Analysing collocations in articles published in coworking blogs in Portuguese: A comparative study on authentic and translated texts

Corpus-based translation studies (CBTS) have undergone significant evolution, transitioning from descriptive methodologies to theoretical and applied approaches in recent years. However, the analysis of corpus-based research outcomes is crucial, and the absence of a unified framework often leads to less experienced researchers overlooking critical factors. This, in turn, results in varied interpretations of the same data, substantially compromising the objectivity and scientific rigor of the approach. Inspired by House’s (2014) model of translation quality assessment, Berman (2009)’s view on translation criticism, and De Sutter and Lefer (2020)’s multi-methodological, multifactorial, and interdisciplinary approach to CBTS, this study proposes a tripartite empirical-analytical framework to help researchers identify the potential factors influencing translator decision-making: textual characteristics, translator’s personal attributes, and the sociocultural context of the target language. To evaluate its utility, utilizing the mixed-effects logistic regression method, a case study is conducted to examine significant factors conditioning the reporting verb say and its Chinese translations in an English-Chinese parallel corpus of news texts, employing Appraisal Theory as the basis to determine equivalences and non-equivalences between the source language and target language. The case study shows that the framework facilitates a comprehensive analysis of the corpus findings by encompassing diverse perspectives within this scaffold. As digital technology, studies in multimodal discourse, and CBTS continue to intersect, the framework can also incorporate non-linguistic elements and AI translation tools, provided there are explicit criteria for examining translation phenomena. This framework equips researchers with a comprehensive set of perspectives, enabling them to consider as many factors as possible, thus bolstering the objectivity and scientific rigor of CBTS. The combined use of the structured framework and the multivariate analysis technique offers a holistic approach and stands as a critical advancement in CBTS by standardizing the analysis process and mitigate the subjective variability inherent in explaining translation phenomena.

Referential translation machines (RTMs) pioneer a language independent approach for predicting translation performance and to all similarity tasks with top performance in both bilingual and monolingual settings and remove the need to access any task or domain specific information or resource. RTMs achieve to become 1st in documentlevel, 4th system at sentence-level according to mean absolute error, and 4th in phrase-level prediction of translation quality in quality estimation task. 1 Referential Translation Machines Prediction of translation performance can help in estimating the effort required for correcting the translations during post-editing by human translators if needed. Referential translation machines achieve top performance in automatic and accurate prediction of machine translation performance independent of the language or domain of the prediction task. Each referential translation machine (RTM) model is a data translation prediction model between the instances in the training set and the test set and translation acts are indicators of the data transformation and translation. RTMs are powerful enough to be applicable in different domains and tasks while achieving top performance in both monolingual (Bicici and Way, 2015) and bilingual settings (Bicici et al., 2015b). Figure 1 depicts RTMs and explains the model building process (Bicici, 2016). RTMs use ParFDA (Bicici et al., 2015a) for selecting instances and interpretants, data close to the task instances for building prediction models and machine translation performance prediction system (MTPPS) (Bicici and Way, 2015) for generating features. We improve our RTM models (Bicici et Figure 1: RTM depiction: ParFDA selects interpretants close to the training and test data using parallel corpus in bilingual settings and monolingual corpus in the target language or just the monolingual target corpus in monolingual settings; an MTPPS uses interpretants and training data to generate training features and another uses interpretants and test data to generate test features in the same feature space; learning and prediction takes place taking these features as input. al., 2015b) with numeric expression identification using regular expressions and replace them with a label (Bicici, 2016). 2 RTM in the Quality Estimation Task We develop RTM models for all of the four subtasks of the quality estimation task (QET) in WMT16 (Bojar et al., 2016) (QET16), which include English to Spanish (en-es), English to German (en-de), and German to English (de-en) translation directions. The subtasks are: sentencelevel prediction (Task 1), word-level prediction (Task 2), phrase-level prediction (Task 2p), and document-level prediction (Task 3). Task 1 is about predicting HTER (human-targeted translation edit rate) (Snover et al., 2006) scores of sentence translations, Task 2 is about binary classification of word-level quality, Task 2p is about binary classification of phrase-level quality, and

