Translation Output Research Articles

Machine Translation and Parallel Corpora: The Case of English-to-Chinese Translation of Relative Clauses

This paper compares translations done by professional human translators (as manifested in the Babel English-Chinese Parallel Corpus) and machine translations produced by Google Translate (with English as the source language and Simplified Chinese as the target language). Capitalizing on the branching direction of these two languages, we investigated translation strategies of restrictive relative clauses by focusing on which-relatives. The specific objectives of the study were to (a) extract relevant data from a parallel corpus; (b) assess machine translation quality; and (c) highlight the similarities and dissimilarities between corpus outputs and machine translation outputs. Of the 147 test materials, 115 Google Translate outputs (78.2%) were rated as ‘successful’ or ‘acceptable’. No significant differences were found between the degree of success in machine translations and linguistic factors (e.g., the active or passive voice in relative clauses). This finding confirms that linguistic knowledge is not required when using statistical machine translation (SMT) such as Google Translate. Another noteworthy finding was that Google Translate did not use the least frequently occurring translation strategy in the parallel corpus. This is not surprising given that SMT systems greatly rely on parallel corpora for training statistical translation models.

Assessment of Machine Translation Output: A comparative study between Human and Automatic Models

Assessment of Machine Translation Output: A comparative study between Human and Automatic Models

Machine Translation Using Semantic Web Technologies: A Survey

A large number of machine translation approaches have recently been developed to facilitate the fluid migration of content across languages. However, the literature suggests that many obstacles must still be dealt with to achieve better automatic translations. One of these obstacles is lexical and syntactic ambiguity. A promising way of overcoming this problem is using Semantic Web technologies. This article presents the results of a systematic review of machine translation approaches that rely on Semantic Web technologies for translating texts. Overall, our survey suggests that while Semantic Web technologies can enhance the quality of machine translation outputs for various problems, the combination of both is still in its infancy.

Making Sense of Foreign Language Posts in Social Media

Many people's social media feeds include posts in languages they do not understand. While previous research has examined bilingual social media users' language choices, little research has focused on how people make sense of foreign language posts. In the present study, we interviewed 23 undergraduate social media users about how they consume and make sense of posts in other languages. Interviewees reported that they often did not pay attention to or engage with foreign language posts, due to a lack of relevance and contextual knowledge. When they did actively engage with foreign language posts, interviewees did not rely solely on machine translation output but instead actively collected and combined various cues from within and outside the post in order to understand what it was about. Interviewees further reported different types of goals for trying to make sense of foreign language posts; some focused on simply extracting and understanding the emotional components of a post while others tried to gain a fuller understanding of a post, including its contextual and cultural meanings. Based on these findings, we suggest design possibilities that could better aid multilingual communication in social media.

Neural versus phrase-based MT quality: An in-depth analysis on English–German and English–French

Within the field of statistical machine translation, the neural approach (NMT) is currently pushing ahead the state of the art performance traditionally achieved by phrase-based approaches (PBMT), and is rapidly becoming the dominant technology in machine translation. Indeed, in the last IWSLT and WMT evaluation campaigns on machine translation, NMT outperformed well established state-of-the-art PBMT systems on many different language pairs. To understand in what respects NMT provides better translation quality than PBMT, we perform a detailed analysis of neural versus phrase-based statistical machine translation outputs, leveraging high quality post-edits performed by professional translators on the IWSLT data. In this analysis, we focus on two language directions with different characteristics: English–German, known to be particularly hard because of morphology and syntactic differences, and English–French, where PBMT systems typically reach outstanding quality and thus represent a strong competitor for NMT. Our analysis provides useful insights on what linguistic phenomena are best modelled by neural models – such as the reordering of verbs and nouns – while pointing out other aspects that remain to be improved – like the correct translation of proper nouns.

Asymmetry and automaticity in translation

Abstract This article adopts an interdisciplinary approach to integrate insights from cognitive psychology and Cognitive Linguistics into translational inquiry by modeling the translator’s operations as alternating between System 1 and System 2 thinking. We analyze trainee output to investigate translatorial decision-making in scenarios employing instances of basic cross-linguistic asymmetry in the partitioning of conceptual material. The objective is to better understand the role that automaticity plays in those contexts: how pronounced it is, how it influences translation output, and to what extent the skill of monitoring it can be regarded a component of translation competence. The study also tests, in one condition, whether the translator trainer can induce System 2 processing in trainees by issuing extra pre-task instruction, thereby helping optimize their performance.

