2020
|
Bizzoni, Yuri; Juzek, Tom S; España-Bonet, Cristina; Chowdhury, Koel Dutta; van Genabith, Josef; Teich, Elke How Human is Machine Translationese? Comparing Human and Machine Translations of Text and Speech Inproceedings The 17th International Workshop on Spoken Language Translation, Seattle, WA, United States, 2020. Abstract | Links | BibTeX @inproceedings{Bizzoni2020,
title = {How Human is Machine Translationese? Comparing Human and Machine Translations of Text and Speech},
author = {Yuri Bizzoni and Tom S. Juzek and Cristina Espa\~{n}a-Bonet and Koel Dutta Chowdhury and Josef van Genabith and Elke Teich
},
url = {http://www.sfb1102.uni-saarland.de/wp/wp-content/uploads/2020/06/IWSLT-b1-B7-final2020.pdf
http://iwslt.org/doku.php},
year = {2020},
date = {2020-07-00},
booktitle = {The 17th International Workshop on Spoken Language Translation},
address = {Seattle, WA, United States},
abstract = {Translationese is a phenomenon present in human translations, simultaneous interpreting, and even machine translations. Some translationese features tend to appear in simultaneous interpreting with higher frequency than in human text translation, but the reasons for this are unclear. This study analyzes translationese patterns in translation, interpreting, and machine translation outputs in order to explore possible reasons. In our analysis we (i) detail two non-invasive ways of detecting translationese and (ii) compare translationese across human and machine translations from text and speech. We find that machine translation shows traces of translationese, but does not reproduce the patterns found in human translation, offering support to the hypothesis that such patterns are due to the model (human vs. machine) rather than to the data (written vs. spoken).},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Translationese is a phenomenon present in human translations, simultaneous interpreting, and even machine translations. Some translationese features tend to appear in simultaneous interpreting with higher frequency than in human text translation, but the reasons for this are unclear. This study analyzes translationese patterns in translation, interpreting, and machine translation outputs in order to explore possible reasons. In our analysis we (i) detail two non-invasive ways of detecting translationese and (ii) compare translationese across human and machine translations from text and speech. We find that machine translation shows traces of translationese, but does not reproduce the patterns found in human translation, offering support to the hypothesis that such patterns are due to the model (human vs. machine) rather than to the data (written vs. spoken). |
2019
|
Bizzoni, Yuri; Teich, Elke Analyzing variation in translation through neural semantic spaces Inproceedings 2019. BibTeX @inproceedings{Bizzoni2019,
title = {Analyzing variation in translation through neural semantic spaces},
author = {Yuri Bizzoni and Elke Teich},
year = {2019},
date = {2019-08-30},
journal = {Special topic: Neural Networks for Building and Using Comparable Corpora, Recent Advances in Natural Language Processing (RANLP), Varna, Bulgaria},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
|
Karakanta, Alina; Menzel, Katrin; Przybyl, Heike; Teich, Elke Detecting linguistic variation in translated vs. interpreted texts using relative entropy Inproceedings Empirical Investigations in the Forms of Mediated Discourse at the European Parliament, Thematic Session at the 49th Poznan Linguistic Meeting (PLM2019), Poznan, 2019. BibTeX @inproceedings{Karakanta2019,
title = {Detecting linguistic variation in translated vs. interpreted texts using relative entropy},
author = {Alina Karakanta and Katrin Menzel and Heike Przybyl and Elke Teich},
year = {2019},
date = {2019-05-05},
booktitle = {Empirical Investigations in the Forms of Mediated Discourse at the European Parliament, Thematic Session at the 49th Poznan Linguistic Meeting (PLM2019), Poznan},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
|
2018
|
Collard, Camille; Przybyl, Heike; Defrancq, Bart Interpreting into an SOV Language: Memory and the Position of the Verb. A Corpus-Based Comparative Study of Interpreted and Non-mediated Speech. Journal Article Meta, 63 (3), pp. 695-716, 2018. Abstract | Links | BibTeX @article{Collard2018,
title = {Interpreting into an SOV Language: Memory and the Position of the Verb. A Corpus-Based Comparative Study of Interpreted and Non-mediated Speech.},
author = {Camille Collard and Heike Przybyl and Bart Defrancq},
editor = {Nathalie K\"{u}blera nd Rudy Loock and Mojca Pecman, Les Presses de l’Universit'{e} de Montr'{e}al},
url = {https://id.erudit.org/iderudit/1060169ar},
doi = {10.7202/1060169ar},
year = {2018},
date = {2018-12-00},
journal = {Meta},
volume = {63},
number = {3},
pages = {695-716},
abstract = {In Dutch and German subordinate clauses, the verb is generally placed after the clausal constituents (Subject-Object-Verb structure) thereby creating a middle field (or verbal brace). This makes interpreting from SOV into SVO languages particularly challenging as it requires further processing and feats of memory. It often requires interpreters to use specific strategies (for example, anticipation) (Lederer 1981; Liontou 2011). However, few studies have tackled this issue from the point of view of interpreting into SOV languages. Producing SOV structures requires some specific cognitive effort as, for instance, subject properties need to be kept in mind in order to ensure the correct subject-verb agreement across a span of 10 or 20 words. Speakers therefore often opt for a strategy called extraposition, placing specific elements after the verb in order to shorten the brace (Hawkins 1994; Bevilacqua 2009). Dutch speakers use this strategy more often than German speakers (Haeseryn 1990). Given the additional cognitive load generated by the interpreting process (Gile 1999), it may be assumed that interpreters will shorten the verbal brace to a larger extent than native speakers.
