinese linguistic worldview, which have occurred as a result of extralinguistic and socio-cultural transformations brought by globalization processes between the 20th and 21st centuries. The starting point of the research is the traditional ideas about freedom, rooted in Confucianism and the Taoist worldview, according to which disobedience and self-will of the individual can lead society and the country to chaos, so freedom should be limited by norms and prohibitions. By studying empirical data with the help of definitional, quantitative, interpretative, and contextual analysis, a total of more than 1000 phrases with the lexical unit of interest were gathered. Thus, the idea of freedom in modern Chinese society was described in great detail. The results obtained were accompanied with the data of the quantitative analysis of the most frequent collocations based on the materials of the Balanced Corpus of Modern Chinese in Peking University. The article provides many examples that prove that the lexeme 自由 zìyóu occurs in phrases that describe numerous and diverse spheres of human activity. The conclusion was made that it has expanded its combinability capabilities as a result of the semantic shift that took place with the penetration of liberal ideas into China from the West. The research is of great importance and benefit for Chinese philologists, as well as for anyone interested in studying globalization processes and their impact on the traditional cultures of the East.

A Multi-dimensional Approach to Register Variations in Mandarin Chinese

Multi-dimensional Analysis (MD) is a quantitative corpus-based approach which describes and interprets patterns of register variations through factor analysis of a set of linguistic fea-tures across text varieties, and reveals their systematic relationships with communicative purposes. The model has been employed to explore language variation in many languages (e.g., English, Somali, Nukulaelae Tuvaluan, Korean, and Spanish), yet insufficient research has been carried out on register variation in Mandarin Chinese on a full scale. In this research, 88 linguistic features are tagged in a balanced corpus composed of 20 Mandarin Chinese spoken and written registers. Through factor analysis, five dimensions which consist of 65 linguistic features are identified and interpreted from linguistic and functional perspectives. The first two dimensions, interactive vs. informational discourse and narrative vs. non-narrative concern, are similar to dimensions that have been claimed to constitute universal parameters of register variation in previous MD studies. The exist-ence of two potential universal dimensions suggests that the basic communicative purposes and functions underlying the different languages are markedly similar, given the existing social, cultural, and linguistic dissimilarities. Dimension 4, casual real-time speech with stance, is identified as a distinctive dimension in Mandarin Chinese. Dimension 3, explicit-ness in cohesion and reasoning, and Dimension 5, abstract information, are found to be as-sociated with foreign influence, and their register variation patterns illustrate how foreign contact affects Chinese register variation in a quantitative manner.

Database of Figurative Expressions with Indicators from the ‘Balanced Corpus of Contemporary Written Japanese’

Database of Figurative Expressions with Indicators from the ‘Balanced Corpus of Contemporary Written Japanese’

Consonant and vowel articulation accuracy in younger and middle-aged Spanish healthy adults.

Children acquire vowels earlier than consonants, and the former are less vulnerable to speech disorders than the latter. This study explores the hypothesis that a similar contrast exists later in life and that consonants are more vulnerable to ageing than vowels. Data was obtained with two experiments comparing the speech of Younger Adults (YAs) and Middle-aged Adults (MAs). In the first experiment an Automatic Speech Recognition (ASR) system was trained with a balanced corpus of 29 YAs and 27 MAs. The productions of each speaker were obtained in a Spanish language word (W) and non-word (NW) repetition task. The performance of the system was evaluated with the same corpus used for training using a cross validation approach. The ASR system recognized to a similar extent the Ws of both groups of speakers, but it was more successful with the NWs of the YAs than with those of the MAs. Detailed error analysis revealed that the MA speakers scored below the YA speakers for consonants and also for the place and manner of articulation features; the results were almost identical in both groups of speakers for vowels and for the voicing feature. In the second experiment a group of healthy native listeners was asked to recognize isolated syllables presented with background noise. The target speakers were one YA and one MA that had taken part in the first experiment. The results were consistent with those of the ASR experiment: the manner and place of articulation were better recognized, and vowels and voicing were worse recognized, in the YA speaker than in the MA speaker. We conclude that consonant articulation is more vulnerable to ageing than vowel articulation. Future studies should explore whether or not these early and selective changes in articulation accuracy might be caused by changes in speech perception skills (e.g., in auditory temporal processing).

