Parallel Corpus Research Articles

Consensus-Based Machine Translation for Code-Mixed Texts

Multilingualism in India is widespread due to its long history of foreign acquaintances. This leads to the presence of an audience familiar with conversing using more than one language. Additionally, due to the social media boom, the usage of multiple languages to communicate has become extensive. Hence, the need for a translation system that can serve the novice and monolingual user is the need of the hour. Such translation systems can be developed by methods such as statistical machine translation and neural machine translation, where each approach has its advantages as well as disadvantages. In addition, the parallel corpus needed to build a translation system, with code-mixed data, is not readily available. In the present work, we present two translation frameworks that can leverage the individual advantages of these pre-existing approaches by building an ensemble model that takes a consensus of the final outputs of the preceding approaches and generates the target output. The developed models were used for translating English-Bengali code-mixed data (written in Roman script) into their equivalent monolingual Bengali instances. A code-mixed to monolingual parallel corpus was also developed to train the preceding systems. Empirical results show improved BLEU and TER scores of 17.23 and 53.18 and 19.12 and 51.29, respectively, for the developed frameworks.

We need world-wide corpus-based typology: A parallel corpus study of restrictives ('only')

Language is a tool for communication in concrete use and there are several thousand languages; hence, approaches that are both corpus-linguistically and typologically informed must play an important role in linguistics. This is demonstrated in an investigation of the generally expressed meaning (GEM) 'only', considered here in translations of the New Testament. It is shown that restrictives are universal (are attested in all 121 languages of a stratified sample from 121 language families and isolates), that the generally expressed meaning (GEM) 'only' differs considerably from the parochially expressed meaning (PEM) of English only, that restriction plays an important role in discourse and that the use of restrictives exhibits hemispheric differences with the Pacific and the Afro-Eurasian linguistic hemispheres reflecting partly different usage patterns. It is argued that these differences are deeply rooted in discourse (parole) rather than grammar and lexicon (langue) and are so common in discourse that they percolate to written Bible translations.

Testing the unique item hypothesis with phrasal verbs in Chinese–English translations of Lu Xun’s short stories: the perspective of translation directionality

The present study revisits the unique item hypothesis (UIH) from the perspective of translation directionality in the Chinese–English(C–E) language pair. Phrasal verb (PV) is used as the linguistic feature to investigate whether UIH holds true in C–E translations and whether translation directionality plays a role in the representation of unique items, based on a self-built parallel corpus of Lu Xun’s short stories and their English translations done by two L1 and two L2 translators, and a reference corpus of BNC short stories as the non-translated reference. It is found PVs are significantly over-represented in C–E translated texts when compared with English non-translated texts, and this overrepresentation is mainly attributed to the remarkable use of PVs by L1 translators; and there is a significant difference in the use of PVs by translators of different directionality, while no significant difference is found within the same direction. Additionally, L2 translators tend to use a limited range of PVs and prefer transparent PVs to semi-transparent and opaque ones. The results falsify the UIH in general and suggest that UIH is a conditional translation tendency constrained by translation directionality, or UIH is directionality-dependent. Gravitational pull model is used to analyze and explain the divergence between different translation directions.

Development and Validation of Parallel Corpus: A Framework for Building Bikol-Filipino Linguistic Resource

Development and Validation of Parallel Corpus: A Framework for Building Bikol-Filipino Linguistic Resource

Expected sum of edge lengths in planar linearizations of trees

Dependency trees have proven to be a very successful model to represent the syntactic structure of sentences of human languages. In these structures, vertices are words and edges connect syntacticallydependent words. The tendency of these dependencies to be short has been demonstrated using random baselines for the sum of the lengths of the edges or their variants. A ubiquitous baseline is the expected sum in projective orderings (wherein edges do not cross and the root word of the sentence is not covered by any edge), that can be computed in time O(n). Here we focus on a weaker formal constraint, namely planarity. In the theoretical domain, we present a characterization of planarity that, given a sentence, yields either the number of planar permutations or an efficient algorithm to generate uniformly random planar permutations of the words. We also show the relationship between the expected sum in planar arrangements and the expected sum in projective arrangements. In the domain of applications, we derive a O(n)-time algorithm to calculate the expected value of the sum of edge lengths. We also apply this research to a parallel corpus and find that the gap between actual dependency distance and the random baseline reduces as the strength of the formal constraint on dependency structures increases, suggesting that formal constraints absorb part of the dependency distance minimization effect. Our research paves the way for replicating past research on dependency distance minimization using random planar linearizations as random baseline.

