Balanced Corpus Research Articles

현대 한국어의 한자ㆍ한자어의 빈도 및 분포에 대한 계량적 분석과 그 활용

This paper is a part of the serial studies which aim to clarify the status of Chino-Korean Words in modern Korean by analyzing text corpora. The author have conversed 3M-words balanced modern Korean corpus into Chino-Koreanized corpus using text editor and python script and Korean Chino-Korean correspondance files, and the ways of creating Chino-Koreanized corpus was demonstrated in chapter 2. In chapter 2, also, it was emphasized that the ways of calculating the frequencies and investigating the distributions of Chino-Korean syllables and Chino-Korean words are to be reconsidered. In chapter 3, we showed the frequencies and distributions of Chino-Korean syllables and Chino-Korean words and clarified that the text coverages of Chino-Korean words are higher than usually thought. In chapter 4, based on the frequency data, the author proposed the lists of Chino-Korean syllables and Chino-Korean words for first step learners in Koran language education.(Yonsei University)

Analysing Lexical Diversity and Lexical Density in Korean Texts

This study is to quantitatively compare and analyze lexical richness of Korean text through actual corpus. Lexical richness measurement is mainly used to evaluate learner`s lexical ability. Lexical richness measurement mainly follows the method discussed in Read (2000) which contains lexical diversity, lexical density, lexical sophistication, number of errors and so on. Lexical diversity and lexical density, especially, can be used to understand characteristics of each text as they vary by the word type. This study centers on lexical diversity and lexical density among other measurement indexes of lexical richness. Lexical density, especially, is a quantitative measurement method that focuses on the words used in spoken and/or written texts rather than individual words. In addition, this study adopts quantitative approach to lexical richness of Korean text by using written and spoken balanced corpus composed of various genres. The contents structure of this study is as follows: first, arranging lexical diversity and lexical density measurement as well as their result previously performed on Korean languages; second, the research targets and methods; third, measurement result of lexical diversity and lexical density; fourth, measurement result of lexical density per various spoken language register.

NAMES OF THE ETHNOGRAPHIC GROUPS IN THE DICTIONARY OF CONTEMPORARY LATVIAN LANGUAGE

Determination of the ethnonymic lexical semantics can be considered as one of the aspects of ethnic studies in linguistics. Who are the Latgalians, Latgals, Suits, Selonians and other ethnographic groups, according to the modern concepts? The answer to this question can be found in the Dictionary of Contemporary Latvian Language (MLVV), which is fully available on the Internet since 2014.The aim of the paper is to analyse the names of Latvian ethnographic groups and the principles of defining meanings in MLVV, especially focusing on issues related to names of Latgalian Latvians. Lexicographic data analysis and contextual approach are applied in this research with the help of which extralinguistic factors (linguistic and situative context) are revealed.One of the most important tasks in lexicographic work is the inventory of the material, in other words, selection of word entries, illustrative texts, collection and classification of information from preceding dictionaries. In the present paper, the electronic version of the eight-volume Dictionary of the Literary Latvian Language (1972–1996) has been used for comparison. The Balanced Corpus of Modern Latvian, consisting of ~4.5 million word use cases, is applied to determine the contextual semantics and the usage frequency of the names of ethnographic groups.MLVV materials show that there is a lack of consequence in the designation of ethnographic groups and definition of their meaning. The names of ethnographic groups (as well as other ethnonyms) as entries are mostly provided in the plural, for instance, latgalieši, suiti, zemgalieši (Latgalians, Suits, inhabitants of Zemgale), etc., some of them in singular forms: kurzemnieks, vidzemnieks (inhabitant of Kurzeme, inhabitant of Vidzeme). In the previously published Dictionary of the Literary Latvian Language, the names are provided in the plural, and the data of the modern Latvian language corpus show that all the analysed lexemes are used in plural form more often.The definitions of the meanings are lacking consistency. Firstly, the scope of definitions is broad: the territorial belonging to a certain Latvian ethnographic region is taken as a basis for all the explanations of the meaning; some definitions have references to linguistic and cultural peculiarity. Secondly, none of definitions has a reference that it is an ethnographic group of Latvians; only in one case, the designation of ethnic origin latvieši (Latvians) is provided. The explanation ‘inhabitants’ (of the region) is applied most often, which indicates the belonging of the lexeme to katoikonyms, not ethnonyms; also, the explanation ‘indigenous inhabitants’ (of the region) indicates indirectly that they might be Latvians.The novelty of the dictionary is that there are two meanings for the lexemes latgaļi (Latgalians) and sēļi (Selonians): the first one characterizes the Baltic tribes, the second – inhabitants/ natives of the cultural historical region (more precisely – ethnographic group of the Latvian nation). At the same time, it is not considered that lexemes latgalieši, kurzemnieki, vidzemnieki, zemgalieši (Latgalians, inhabitants of Kurzeme, Vidzeme, Zemgale) in contemporary Latvian also have at least two meanings: the meaning of ‘ethnographic group of the nation’ in modern texts, especially in the media, is complemented by a more general meaning of ‘inhabitant of the region’, in other words, the ethnonym becomes a katoikonym.The advantage of the electronic dictionary of contemporary Latvian is the possibility to edit it constantly; therefore, it is worth to pay attention to separate lexical semantic groups when developing the dictionary. In that way, it will be possible to avoid inconsistency in definitions of meanings. In order to separate ethnonyms from katoikonyms, in the definitions of the ethnonyms analysed in the paper, the designation of ethnic community ‘ethnographic group of Latvian nation’ can be applied as the main component, supplemented by semes indicating the connection with a certain territory, and possibly, the characterization of cultural, religious and linguistic peculiarity.