Detecting misogyny in Spanish tweets. An approach based on linguistics features and word embeddings

Online social networks allow powerless people to gain enormous amounts of control over particular people’s lives and profit from the anonymity or social distance that the Internet provides in order to harass other people. One of the most frequently targeted groups comprise women, as misogyny is, unfortunately, a reality in our society. However, although great efforts have recently been made to identify misogyny, it is still difficult to distinguish as it can sometimes be very subtle and deep, signifying that the use of statistical approaches is not sufficient. Moreover, as Spanish is spoken worldwide, context and cultural differences can complicate this identification. Our contribution to the detection of misogyny in Spanish is two-fold. On the one hand, we apply Sentiment Analysis and Social Computing technologies for detecting misogynous messages in Twitter. On the other, we have compiled the Spanish MisoCorpus-2020, a balanced corpus regarding misogyny in Spanish, and classified it into three subsets concerning (1) violence towards relevant women, (2) messages harassing women in Spanish from Spain and Spanish from Latin America, and (3) general traits related to misogyny. Our proposal combines a classification based on average word embeddings and linguistic features in order to understand which linguistic phenomena principally contribute to the identification of misogyny. We have evaluated our proposal with three machine-learning classifiers, achieving the best accuracy of 85.175%. Finally the proposed approach is also validated with existing corpora for misogyny and aggressiveness detection such as AMI and HatEval obtaining good results

Detecting hate speech against politicians in Arabic community on social media

Purpose This paper aims to propose an approach for hate speech detection against politicians in Arabic community on social media (e.g. Youtube). In the literature, similar works have been presented for other languages such as English. However, to the best of the authors’ knowledge, not much work has been conducted in the Arabic language. Design/methodology/approach This approach uses both classical algorithms of classification and deep learning algorithms. For the classical algorithms, the authors use Gaussian NB (GNB), Logistic Regression (LR), Random Forest (RF), SGD Classifier (SGD) and Linear SVC (LSVC). For the deep learning classification, four different algorithms (convolutional neural network (CNN), multilayer perceptron (MLP), long- or short-term memory (LSTM) and bi-directional long- or short-term memory (Bi-LSTM) are applied. For extracting features, the authors use both Word2vec and FastText with their two implementations, namely, Skip Gram (SG) and Continuous Bag of Word (CBOW). Findings Simulation results demonstrate the best performance of LSVC, BiLSTM and MLP achieving an accuracy up to 91%, when it is associated to SG model. The results are also shown that the classification that has been done on balanced corpus are more accurate than those done on unbalanced corpus. Originality/value The principal originality of this paper is to construct a new hate speech corpus (Arabic_fr_en) which was annotated by three different annotators. This corpus contains the three languages used by Arabic people being Arabic, French and English. For Arabic, the corpus contains both script Arabic and Arabizi (i.e. Arabic words written with Latin letters). Another originality is to rely on both shallow and deep leaning classification by using different model for extraction features such as Word2vec and FastText with their two implementation SG and CBOW.

신어의 빈도와 관련한 몇 가지 문제

This paper presents the current research trends in neologisms and discusses issues regarding the use of frequency criteria for the analysis of Korean neologisms from the perspective of corpus linguistics and lexicography. The literature on neologisms usually agrees that frequency plays a central role in the inclusion of neologisms in the dictionary; however, frequency is more complex than it appears as it entails a number of factors such as the time span of a word’s use as well as the variety of registers and aspects of frequency. The use of Web data to extract neologisms – instead of a balanced corpus – has brought about a new range of more complex issues that call for new ways to be tackled. This study investigates the issues raised by the application of frequency criteria to the neologism analysis when using Web corpus methodology. Chapter 2 reviews previous research trends related to neologism frequency from the point of view of linguistics and neologism studies. Chapter 3 examines and discusses frequency issues in terms of the unit and object of neologisms (3.1), Web register variations (3.2), and the relationship between frequency and users’ perception (3.3).