Same yet different

Abstract The article compares the distributional differences in the use of the partitive object cases in Estonian and Finnish via multifactorial modeling in contrastive research using the European Parliament parallel text corpus. Based on previous contrastive research on Finnic, we expected the principles of object case marking to be similar for Estonian and Finnish (confirmed), and the partitive objects to be more numerous in Estonian than in Finnish (not confirmed, as countable objects with scalar verbs proved less likely to be partitive in Estonian). We hypothesized that multifactorial modeling in contrastive research design could help identify the causes for variation and unfold subtle differences between related language systems. Since preferences related to grammatical voice and constituent order revealed subtle differences between the systems, this hypothesis was confirmed.

Concessive subordination in English and Norwegian

Abstract This paper investigates concessive markers, primarily subordinators, in the English-Norwegian Parallel Corpus. The initial question is how English and Norwegian concessive markers compare with regard to their frequencies and syntactic functions, both intra- and cross-linguistically. Overall, the languages are relatively similar. However, individual concessive markers differ in frequency, syntactic flexibility and possibly formality in original texts in both languages. The second research question concerns the placement of concessive clauses. Again the languages are rather similar, but there is variation within both languages, with subordinators having individual positional preferences. Finally, the translations of the subordinators are examined. Concessive markers are overrepresented in Norwegian translations and underrepresented in English translations compared to originals. While translation correspondences are often congruent, the mutual correspondence between pairs of subordinators is low. Changes made in translation suggest that Norwegian chooses coordination more often than English, thus displaying less syntactic complexity.

English complex premodifiers and their German and Swedish correspondences

Abstract This study concerns English hyphenated premodifiers (science-based targets; lower-back pain) contrasted with their German and Swedish correspondences. The data stem from the Linnaeus University English-German-Swedish corpus (LEGS), which contains non-fiction texts, but comparisons are also made to fiction texts from the English-Swedish Parallel Corpus (ESPC). The study shows that these condensed and complex premodifiers are more frequent in English originals than in English translations, and more typical of the non-fiction genre than that of fiction. Information density and terminological precision thus seem to be more important factors for the use of hyphenated premodifiers than creativity and expressiveness. In original English, two-thirds of the right-hand elements are either nouns or ed-participles. In translated English, numerals as left-hand elements (three-page document) are less frequent than in original English. Regarding German and Swedish correspondences, around half are premodifiers. Postmodifiers in the form of prepositional phrases and relative clauses are more frequent in Swedish than in German, which instead “overuses” premodifying extended attributes. Compound adjectives/participles and compound nouns are the most frequent correspondences in both German and Swedish. In almost half the instances, German and Swedish translators choose the same correspondents, indicating a high degree of similarity in the structural preferences in the two target languages.

Simple and complex help constructions in English and Norwegian

Abstract This paper reports on a study of verbal help constructions in English and Norwegian. It is based on data from the English–Norwegian Parallel Corpus, and discusses 11 constructions in all, nine of which have a close parallel in the other language. The constructions vary in syntactic complexity from the simple intransitive, on the one-hand, to complex-transitives containing infinitive complements, on the other. The hypothesis is advanced that the simpler the basic syntactic structure of a construction, the more likely it is to be translated by a construction with a similar syntax. This hypothesis receives no support from the data. On the contrary, it is more complex constructions, containing an explicit helpee, that are more likely to be translated by a syntactically similar construction.