TAMEEM V1.0: speakers and text independent Arabic automatic continuous speech recognizer

This research work aims to disseminate the efforts towards developing and evaluating TAMEEM V1.0, which is a state-of-the-art pure Modern Standard Arabic (MSA), automatic, continuous, speaker independent, and text independent speech recognizer using high proportion of the spoken data of the phonetically rich and balanced MSA speech corpus. The speech corpus contains speech recordings of Arabic native speakers from 11 Arab countries representing Levant, Gulf, and Africa regions of the Arabic World, which make about 45.30 h of speech data. The recordings contain about 39.28 h of 367 sentences that are considered phonetically rich and balanced, which are used for training TAMEEM V1.0 speech recognizer, and another 6.02 h of another 48 sentences that are used for testing purposes, which are mostly text independent and foreign to the training sentences. TAMEEM V1.0 speech recognizer is developed using the Carnegie Mellon University (CMU) Sphinx 3 tools in order to evaluate the speech corpus, whereby the speech engine uses three-emitting state Continuous Density Hidden Markov Model for tri-phone based acoustic models, and the language model contains uni-grams, bi-grams, and tri-grams. Using three different testing data sets, this work obtained 7.64% average Word Error Rate (WER) for speakers dependent with text independent data set. For speakers independent with text dependent data set, this work obtained 2.22% average WER, whereas 7.82% average WER is achieved for speakers independent with text independent data set.

Reconciling structure and agency in strategy - as-practice research: Towards a strong-structuration theory approach

An overwhelming focus of research on the micro agency of strategic actors has led to the literature being characterized as demonstrating a micro-myopia, resulting in a micro-isolationism. This means we know little about how the micro interrelates with the macro in strategy work. We address this problem in our conceptual article which adopts a structurationist stance to explicate how strategy-as-practice (SaP) research could be enhanced and extended by paying equal attention to both agency and structure. Specifically, we advance strong structuration theory (SST), a promising development from Giddens’ seminal work on structuration theory, to show how strategic activity can be understood as an ongoing process of structuration unfolding over time. We argue for the use of both types of methodological bracketing (context and conduct analysis), advocating systematic attention to the interplay between macro-societal and micro-local levels of analysis. Our discussion concludes with guidance for researchers inviting them to undertake empirical fieldwork that overcomes SaP’s current micro-myopia, creating a more balanced corpus of work.

Social Media – Processing Romanian Chat and Discourse Analysis

In order to obtain a balanced corpus, a sub-corpus of 2,576 sentences illustrating contemporary social media language has been added to the Dependency Treebank for Romanian. The texts were taken from the chat. The subject of this paper is to describe the second step of processing non-standard texts with a hybrid POS-tagger for Romanian and with a Malt parser, both until now trained on standard language and on other styles of communication. The results obtained show that the UAIC tools are comparable with the tools for other languages trained on similar corpora. Another purpose is to develop this resource, the Dependency Treebank for Romanian, not only quantitatively, doubling its dimension in a year, but also changing its format with a new one, compatible with other similar foreign corpora, and adding new, more complex annotation layers. A semantic layer and a discursive annotation will be added, permitting the study of discursive and conversational particularities. Finally, examples illustrating discursive particularities of the chat communication are discussed.