Public Service Translation has for long been the ‘forgotten voice’ in PSI studies but it is arguably a valuable linguistic support for legal institutions and for training interpreters in the legal sector. Given that interpreters in the legal system in Italy often tend to ‘double-up’ as legal translators (to make a living) the line between the two is often hazy. Hybrid modalities like sight translation of legal and administrative documents is also a ‘borderline’ feature of these intertwined professions. The main aim of this paper is to describe how parallel and monolingual corpora can be used to train public service interpreters in double roles (translators, interpreters), namely by using corpora to translate, in multiple community languages. To this purpose, a computerized corpus has been constructed as a representative sample of learners’ renditions of legal texts. Then, other two corpora, monolingual and parallel corpora, have been used to verify the stumbling blocks dialogue interpreters struggle with, e.g. discourse markers and phraseological constructions. Corpus data are used descriptively (analyzing data) and prescriptively (providing examples of correct phraseological language usage in the languages at issue). In other words, I will describe how this methodology – through the collection of voice-recorded parallel corpora – is an invaluable tool in the training of legal (dialogue) interpreters. My ultimate aim is to provide concrete tools for legal interpreters and their trainers to facilitate their task primarily by constructing a multilingual parallel corpus as a resource for both academic research and PSIT practitioners.

Translation as a process of meaning making activity requires a cognitive process one of which is realized in a pause, a temporary stop or a break indicating doing other than typing activities in a certain period of translation process. Scholars agree that pauses are an indicator of cognitive process without which there will never be any translation practices. Despite such agreement, pauses are debatable as well, either in terms of their length or in terms of the activities managed by a translator while taking pauses. This study, in particular, aims at finding out how student translators and professional translators managed the pauses in a translation process. This was a descriptive research taking two student translators and two professional translators as the participants who were asked to translate a text from English into bahasa Indonesia. The source text (ST) was a historical recount text entitled ‘Early History of Yellowstone National Park’ downloaded from http://www.nezperce.com/yelpark9.html composed of 230-word long from English into bahasa Indonesia. The data were collected using Translog protocols, think aloud protocols (TAPs) and screen recording. Based on the data analysis, it was found that student translators took the longest pauses in the drafting phase spent to solve the problems related to finding out the right equivalent for the ST words or terms and to solve the difficulties encountered in encoding their ST understanding in the TL; meanwhile, professional translators took the longest pauses in the pos-drafting phase spent to ensure whether their TT had been natural and whether their TT had corresponded to the prevailing grammatical rules of the TL.

In this article we present a corpus-based statistical approach to measuring translation quality, more particularly translation acceptability, by comparing the features of translated and original texts. We discuss initial findings that aim to support and objectify formative quality assessment. To that end, we extract a multitude of linguistic and textual features from both student and professional translation corpora that consist of many different translations by several translators in two different genres (fiction, news) and in two translation directions (English to French and French to Dutch). The numerical information gathered from these corpora is exploratively analysed with Principal Component Analysis, which enables us to identify stable, language-independent linguistic and textual indicators of student translations compared to translations produced by professionals. The differences between these types of translation are subsequently tested by means of ANOVA. The results clearly indicate that the proposed methodology is indeed capable of distinguishing between student and professional translations. It is claimed that this deviant behaviour indicates an overall lower translation quality in student translations: student translations tend to score lower at the acceptability level, that is, they deviate significantly from target-language norms and conventions. In addition, the proposed methodology is capable of assessing the acceptability of an individual student’s translation – a smaller linguistic distance between a given student translation and the norm set by the professional translations correlates with higher quality. The methodology is also able to provide objective and concrete feedback about the divergent linguistic dimensions in their text.

Human Evaluation of Machine Translation is the most important aspect of improving accuracy of translation output which can be used for text categorization ahead. In this article we describe approach of text classification based on parallel corpora and natural language processing techniques. A text classifier is built on multilingual texts by translating different features of the model using the Expectation Maximization Algorithm. Cross-lingual text classification is the process of classifying text into different languages during translation by using training data. The main idea underlying this mechanism is using training data from parallel corpus and applying classification algorithms for reducing the distortion and alignment errors in Machine translation. In this chapter a Classification Model is trained which directs source language to target language on the basis of translation knowledge and parameters defined. The Algorithm adopted here is Expectation Maximization Algorithm which removes ambiguity in parallel corpora by aligning source sentence to target sentence. It considers possible translations from source to target language and selects the one that fits the model on the basis of BLEU (bilingual evaluation understudy) score. The only requirement of this learning is unlabelled data in the target language. The algorithm can be evaluated accurately by running a separate classifier on different parallel corpora. We use Monolingual Corpora and Machine Translation in our study to see the effect of both the models on our parallel corpora.