What Do Professional Translators Do when Post-Editing for the First Time? First Insight into the Spanish-Basque Language Pair

Machine translation post-editing is becoming commonplace and professional translators are often faced with this unknown task with little training and support. Given the different translation processes involved during post-editing, research suggests that untrained translators do not necessarily make good post-editors. Besides, the post-editing activity will be largely influenced by numerous aspects related to the technology and texts used. Training material, therefore, will need to be tailored to the particular conditions under which post-editing is bound to happen. In this work, we provide a first attempt to uncover what activity professional translators carry out when working from Spanish into Basque. Our initial analysis reveals that when working with moderate machine translation output post-editing shifts from the task of identifying and fixing errors, to that of “patchwork” where post-editors identify the machine translated elements to reuse and connect them using their own contributions. Data also reveal that they primarily focus on correcting machine translation errors but often fail to restrain themselves from editing correct structures. Both findings have clear implications for training and are a step forward in tailoring sessions specifically for language combinations of moderate quality.

What Do Professional Translators Do when Post-editing for the First Time? First Insight into the Spanish-Basque Language Pair

Machine translation post-editing is becoming commonplace and professional translators are often faced with this unknown task with little training and support. Given the different translation processes involved during post-editing, research suggests that untrained translators do not necessarily make good post-editors. Besides, the post-editing activity will be largely influenced by numerous aspects related to the technology and texts used. Training material, therefore, will need to be tailored to the particular conditions under which post-editing is bound to happen. In this work, we provide a first attempt to uncover what activity professional translators carry out when working from Spanish into Basque. Our initial analysis reveals that when working with moderate machine translation output post-editing shifts from the task of identifying and fixing errors, to that of “patchwork” where post-editors identify the machine translated elements to reuse and connect them using their own contributions. Data also reveal that they primarily focus on correcting machine translation errors but often fail to restrain themselves from editing correct structures. Both findings have clear implications for training and are a step forward in tailoring sessions specifically for language combinations of moderate quality.

Visualizing Neural Machine Translation Attention and Confidence

Abstract In this article, we describe a tool for visualizing the output and attention weights of neural machine translation systems and for estimating confidence about the output based on the attention. Our aim is to help researchers and developers better understand the behaviour of their NMT systems without the need for any reference translations. Our tool includes command line and web-based interfaces that allow to systematically evaluate translation outputs from various engines and experiments. We also present a web demo of our tool with examples of good and bad translations: http://ej.uz/nmt-attention.

Predictor-Estimator

Recently, quality estimation has been attracting increasing interest from machine translation researchers, aiming at finding a good estimator for the “quality” of machine translation output. The common approach for quality estimation is to treat the problem as a supervised regression/classification task using a quality-annotated noisy parallel corpus, called quality estimation data , as training data. However, the available size of quality estimation data remains small, due to the too-expensive cost of creating such data. In addition, most conventional quality estimation approaches rely on manually designed features to model nonlinear relationships between feature vectors and corresponding quality labels. To overcome these problems, this article proposes a novel neural network architecture for quality estimation task—called the predictor-estimator —that considers word prediction as an additional pre-task. The major component of the proposed neural architecture is a word prediction model based on a modified neural machine translation model—a probabilistic model for predicting a target word conditioned on all the other source and target contexts. The underlying assumption is that the word prediction model is highly related to quality estimation models and is therefore able to transfer useful knowledge to quality estimation tasks. Our proposed quality estimation method sequentially trains the following two types of neural models: (1) Predictor : a neural word prediction model trained from parallel corpora and (2) Estimator : a neural quality estimation model trained from quality estimation data. To transfer word a prediction task to a quality estimation task, we generate quality estimation feature vectors from the word prediction model and feed them into the quality estimation model. The experimental results on WMT15 and 16 quality estimation datasets show that our proposed method has great potential in the various sub-challenges.

Identifying the Machine Translation Error Types with the Greatest Impact on Post-editing Effort.

Translation Environment Tools make translators’ work easier by providing them with term lists, translation memories and machine translation output. Ideally, such tools automatically predict whether it is more effortful to post-edit than to translate from scratch, and determine whether or not to provide translators with machine translation output. Current machine translation quality estimation systems heavily rely on automatic metrics, even though they do not accurately capture actual post-editing effort. In addition, these systems do not take translator experience into account, even though novices’ translation processes are different from those of professional translators. In this paper, we report on the impact of machine translation errors on various types of post-editing effort indicators, for professional translators as well as student translators. We compare the impact of MT quality on a product effort indicator (HTER) with that on various process effort indicators. The translation and post-editing process of student translators and professional translators was logged with a combination of keystroke logging and eye-tracking, and the MT output was analyzed with a fine-grained translation quality assessment approach. We find that most post-editing effort indicators (product as well as process) are influenced by machine translation quality, but that different error types affect different post-editing effort indicators, confirming that a more fine-grained MT quality analysis is needed to correctly estimate actual post-editing effort. Coherence, meaning shifts, and structural issues are shown to be good indicators of post-editing effort. The additional impact of experience on these interactions between MT quality and post-editing effort is smaller than expected.