DSISA: A New Neural Machine Translation Combining Dependency Weight and Neighbors

Most of the previous neural machine translations (NMT) rely on parallel corpus. Integrating explicitly prior syntactic structure information can improve the neural machine translation. In this article, we propose a Syntax Induced Self-Attention (SISA) which explores the influence of dependence relation between words through the attention mechanism and fine-tunes the attention allocation of the sentence through the obtained dependency weight. We present a new model, Double Syntax Induced Self-Attention (DSISA), which fuses the features extracted by SISA and a compact convolution neural network (CNN). SISA can alleviate long dependency in sentence, while CNN captures the limited context based on neighbors. DSISA utilizes two different neural networks to extract different features for richer semantic representation and replaces the first layer of Transformer encoder. DSISA not only makes use of the global feature of tokens in sentences but also the local feature formed with adjacent tokens. Finally, we perform simulation experiments that verify the performance of the new model on standard corpora.

A comparative analysis of lexical-based automatic evaluation metrics for different Indic language pairs

<p>With the rise of machine translation systems, it has become essential to evaluate the quality of translations produced by these systems. However, the existing evaluation metrics designed for English and other European languages may not always be suitable or apply to other Indic languages due to their complex morphology and syntax. Machine translation evaluation (MTE) is a process of assessing the quality and accuracy of the machine-translated text. MTE involves comparing the machine-translated output with the reference translation to calculate the level of similarity and correctness. Therefore, this study evaluates different metrics, namely, BLEU, METEOR, and TER to identify the most suitable evaluation metric for Indic languages. The study uses datasets for Indic languages and evaluates the metrics on various translation systems. The study contributes to the field of MT by providing insights into suitable evaluation metrics for Indic languages. This research paper aims to study and compare several lexical automatic machine translation evaluation metrics for Indic languages. For this research analysis, we have selected five language pairs of parallel corpora from the low-resource domain, such as English–Hindi, English-Punjabi, English-Gujarati, English-Marathi, and English-Bengali. All these languages belong to the Indo-Aryan language family and are resource-poor. A comparison of the state of art MT is presented and shows which translator works better on these language pairs. For this research work, the natural language toolkit tokenizers are used to assess the analysis of the experimental results. These results have been performed by taking two different datasets for all these language pairs using fully automatic MT evaluation metrics. The research study explores the effectiveness of these metrics in assessing the quality of machine translations between various Indic languages. Additionally, this dataset and analysis will make it easier to do future research in Indian MT evaluation.</p>

Building a Parallel Corpus for Chinese Folk Songs Translation Studies: A Case Study of Northern Shaanxi and Hua’er Folk Songs

Folk songs, collaboratively created by the public and transmitted orally, have gained widespread popularity. The translation of folk songs primarily centers on lyrics translation, a subset of literary translation. Recent advancements in corpus technology have highlighted the significance of corpus-based research approaches for the analysis of literary translation. The corpus method, now employed as a hybrid research approach, enables the generation of quantitative data for descriptive translation studies. Scholars are increasingly using parallel corpora containing both the source text (ST) and the target text (TT) to explore translation universals across diverse texts. Despite the growing body of literature on the translation of Chinese folk songs, most studies have involved straightforward analyses of a limited number of translated texts without the utilization of quantitative approaches. This article aims to bridge this gap by presenting a prospective study on the creation of the Chinese-English Parallel Corpus of Northern Shaanxi and Hua’er Folk Songs (CEPCNSHFS). The study covers essential aspects such as sampling, corpus structure, corpora selection, and corpora processing. Moreover, to assess the practical utility of the CEPCNSHFS, a pilot study was conducted. The primary contributions of this article reside in the potential of the CEPCNSHFS to support diverse research topics, including the exploration of translation language characteristics, styles, and methods employed in translating Northern Shaanxi and Hua’er folk songs, both of which hold significant positions within Chinese folk song traditions.

A Multidimensional Analysis of the Discourse Features of the English Translation of Heavy Wings

Ideology and translation go hand in hand. Translation is a rewriting of the source text, reflecting the ideology and poetics of a particular society. This paper establishes a Chinese-English parallel corpus based on two English translations of Heavy Wings, which takes ideology as the object of study, using multidimensional analysis to examine register differences and to compare the represented features in ideological markers between the two translations in the four dimensions, namely nominalization, transitivity, modality and coherence. Based on system functional linguistics, this paper examines the relationship between ideological markers and the process of the event, the participants and the environment. It was found that: (1) There is a great significance in register of two translations, with 31 out of 67 lexico-grammatical features. Gladys Yang’s translation shows a higher degree of abstract information and explicit; (2) Gladys Yang’s translation mainly focus on the reform process and participants, deliberately distancing the two parties from each other, while Goldblatt’s translation tends to attribute the factional conflicts in the reform to various environmental components, opening up more evaluation space for the textual content; (3) The differences in the use of linguistic and ideological markers are affected by both intra-linguistic and extra-linguistic factors, namely source text factors, social context, ideology, translator subjectivity, etc.