BCCWJ-EyeTrack : 『現代日本語書き言葉均衡コーパス』に対する読み時間付与とその分析

BCCWJ-EyeTrack : 『現代日本語書き言葉均衡コーパス』に対する読み時間付与とその分析

Classification of regional and genre varieties of Chinese: A correspondence analysis approach based on comparable balanced corpora

AbstractThis paper proposes a robust text classification and correspondence analysis approach to identification of similar languages. In particular, we propose to use the readily available information of clauses and word length distribution to model similar languages. The modeling and classification are based on the hypothesis that languages are self-adaptive complex systems and hence can be classified by dynamic features describing the system, especially in terms of distributional relations of constituents of a system. For similar languages whose grammatical differences are often subtle, classification based on dynamic system features should be more effective. To test this hypothesis, we considered both regional and genre varieties of Mandarin Chinese for classification. The data are extracted from two comparable balanced corpora to minimize possible confounding factors. The two corpora are the Sinica Corpus from Taiwan and the Lancaster Corpus of Mandarin Chinese from Mainland China, and the two genres are reportage and review. Our text classification and correspondence analysis results show that the linguistically felicitous two-level constituency model combining power functions between word and clauses effectively classifies the two varieties of Chinese for both genres. In addition, we found that genres do have compounding effect on classification of regional varieties. In particular, reportage in two varieties is more likely to be classified than review, corroborating the complex system view of language variations. That is, language variations and changes typically do not take place evenly across the board for the complete language system. This further enhances our hypothesis that dynamic complex system features, such as the power functions captured by the Menzerath–Altmann law, provide effective models in classifications of similar languages.

分類語彙表番号を付与した『現代日本語書き言葉均衡コーパス』の書籍・新聞・雑誌データ

分類語彙表番号を付与した『現代日本語書き言葉均衡コーパス』の書籍・新聞・雑誌データ

Confusion analysis in phoneme based speech recognition in Hindi

Phoneme recognition is an essential step in the development of a speech recognition system (SRS), as phonemes are fundamental building blocks in a spoken language. This research work aimed to present phoneme recognition with systematic confusion analysis for the Hindi language. The accuracy of phoneme recognition is the foundation for developing an efficient SRS. Therefore, the systematic confusion analysis for phoneme recognition is essential to improve speech recognition performance. Experiments conducted on Continuous Hindi speech corpus for phoneme recognition with speaker-dependent mode using Hidden Markov Model (HMM) based tool kit HTK. Feature extraction technique Perceptual Linear Predictive Coefficient (PLP) was used with five states Monophones HMM model. Tests were performed for exploring the recognition of Hindi vowels and consonants. Confusion matrices were presented for both vowels and consonants with analysis and possible solutions. During systematic analysis, the vowels were divided into front, middle, and back vowels while consonants were categorized based on place of articulation and manner of articulation. Research findings show that some Hindi phonemes have significant effects on speech recognition. The investigations also reveal that some Hindi phonemes are mostly confused, and some phonemes have more deletions and insertions. The research further demonstrates that the words made of less number of phonemes show more insertion errors. It was also found that most of the Hindi sentences end with some specific words. These particular words can be used to reduce the search place in language modeling for improving speech recognition. The research findings can be utilized to enhance the performance of the speech recognition system by selecting suitable feature extraction techniques and classification techniques for phonemes. The outcome of the research can also be used to develop improved pronunciation dictionaries and designing the text for developing phonetically balanced speech corpus for improvement in speech recognition. Experimental results show an average corrected recognition score of 70% for vowel class and consonant categories, the maximum average corrected recognition score of 94% was obtained with palatal sounds, and the lowest average corrected recognition score of 54% was achieved with liquid sounds. The comparative analysis of the presented work was made to similar existing works.