This study aims to examine University of Bahrain graduating translation students’ use of creative microstrategies in rendering into English a news text and compare it with the translation of a Bahrain News Agency professional translator. The study seeks to identify the students’ creative microstrategies based on the classification proposed by Anne Schjoldager’s‘ (2008) model of macrostrategies. The participants were 15 English language students with a minor in translation who were expected to graduate in the semester during which the study was conducted. They were required to translate a news text from Arabic to English in order to reveal the creative microstrategies used and then their performance was compared with that done by a professional translator employed by Bahrain News Agency (BNA)and published on its official website . The findings of the study have shown that translation students are inclined to focus more on the syntactic microstrategies rather than on the semantic and pragmatic ones when processing and rendering the source text into the target language. Unlike the professional translator, students' lack of negotiation with and deeper analysis of the text has deprived their performance to a certain extent from the creativity required in translation and rendered it into a mechanical exercise.

According to books of grammar, a causative form is an expression of an agent causing or forcing a person to perform an action. Translation of English causatives into Persian seems to be one of the biggest problems that Translation students and novice translators usually come across. Therefore, the purpose of this study was to investigate the translation strategies applied by the professional translator and translation trainees while translating English causatives into Persian. In this descriptive corpus-based study, the present researcher examined sixty causative constructions of novel Lord of The Flies by Gerald (1991) and their Persian translation by Mansouri (2003). In addition, twenty causative constructions from the novel were given to the twenty Translation students in order to analyze their Persian translations of causative constructions. Based on the finding, the professional translator has used Non-causative and Positive Implication strategies most frequently, whereas the students have used Auxiliary and Noncausative strategies most frequently. It can be concluded that there is a strategy behind every choice, and a reason behind every strategy, and translators should try their best to transfer all the components of a causative verb as well as possible, because each word or verb has its own value. The translator's mastery over the causative construction in the language pair explores throughout this study reminds us of a point of paramount significance. The main implication of this research may make the translators, at any level, better understand the English causative sentences and avoid producing translations that hinder communication between the translator and the readers.

This paper describes an empirical corpus analysis of English translation strategies for German non-SVO sentence beginnings in two small corpora, one of which is a corpus of 5 professional translations, the other a smaller corpus of 5 texts with 4 student versions for each text. For the analysis, two related but distinct methods of analysis were selected, one based on Rogers (2006), who investigated whether pragmatic word order or grammatical word order is given preference in English translations of German non-SVO sentences. The other is based on Firbas’s (1992) concept of functional sentence perspective as elaborated and operationalized by Doherty (1991, 2002), in order to determine whether communicative dynamism (CD) has been preserved in English translations. The analysis shows that professional translators achieve pragmatic word order in between 60–70% of cases, and maintain CD in an even higher percentage of cases (80–100%). Student translators, who are further differentiated as to English native speakers (EN), German native speakers (GN) or non-native speakers of either language (NN), vary more widely. GNs tend to achieve more pragmatic word order translations than ENs and even higher CD preservation, but there is also greater variety between different ENs. On average, functional sentence perspective is less well preserved overall in student translations than in professional translations.

The translation industry has witnessed rapid technological improvements in recent years. This rapid improvement is ascribed to a huge demand for the workload. Using a computer in the field of translation is very important due to the huge demand for fast and accurate translation. Translation tools came to existence due to the low proficiency of machine translation. CAT tools have become essential for many institutions, companies, and organizations. CAT tools increase productivity and minimize translation costs. The purpose of this study is to scrutinise the attitudes of professional translators and translation students towards CAT tools in Yemen. The questionnaire of this study was composed of 27 statements distributed to four constructs. The link of the questionnaire was distributed via WhatsApp to the participants. The other tool was an online interview about some issues related to CAT tools. These responses were analysed qualitatively. The researchers distributed the questionnaire to 250 people, and the valid responses of the participants were 56. The analysis shows that they have positive attitudes towards CAT tools. The results of the study show that professional translators and translation students show a positive attitude. Unexpectedly, the profiles of the participants do not play any role in their attitudes towards CAT tools.

Consistent demand for highly professional translators determines continuous attempts of researchers and programmers to develop and propose reliable tools for both improvement of translation quality and facilitation of translators’ work. Last ten years have brought the parallel and comparable corpora into the focus of Ukrainian scientists’ attention. The aim of the paper is to specify the prerequisites for building the English-Ukrainian parallel corpus and describe its application in Translation Studies. A parallel corpus as a separate type of linguistic corpora cannot be built without alignment that enables placing and extracting corresponding sentences/paragraphs of source and target texts in one space. To create parallel corpora, it is necessary to perform additional text preparation. The Sketch Engine system (an example of a web-oriented system for work with corpora) can offer the solution for annotation with Excel. However, Sketch Engine lacks artificial intelligence techniques for further word processing. There is probability that employment of a neural network in the future will enable text alignment in parallel corpora instead of system users. Data from parallel corpora can be used in translation lexicography, comparative lexico-grammatical works, studies in the theory and practice of translation, language teaching, and development of machine translation systems. Corpus-based translation analysis is extremely relevant to identifying translation solutions that can only be explored on the basis of translation products. It is stipulated by rather frequent absence of dictionary equivalents in most contexts and ready evidence of possible translation variants in parallel corpora that provide the usage of a language unit in a wide range of contexts.