현대일본 문장어의 ｢男｣ 관련어휘 사용실태 ─ ｢現代日本語書き言葉均衡コーパス｣를 분석대상으로 ─

현대일본 문장어의 ｢男｣ 관련어휘 사용실태 ─ ｢現代日本語書き言葉均衡コーパス｣를 분석대상으로 ─

Improving selection of synsets from WordNet for domain-specific word sense disambiguation

Word Sense Disambiguation (WSD) is a fundamental task useful for Information Retrieval, Information Extraction, web search, and indexing, among others. In the literature there exist several works dedicated to generic WSD task, but in recent years domain-specific WSD has attracted the attention of several researchers. In this sense, this paper describes an approach for domain-specific WSD by selecting the predominant sense (synset from WordNet) of ambiguous words. To achieve it the method uses two corpora: the domain-specific test corpus (containing target ambiguous words) and a domain-specific auxiliary corpus (obtained by using relevant words from the domain-specific test corpus). The approach has four main stages: (1) auxiliary corpus generation; (2) related features extraction (from the auxiliary corpus); (3) test features extraction (from the test corpus); and (4) features integration. The proposed approach has been tested on domain-specific corpora (Sports and Finance) and on one balanced corpus, BNC. Even though our WSD approach showed some limitations when dealing with the general-domain corpus, the obtained results for domain-specific corpora, which are our main interest, were better than those reported in previous works.

How do English translations differ from non-translated English writings? A multi-feature statistical model for linguistic variation analysis

Abstract This paper discusses the debatable hypotheses of “Translation Universals”, i. e. the recurring common features of translated texts in relation to original utterances. We propose that, if translational language does have some distinctive linguistic features in contrast to non-translated writings in the same language, those differences should be statistically significant, consistently distributed and systematically co-occurring across registers and genres. Based on the balanced Corpus of Translational English (COTE) and its non-translated English counterpart, the Freiburg-LOB corpus of British English (FLOB), and by deploying a multi-feature statistical analysis on 96 lexical, syntactic and textual features, we try to pinpoint those distinctive features in translated English texts. We also propose that the stylo-statistical model developed in this study will be effective not only in analysing the translational variation of English but also be capable of clustering those variational features into a “translational” dimension which will facilitate a crosslinguistic comparison of translational languages (e. g. translational Chinese) to test the Translation Universals hypotheses.

Towards Innovative Resources for Medieval Latin

The medieval civilisation of Europe can only be investigated indirectly, by means of the diligent study of numerous traces that have survived to our times. The best source of our knowledge is still numerous surviving texts, due both to their huge quantity and surprising variety. Written mainly in Medieval Latin, within a social context that had nothing in common either with ancient or our own times, they have not benefited as they deserve from recent advances in computational linguistics or text mining. This is due, among other things, to the generally poor quality of existing resources, inadequate design of user search interfaces and unsatisfactory application of Natural Language Processing and Digital Humanities methods to the study of ancient texts. To challenge this situation we propose to build a large, representative and balanced corpus of Medieval Latin texts composed between 500 and 1500 AD all across Europe. The corpus will be annotated with PoS, lemma, time and place labels and enriched by linking it closely to a collection of dictionaries and encyclopaedias. For both textual and lexicographical resources, tools allowing efficient statistical analysis and data visualisation will be developed, aimed at revealing cultural and societal patterns that are still to be discovered from the Latin words.