Issues in Machine Translation

AbstractMachine translation (MT) is still a huge challenge for both IT developers and users. From the beginning of machine translation, problems at the syntactic and semantic levels have been faced. Today despite progress in the development of MT, its systems still fail to recognise which synonym, collocation or word meaning should be used. Although mobile apps are very popular among users, errors in their translation output create misunderstandings. The paper deals with the analysis of machine translation of general everyday language in Lithuanian to English and English to Lithuanian language pairs. The results of the analysis show that more than two thirds of all the sentences were translated incorrectly, which means that there is a relatively small possibility that a mobile app will translate sentences correctly. The results are disappointing, because even after almost 70 years of MT research and improvement, researchers still cannot offer a system that would be able to translate with at least 50% correctness.

The processing of German modal particles in post-editing tasks

German Modal Particles (MPs) are linguistic elements that pose severe difficulties for the translation and post-editing processes. One of the major reasons for such hindrance is the context-dependence function of the MP. This work is an effort to collect relevant information about the translation and decision-making processes in the German/Portuguese language pair. It presents the results of an experimental study on post-editing the MPs doch and wohl into Brazilian Portuguese. In order to empirically investigate the cognitive effort required to postedit machine translated modal particles, three research instruments were used, namely: the software Translog-II; the eye tracker Tobii T60; a prospective questionnaire (free and guided). In the experiment, twenty Brazilian participants were asked to post-edit five Portuguese machine translation outputs with MP’s doch and wohl. The results here presented tend to confirm the hypothesis by GUTT (1998) and the study on processing analysis performed by ALVES (2007), which show that the relation between effort and effect does not obey a linear relation among themselves. Therefore, the analysis of how modal particles are processed in post-editing tasks tends to show that different cognitive environments imply distinct allocation of the minimum cognitive effort needed to achieve a relevant contextual effect

MTOR Inhibition Mitigates Enhanced mRNA Translation Associated with the Metastatic Phenotype of Osteosarcoma Cells In Vivo.

To successfully metastasize, tumor cells must respond appropriately to biological stressors encountered during metastatic progression. We sought to test the hypothesis that enhanced efficiency of mRNA translation during periods of metastatic stress is required for metastatic competence of osteosarcoma and that this metastasis-specific adaptation is amenable to therapeutic intervention. We employ novel reporter and proteomic systems that enable tracking of mRNA translation efficiency and output in metastatic osteosarcoma cells as they colonize the lungs. We test the potential to target mRNA translation as an antimetastatic therapeutic strategy through pharmacokinetic studies and preclinical assessment of the prototypic mTOR inhibitor, rapamycin, across multiple models of metastasis. Metastatic osteosarcoma cells translate mRNA more efficiently than nonmetastatic cells during critical stressful periods of metastatic colonization of the lung. Rapamycin inhibits translational output during periods of metastatic stress, mitigates lung colonization, and prolongs survival. mTOR-inhibiting exposures of rapamycin are achievable in mice using treatment schedules that correspond to human doses well below the MTDs defined in human patients, and as such are very likely to be tolerated over long exposures alone and in combination with other agents. Metastatic competence of osteosarcoma cells is dependent on efficient mRNA translation during stressful periods of metastatic progression, and the mTOR inhibitor, rapamycin, can mitigate this translation and inhibit metastasis in vivo Our data suggest that mTOR pathway inhibitors should be reconsidered in the clinic using rationally designed dosing schedules and clinical metrics related to metastatic progression. Clin Cancer Res; 22(24); 6129-41. ©2016 AACR.

The ribosome-engaged landscape of alternative splicing.

High-throughput RNA sequencing (RNA-seq) has revealed an enormous complexity of alternative splicing (AS) across diverse cell and tissue types. However, it is currently unknown to what extent repertoires of splice-variant transcripts are translated into protein products. Here, we surveyed AS events engaged by the ribosome. Notably, at least 75% of human exon-skipping events detected in transcripts with medium-to-high abundance in RNA-seq data were also detected in ribosome profiling data. Furthermore, relatively small subsets of functionally related splice variants are engaged by ribosomes at levels that do not reflect their absolute abundance, thus indicating a role for AS in modulating translational output. This mode of regulation is associated with control of the mammalian cell cycle. Our results thus suggest that a major fraction of splice variants is translated and that specific cellular functions including cell-cycle control are subject to AS-dependent modulation of translation output.