The paper investigates the notion of Translation Units (TUs) from a cognitive angle. A TU is defined as the translator’s focus of attention at a time. Since attention can be directed towards source text (ST) understanding and/or target text (TT) production, we analyze the activity data of the translators’ eye movements and keystrokes. We describe methods to detect patterns of keystrokes (production units) and patterns of gaze fixations on the source text (fixation units) and compare translation performance of student and professional translators. Based on 24 translations from English into Danish of a 160 word text we find major differences between students and professionals: Experienced professional translators are better able to divide their attention in parallel on ST reading (comprehension) and TT production, while students operate more in an alternating mode where they either read the ST or write the TT. In contrast to what is frequently expected, our data reveals that TUs are rather coarse units as compared to the notion of ‘translation atom,’ which coincide only partially with linguistic units.

Corpus-based translation studies (CTS) is a very new technique in translation studies. In Bible translation, researchers have been collating corpora of source texts since the eighteenth century. In recent times, corpora of translated Bible texts have also been generated by a number of researchers. Thus the notion of working with corpora rather than with individual texts is not new to Bible translation. However, until very recently these corpora were examined manually, so that the introduction of computer-aided tools for corpus analysis represents a significant advancement for the Bible translator or researcher. By its very nature, the Bible lends itself remarkably well to (computerised) corpus projects. It is by definition a corpus of 66 books, and the pre-existing divisions into books, chapters and verses gives it an unusual advantage in annotation over literary works in the context of a parallel corpus. This article looks at the ways in which the Bible and its translations lend themselves to the formation of a specific corpus, the types of corpora and texts that could be considered as well as the challenges involved in producing these parallel corpora. The article then goes on to examine the potential applications of such a corpus for both the production of new Bible translations and the evaluation of existing ones. Applications to other nonreligious linguistic problems are also explored. In assessing these challenges, it becomes evident that CTS indeed redefines the nature of translation research as well as the role of the researcher. Finally, the status of existing projects to produce a parallel corpus based on Biblical texts will be reported.

A Twi-English parallel corpus is certainly an important resource for Machine Translation of Twi (ISO 639-3), a Low- Resource Language (LRL) which is mainly spoken in Ghana and Ivory Coast. Currently large-scale multidomain Twi- English parallel corpus is still unavailable partly due to the difficulties and the arduous efforts required in its design. A digital Twi lexicon curated purposely for linguistic research is also not available. In this paper, we present TWIENG – Twi English corpus, a large-scale multi-domain Twi-English parallel corpus and Twi lexicon, a digital Twi Dictionary. We discuss the data collection methodology, translation, alignment and compilation of the Twi-English parallel sentences and the technology we used to compile and host the corpus. Today’s parallel corpora are crawled from the web using web crawlers, the sentence pairs are processed, aligned, tokenized and compiled to create the corpus. We crawled English sentences from Ghanaian indigenous electronic news portals, Ghanaian Parliamentary Hansards, standard literature and also used crowdsourcing. The sentences are translated by professional translators and linguists, then aligned, tokenized and compiled. The corpus is curated using the sketch engine, a corpus manager and analysis software developed by Lexical Computing Limited. The corpus is manually evaluated by Twi professional linguists. The Corpus has 5,419 parallel sentences which were curled from local news portals, Ghana Parliament Hansard, The New Testament of the Twi Bible and through crowdsourcing via social media sites. CCS CONCEPTS • Computing Methodologies • Artificial Intelligence • Natural Language Processing

This paper surveys the strategies that the Contrastive, Typological, and Translation Mining parallel corpus traditions rely on to deal with the issue of target language representativeness of translations. On the basis of a comparison of the corpus architectures and research designs of the three traditions, we argue that they have each developed their own representativeness strategies: (i) monolingual control corpora (Contrastive tradition), (ii) limits on the scope of research questions (Typological tradition), and (iii) parallel control corpora (Translation Mining tradition). We introduce normalized pointwise mutual information (NPMI) as a bi-directional measure of cross-linguistic association, allowing for an easy comparison of the outcomes of different traditions and the impact of the monolingual and parallel control corpus representativeness strategies. We further argue that corpus size has a major impact on the reliability of the monolingual control corpus strategy and that a sequential parallel control corpus strategy is preferable for smaller corpora.