Chronique bibliographique polonaise : 2012-2016

The medieval civilisation of Europe can only be investigated indirectly, by means of the diligent study of numerous traces that have survived to our times. The best source of our knowledge is still numerous surviving texts, due both to their huge quantity and surprising variety. Written mainly in Medieval Latin, within a social context that had nothing in common either with ancient or our own times, they have not benefited as they deserve from recent advances in computational linguistics or text mining. This is due, among other things, to the generally poor quality of existing resources, inadequate design of user search interfaces and unsatisfactory application of Natural Language Processing and Digital Humanities methods to the study of ancient texts. To challenge this situation we propose to build a large, representative and balanced corpus of Medieval Latin texts composed between 500 and 1500 AD all across Europe. The corpus will be annotated with PoS, lemma, time and place labels and enriched by linking it closely to a collection of dictionaries and encyclopaedias. For both textual and lexicographical resources, tools allowing efficient statistical analysis and data visualisation will be developed, aimed at revealing cultural and societal patterns that are still to be discovered from the Latin words.

Cueing musical emotions: An empirical analysis of 24-piece sets by Bach and Chopin documents parallels with emotional speech

Acoustic cues such as pitch height and timing are effective at communicating emotion in both music and speech. Numerous experiments altering musical passages have shown that higher and faster melodies generally sound “happier” than lower and slower melodies, findings consistent with corpus analyses of emotional speech. However, equivalent corpus analyses of complex time-varying cues in music are less common, due in part to the challenges of assembling an appropriate corpus. Here, we describe a novel, score-based exploration of the use of pitch height and timing in a set of “balanced” major and minor key compositions. Our analysis included all 24 Preludes and 24 Fugues from Bach’s Well-Tempered Clavier (book 1), as well as all 24 of Chopin’s Preludes for piano. These three sets are balanced with respect to both modality (major/minor) and key chroma (“A,” “B,” “C,” etc.). Consistent with predictions derived from speech, we found major-key (nominally “happy”) pieces to be two semitones higher in pitch height and 29% faster than minor-key (nominally “sad”) pieces. This demonstrates that our balanced corpus of major and minor key pieces uses low-level acoustic cues for emotion in a manner consistent with speech. A series of post hoc analyses illustrate interesting trade-offs, with sets featuring greater emphasis on timing distinctions between modalities exhibiting the least pitch distinction, and vice-versa. We discuss these findings in the broader context of speech-music research, as well as recent scholarship exploring the historical evolution of cue use in Western music.

Context-dependent Variation of LSP Collocations: A Corpus-based Analysis

This paper concerns a corpus-based study on LSP collocational variation, a phenomenon that is largely underrepresented in bilingual LSP lexicographic resources. Terminological variation on the formal (i.e. paradigmatic/syntagmatic) and semantic level is analysed in texts belonging to the technical domain (building and energy subfield). With the help of a balanced monolingual LSP corpus, variation is observed as a context-dependent phenomenon, with context interpreted as the combination of topic, genre-specific and communicative functions related to the LSP vertical dimension. A model entry from an English-Italian technical dictionary shows how the treatment of LSP collocations as autonomous terms and a corpus-based description of their variational behaviour can provide users with a means for efficiently tackling text production tasks, both in the native and in the foreign language.

On the Merger of Korean Mid Front Vowels: Phonetic and Phonological Evidence

This paper investigates the status of the merger between the mid front unrounded vowels ㅔ[e] and ㅐ[<TEX>${\varepsilon}$</TEX>] in contemporary Korean. Our analysis is based on a balanced corpus of production and perception data from young subjects from three dialectal areas (Seoul, Daegu and Gwangju). Except for expected gender differences, the production data display no difference in the realization of these vowels, in any of the dialects. The perception data, while mostly in line with the production results, show that Seoul females tend to better discriminate the two vowels in terms of perceived height: vowels with a lower F1 are more likely to be categorized as ㅔ by this group. We then investigate the possible causes of this merger: based on an empirical study of transcribed spoken Korean, we show that the pair of vowels ㅔ/ㅐ has a very low functional load. We argue that this factor, together with the phonetic similarity of the two vowels, may have been responsible for the observed merger.

현대일본 문장어의 ｢女｣ 관련어휘 사용실태 - ｢現代日本語書き言葉均衡コーパス｣를 분석대상으로 -

현대일본 문장어의 ｢女｣ 관련어휘 사용실태 - ｢現代日本語書き言葉均衡コーパス｣를 분석대상으로 -

Context Expansion for Domain-Specific Word Sense Disambiguation

Two novel techniques based on term expansion for domain-specific Word Sense Disambiguation are presented in this paper. A list of terms correlated with the ambiguous word is obtained by means of correlation methods such as Pointwise Mutual Information and Chi-square. The list of correlated terms is used for expanding the context in which an ambiguous word occurs, thus leading to a better number of terms which may be further used for determining the correct sense of the target ambiguous word. The proposed method has been tested in domain-specific corpora (Finance and Sports) and in one balanced corpus, the British National Corpus (BNC). The obtained results for domain-specific corpora are competitive (Finance) and ever better (Sports and BNC) than those reported in previous works.