The present study is based on a corpus of interpreted and non-mediated speeches at the European Parliament and compares middle field lengths as well as extraposition in Dutch and German subordinate clauses. Results from 3460 subordinate clauses confirm that interpreters of both languages shorten the middle field more than native speakers. The study also shows that German interpreters use extraposition more often than native speakers, but this is not the case for Dutch interpreters. Dutch and German interpreters appear to use extraposition partly because they imitate the clause word order of the source speech, showing that, in this case, extraposition can be considered an effort-saving tool.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In Dutch and German subordinate clauses, the verb is generally placed after the clausal constituents (Subject-Object-Verb structure) thereby creating a middle field (or verbal brace). This makes interpreting from SOV into SVO languages particularly challenging as it requires further processing and feats of memory. It often requires interpreters to use specific strategies (for example, anticipation) (Lederer 1981; Liontou 2011). However, few studies have tackled this issue from the point of view of interpreting into SOV languages. Producing SOV structures requires some specific cognitive effort as, for instance, subject properties need to be kept in mind in order to ensure the correct subject-verb agreement across a span of 10 or 20 words. Speakers therefore often opt for a strategy called extraposition, placing specific elements after the verb in order to shorten the brace (Hawkins 1994; Bevilacqua 2009). Dutch speakers use this strategy more often than German speakers (Haeseryn 1990). Given the additional cognitive load generated by the interpreting process (Gile 1999), it may be assumed that interpreters will shorten the verbal brace to a larger extent than native speakers.
The present study is based on a corpus of interpreted and non-mediated speeches at the European Parliament and compares middle field lengths as well as extraposition in Dutch and German subordinate clauses. Results from 3460 subordinate clauses confirm that interpreters of both languages shorten the middle field more than native speakers. The study also shows that German interpreters use extraposition more often than native speakers, but this is not the case for Dutch interpreters. Dutch and German interpreters appear to use extraposition partly because they imitate the clause word order of the source speech, showing that, in this case, extraposition can be considered an effort-saving tool. |
Karakanta, Alina; Vela, Mihaela; Teich, Elke EuroParl-UdS: Preserving and Extending Metadata in Parliamentary Debates Inproceedings ParlaCLARIN workshop, 11th Language Resources and Evaluation Conference (LREC2018), Miyazaki, Japan, 2018. Abstract | Links | BibTeX @inproceedings{Karakanta2018b,
title = {EuroParl-UdS: Preserving and Extending Metadata in Parliamentary Debates},
author = {Alina Karakanta and Mihaela Vela and Elke Teich},
url = {http://lrec-conf.org/workshops/lrec2018/W2/pdf/10_W2.pdf},
year = {2018},
date = {2018-00-00},
booktitle = {ParlaCLARIN workshop, 11th Language Resources and Evaluation Conference (LREC2018)},
address = {Miyazaki, Japan},
abstract = {Multilingual parliaments have been a useful source for monolingual and multilingual corpus collection. However, extra-textual information about speakers is often absent, and as a result, these resources cannot be fully used in translation studies. In this paper we present a method for processing and building a parallel corpus consisting of parliamentary debates of the European Parliament for English into German and English into Spanish, where original language and native speaker information is available as metadata. The paperdocumentsallnecessary(pre-andpost-)processingstepsforcreatingsuchavaluableresource. Inadditiontotheparallelcorpora, we collect monolingual comparable corpora for English, German and Spanish using the same method.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Multilingual parliaments have been a useful source for monolingual and multilingual corpus collection. However, extra-textual information about speakers is often absent, and as a result, these resources cannot be fully used in translation studies. In this paper we present a method for processing and building a parallel corpus consisting of parliamentary debates of the European Parliament for English into German and English into Spanish, where original language and native speaker information is available as metadata. The paperdocumentsallnecessary(pre-andpost-)processingstepsforcreatingsuchavaluableresource. Inadditiontotheparallelcorpora, we collect monolingual comparable corpora for English, German and Spanish using the same method. |
Karakanta, Alina; Przybyl, Heike; Teich, Elke Exploring Variation in Translation with Relative Entropy Inproceedings International Symposium on Parallel Corpora ECETT / PaCor 2018, Madrid, Spain, 2018. BibTeX @inproceedings{Karakanta2018b,
title = {Exploring Variation in Translation with Relative Entropy},
author = {Alina Karakanta and Heike Przybyl and Elke Teich},
year = {2018},
date = {2018-00-00},
booktitle = {International Symposium on Parallel Corpora ECETT / PaCor 2018},
address = {Madrid, Spain},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
|
Karakanta, Alina; Vela, Mihaela; Teich, Elke Preserving Metadata from Parliamentary Debates Inproceedings Fišer, Darja; Eskevich, Maria; de Jong, Franciska (Ed.): Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), European Language Resources Association (ELRA), Paris, France, 2018. BibTeX @inproceedings{Karakanta2018,
title = {Preserving Metadata from Parliamentary Debates},
author = {Alina Karakanta and Mihaela Vela and Elke Teich},
editor = {Darja Fi\v{s}er and Maria Eskevich and Franciska de Jong},
year = {2018},
date = {2018-00-00},
booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
publisher = {European Language Resources Association (ELRA)},
address = {Paris, France},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
|