HYBRID DISTANCE-STATISTICAL-BASED PHRASE ALIGNMENT FOR ANALYZING PARALLEL TEXTS IN STANDARD MALAY AND MALAY DIALECTS

Parallel texts corpora are essential resources in linguistics and natural language processing, especially in translation and multilingual information retrieval. The publicly available parallel text corpora are limited to certain genres, types and domains. Furthermore, the parallel dialect text is scarce, even though they are important in the analysis and study of a dialect. Collecting parallel dialect text is challenging because dialects typically appear in the form of speech and very limited dialectic texts exist. Moreover, there is no standard orthography in most dialects. The contributions of this paper are threefold. First, the paper describes a methodology in acquiring a parallel text corpus of Standard Malay and Malay dialects, particularly Kelantan Malay and Sarawak Malay. Second, we propose a hybrid of distance based and statistical-based alignment algorithm to align words and phrases the parallel text. The results show that the precision and recall values of the proposed alignment algorithm are more than 95% and better than the state-of the-art GIZA++. Third, the alignment obtained were compared to find out the lexical similarities and differences between Standard Malay and the two studied Malay dialects, contributing valuable insights into the linguistic variations within the Malay language family.

Bilingualism and Translation: A Teaching Proposal

In Italy the teaching of translation as a specific skill for bilingual speakers is normally not included in the programs of Russian special schools, the so-called subbotnie školy, where the development of language skills (especially listening, writing and grammar accuracy) in heritage speakers generally takes top priority. As already pointed out in other works (Perotto 2020, 2022), the activity of translation exerts an important influence on the cognitive development of the child, especially if bilingual. It helps him/her not only to differentiate and develop two parallel linguistic and cultural systems, but to approach them with a greater metalinguistic and cultural awareness. On the basis of translations made by bilingual participants to the international Cultural Bridge competition, Russian-Italian an Russian-English parallel corpora were created for the analysis of translation difficulties of heritage speakers living in Italy and USA. These difficulties will be analyzed separately or comparatively on the Sketch engine platform, providing some basic indications for a more targeted teaching of translation. The use of corpora can be an important teaching tool to improve heritage speakers’ linguistic competence and translation awareness.

Epistemic modality in English and Thai medical abstracts: a corpus-based study

ABSTRACT Epistemicity is roughly understood as the linguistic expression of a speaker’s commitment to what they say or write. Against a backdrop of scant knowledge about how the phenomenon works in many non-European languages, this study seeks to explore epistemicity in translated and original Thai. Two specialized corpora were built. The first, a parallel corpus of medical abstracts translated from English into Thai (about 100,000 English and 150,000 Thai words), was used to investigate translated Thai and its relationship with the English source texts. The second, a 250,000-word monolingual corpus of medical abstracts originally written in Thai, was used to compare translated and non-translated texts. The distribution and strength of epistemic modal expressions in translated Thai abstracts were found to follow closely patterns established in the English source texts. They differed significantly from patterns observed in original Thai, however, leading to the tentative conclusion that translated Thai could impart less certainty on the writers’ side than would normally be perceived by readers of medical abstracts originally written in Thai.