연어패턴추출에 의한 유의형용사의 구별-「大きい·強い·深い」의 한정용법을 중심으로-

In terms of modifying usage of OKII, TSUYOI, HUKAI, synonymous adjective meaning , appered in balanced corpus, it can be presumed that OKII, TSUYOI, HUKAI are distinguished based on the semantic categorization of modified nouns(collocation pattern) in the rear as follows. (1) TSUYOI ①mental activities such as will, hope, trust, interest, expectation, criticism, attitude, sense./ Negative sentiments such as anxiety, worry, shock, stress. ② natural phenomena such as wind, light, smell./ properties such as power, chemical, attribute, one`s eyes, tone, stimulus. ③ personal individuality or tendency of a group. ④ relationship, influence. (2) FUKAI ①Physical properties such as vertical depth, density, place, section, tone, breath, sleep, silence, taste. ② Mental activities such as move, love, respect, interest, trust, gratitude, sadness, despair, resentment, reflection. ③ Logical reason, understanding, meaning. ④ relationship, relation, influence. (3) OKII ①Size of object, life, organization, sound, natural phenomena. ② Relative size of side, order, portion, case. ③ problem, influence.

Social Stereotypes in Communicative Formulae: Sociometric Approach

The focus of the article is centered on society through the prism of communication. Modern data extraction and information retrieval methods allow building a new vision of communicative process. The article is focused on the example of language idiom representations and omnibus survey, which help concentrate on the most stable human society ways of expression. There is also an attempt to carry out a comparative analysis of social features of the East and the West with the help of on-line national languages’ corpuses. Omnibus survey results testify to the fact that low income people are reluctant to admit the influence of idioms on their day-to-day communicative practices, while rich people stress the significance of stable communicative formulae in their life. Societies are described through their attitude to labor, expressed in the idioms with a ‘hand’ component. With the help of electronic linguistic corpuses (Corpus of the Internet and business Chinese), KOTONOHA (Balanced Corpus of Contemporary Written Japanese), BNC (British National Corpus), COCA (Corpus of Contemporary American English) the research analyzes labor stereotypes on the basis of idiom frequency indexes. In practice the results of this study can be implemented in a special socio-cultural dictionary, where the most frequent idioms are given as social stereotypes and the most powerful symbolic tools of influence and manipulation. The results of these findings are relevant to multicultural societies, migration adaptation practices and global business development. The research results have been processed into a database, marked with the Rospatent Certificate (No. 2013620397, dated 03/13/2013. Rightholder: Limited Liability Company 'Content-Press' (RU). First Author: Yakovlev Igor Gennadievich (RU). Second Author: Zavyalova Natalya Alekseevna (RU)).

The Analysis of Synonym Using Balanced Corpus of Contemporary Written Japanese : In Example of TORIKESHI, KYANSERU, KAIYAKU

The Analysis of Synonym Using Balanced Corpus of Contemporary Written Japanese : In Example of TORIKESHI, KYANSERU, KAIYAKU

DWT features performance analysis for automatic speech recognition of Urdu.

This paper presents the work on Automatic Speech Recognition of Urdu language, using a comparative analysis for Discrete Wavelets Transform (DWT) based features and Mel Frequency Cepstral Coefficients (MFCC). These features have been extracted for one hundred isolated words of Urdu, each word uttered by ten different speakers. The words have been selected from the most frequently used words of Urdu. A variety of age and dialect has been covered by using a balanced corpus approach. After extraction of features, the classification has been achieved by using Linear Discriminant Analysis. After the classification task, the confusion matrix obtained for the DWT features has been compared with the one obtained for Mel-Frequency Cepstral Coefficients based speech recognition. The framework has been trained and tested for speech data recorded under controlled environments. The experimental results are useful in determination of the optimum features for speech recognition task.