Machine Translation and Automatic Post-Editing in Translation of Business Letters and Contracts

In this article the feasibility of the translation of the business letters and contracts through Google Translate as the web-based machine translation program and Grammarly and LanguageTool as the automatic post-editors and also the quality level of the final output of the machine translation and automatic post-editing in comparison with the human translation are studied. The researchers conducted a careful comparative content analysis with both a qualitative and a quantitative design. After comparing the final output of the machine translation and post-editing with the reference materials, the researchers found that the above-mentioned texts can be translated through machine translation and post-editing only under the supervision of a professional translator, and testing the null hypothesis by using t-test indicated that there was a significant difference between this kind of translation and human translation based on the data obtained from this research.

Abstract SY29-01: Deciphering oncogenic translation programs

Abstract New technologies provide precise measures of translation across the transcriptome. In particular, transcriptome-scale ribosome foot printing in conjunction with deep RNA sequencing developed by N. Ignolia and colleagues has shifted the focus in the field towards the output of mRNA translation. We can now measure differential mRNA translation programs and determine RNA sequence and structural features that guide translational responses to cellular and environmental cues. This has direct implications to cancer and beyond. It also extends our understanding of gene expression from transcriptional control to the regulation of protein production. We will provide examples of how these new approaches to studying mRNA translation are leading to new insight into cancer biology and therapy. Citation Format: Hans-Guido Wendel. Deciphering oncogenic translation programs. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr SY29-01.

Opera goer or Scrabble player: what makes a good translator?

How should you select a person to carry out a translation? One approach is to request a sample translation and to evaluate it by hand. Quality Estimation addresses the problem of evaluation at least for Machine Translation output as a prediction task. This approach facilitates low-cost evaluation of MT outputs without expensive reference translations. However, the prediction of human translation in this way is difficult due to its subtlety of expression. We aimed to find out whether the qualifications, hobbies or personality traits of a person could predict their proficiency at translation. First, we gathered information about 82 participants; for each one we established the values of 146 attributes via a questionnaire. Second, we asked them to carry out some Japanese-to-English translations which we scored by hand. Third, we used the attributes as input and the translation scores as output to train the J48 decision-tree algorithm in order to predict the score of a translator from their attributes. This was then evaluated using ten-fold cross validation. When limiting to professional translators in Experiment 6, the best F-score was with Wrapper selection (0.775). The result was statistically significant ($$p < 0.05$$p<0.05). This classifier also showed the second highest Precision on Good (0.813). The second best F-score (0.737) has the highest Precision on Good (0.909), using Manual feature selection. Once again this was significant ($$p < 0.05$$p<0.05). The results suggest that certain attributes affect the prediction; in addition to experience and qualifications in translating into the target language, interest in going to the Opera, playing Scrabble or Contract Bridge, or enjoyment of cryptic crossword puzzles are important factors as well.

Exploration of Corpus Augmentation Approach for English-Hindi Bidirectional Statistical Machine Translation System

Even though lot of Statistical Machine Translation(SMT) research work is happening for English-Hindi language pair, there is no effort done to standardize the dataset. Each of the research work uses different dataset, different parameters and different number of sentences during various phases of translation resulting in varied translation output. So comparing these models, understand the result of these models, to get insight into corpus behavior for these models, regenerating the result of these research work becomes tedious. This necessitates the need for standardization of dataset and to identify the common parameter for the development of model. The main contribution of this paper is to discuss an approach to standardize the dataset and to identify the best parameter which in combination gives best performance. It also investigates a novel corpus augmentation approach to improve the translation quality of English-Hindi bidirectional statistical machine translation system. This model works well for the scarce resource without incorporating the external parallel data corpus of the underlying language. This experiment is carried out using Open Source phrase-based toolkit Moses. Indian Languages Corpora Initiative (ILCI) Hindi-English tourism corpus is used. With limited dataset, considerable improvement is achieved using the corpus augmentation approach for the English-Hindi bidirectional SMT system.

The Adequacy and Acceptability of Machine Translation in Translating the Islamic Texts

&lt;p&gt;Islamic translation is considered as a special distinguished sub-discipline of applied linguistics. It is one of the most important areas of translation because it carries the values and eternal message. Through the history, the first translation work was of religious books. This study attempts to evaluate the adequacy and acceptability of four machine translation (MT) systems (World lingo, Babylon translation, Google translate, Bing translator) in translating the Islamic texts. In addition, it aims to evaluate the Islamic translation outputs based on functional characteristics (accuracy, suitability, and well-formedness) and sub-characteristics (syntax, terminology, reliability, and fidelity). The findings indicted that Google Translate System is the most adequate and acceptable among the other three systems (World lingo, Babylon translation, Bing translator) in translating the Islamic texts. The findings also revealed that Google Translate is acceptable in producing Islamic translation outputs in regard to the following functional characteristics (accuracy, suitability, and well-formedness) and sub-characteristics (syntax, terminology, reliability and fidelity) due to Google Translate advancement.&lt;/p&gt;&lt;strong&gt;&lt;/strong&gt;