The development of Japanese-English and English-Japanese parallel corpus, including search system and its utilisation

Parallel corpora play a central role in translation studies, contrastive linguistics, bilingual lexicography and language teaching/learning, facilitating the comparison of linguistic features across languages. But it is challenging to construct a parallel corpus and there are few of them. Also, they can quickly become outdated and unusable. Professor Yasunori Nishina, Faculty of Global Communication, Kobe Gakuin University, Japan, is an expert in applied translation, corpus linguistics and applied linguistics working to develop a website for searching Japanese-English and English-Japanese parallel corpora. He is revitalising an existing unusable, unsearchable corpus with assistance from Lago NLP. He has a keen focus on design as he wants to ensure diversity of genres and translation directions and, in a world first, is incorporating the ability to analyse parallel corpora. The tool that Nishina and Lago NLP are developing with support from the JSPS KAKENHI grant number 20K00692 (and 23K00599 from 2023) is called Parallel Link. Nishina will also collaborate with a researcher at the University of Oxford, UK, where experts who developed the Japanese corpus online search tool are based. The University has established a collaborative relationship with the National Institute for Japanese Language and Linguistics (NINJAL), which is the centre for Japanese language research, and has carried out joint projects of the Oxford NINJAL Corpus of Old Japanese (ONCOJ). Nishina has completed the development of version 1.20 of Parallel Link and intends to develop it to version 2.0 in the coming years.

The diachronic evolution of artistic terminology in translation. Building a parallel corpus of Giorgio Vasari’s Le Vite

The diachronic evolution of artistic terminology in translation. Building a parallel corpus of Giorgio Vasari’s Le Vite

Exploring Web-Based Translation Resources Applied to Hindi-English Cross-Lingual Information Retrieval

Internet users perceive a multilingual web but are unfamiliar with it due to communication in their regional language called Cross-Lingual Information Retrieval (CLIR). In CLIR, a translation technique is used to translate the user queries into the target document’s language. Conventional translation techniques are based on either a manual dictionary or a parallel corpus, whereas the trending Statistical Machine Translation (SMT) and Neural Machine Translation (NMT) techniques are trained on a parallel corpus. NMT is not so mature for Hindi-English translation, according to the literature, and SMT performs better than the NMT. SMT provides a static translation due to the limited vocabularies in the available parallel corpus. It may not provide the translations for missing or unseen words, whereas the web provides a dynamic interface where multiple users are updating information at the same time. The web may provide the translations for missing or unseen words, and therefore the web is effectively used for technically developed languages like English, German, Spanish, Russian, and Chinese. In this article, different web resources such as Wikipedia, Hindi WordNet and Indo WordNet, ConceptNet, and online dictionary based translation techniques are proposed and applied to Hindi-English CLIR. Wikipedia-based translation approach incorporates three modules—exactly matched, partially matched, and disambiguation—to address the issues of wrong inter-wiki links, partially matched terms, and ambiguous articles. Hindi WordNet and Indo WorNet attribute “English synset” and ConceptNet attributes “Related term” & “Synonymy” are used for obtaining translations. Further, WordNet path similarity is used to disambiguate translations. Various online dictionaries are available that return multiple relevant and irrelevant translations. The proposed approaches are compared to the SMT where the Wikipedia-based approach achieves approximately similar mean average precision to SMT.

A Corpus-based Study on China's Image in Emotions of Daodejing's English Translations from the Perspective of Communication Studies

Grounded in the Chinese-English parallel corpus of Daodejing, this study delves into the intricate interplay between emotional expression, communication effects, and the nuanced portrayal of China, unveiling the subtle impacts of translation in presenting the Chinese image. Through the establishment and analysis of the corpus, the research provides insights into the evolution of emotional elements in multiple translation versions spanning from 1868 to 2009. Quantitative analysis utilizing Pearson correlation coefficients reveals weak correlations between emotional expression and communication effects, both outbound and inbound . Despite a lack of statistical significance, qualitative analysis sheds light on valuable insights. In outbound communication, the nuanced handling of emotional expression, a weaker emphasis on positive emotions, and a deliberate avoidance of excessive negative emotions signify efforts to challenge traditional perceptions and portray a more complex image. In inbound communication, a tendency toward neutral emotional expression aims to offer Western readers a genuine and comprehensive understanding of China. Motivational analysis highlights the influence of cultural differences, historical contexts, reader expectations, and global dynamics on the translator's choices in emotional expression. Cultural nuances play a crucial role in shaping the Chinese image, emphasizing the translator's importance in flexibly navigating cross-cultural communication. In the era of globalization, this study aims to underscore the growing awareness among translators of the necessity to present cultural commonalities in emotional expression, fostering a broader understanding and acceptance of Chinese philosophy and values on the global stage.