Going Beyond Google Translate?

Visualizzazione ridotta
dc.contributor.authorChessa, Francesca
dc.contributor.authorBrelstaff, Gavin
dc.date.accessioned2014-05-13T06:32:46Z
dc.date.available2014-05-13T06:32:46Z
dc.date.issued2011-09-15
dc.descriptionCiclo 2012 di seminari interni CRS4, Number 20120229.IT
dc.description.abstractWe motivate and describe the design and implementation of a web-based system for the alignment of parallel texts. It builds on the interactive color-highlight interface now deployed at Google Translate. By a series of simple point and click operations translators can mark up equivalent text-ranges in their own translation and in the original. When successful, the visual cues created by this activity should benefit the understanding of readers of limited degrees of bilingualism -- and may also capture aspects of semantic context not readily available to algorithmic statistical machine translation. We provide a working demonstration that treats poetic texts.IT
dc.description.abstractStatistical machine translation (SMT) delivers texts unacceptable for literary or academic purposes since generally, it cannot assimilate adequate context: Yet how might one ever articulate such context? Here rather than taking a theoretical perspective we adopt an spatio-visual approach made possible by recent advances in the electronic presentation of multilingual texts:– we allow the translator supply the colour higlights... But how? Semantic units don't respect lexical boundaries and they occur at different scales. Any translator, committed to provide a definitive version of a text, eventually arrives at irreversible order of words – and may actually wish to justify their choices by documenting the correspondence between their version and the original. We focus on verse – an extreme challenge for SMT – with the eventual aim of expressing elusive aspects of semantic communication in order to differentiate those that can be articulated via spatio-visual cues. In verse a deviation from a literal correspondence is essential to reestablish in the translation a "decorum" appropriate to the original so that readers are encouraged to achieve an equivalent respect for its author also from the translated works. We use jQuery to provide an interface that lets the human translator mark up what they consider a correct alignment between words, or groups of words, in the original and their own translation – with a view to articulating context that may not be readily available to SMT. We detail below how the interface runs off a web-page and allows the alignment of equivalent ranges in parallel texts via a simple point-and-click action. Alignments created by the user are instantaneously made visible using a variant of the interactive color-highlight system mentioned above. Key to reducing the complexity of the implementation of the interface is our systematic deployment of open-standard, non-proprietary, web technologies.IT
dc.description.conferencedate2011-09-15
dc.description.conferencelocationAlgheroIT
dc.description.conferencetitleCHItaly2011, 13-16 settembre 2011, AlgheroIT
dc.identifier.urihttp://hdl.handle.net/11050/883
dc.language.isoenIT
dc.subjectmultilingual webIT
dc.subjectHCIIT
dc.subjectjQueryIT
dc.subjectTEIIT
dc.subjectXMLIT
dc.subjectparallel textsIT
dc.subject.een-cordisEEN CORDIS::ELETTRONICA, INFORMATICA E TELECOMUNICAZIONI::Elaborazione dell'informazione, sistema informativo, gestione del flusso di lavoro::Architetture di sistemi avanzatiIT
dc.subject.een-cordisEEN CORDIS::ELETTRONICA, INFORMATICA E TELECOMUNICAZIONI::Elaborazione dell'informazione, sistema informativo, gestione del flusso di lavoro::Computer softwareIT
dc.titleGoing Beyond Google Translate?IT
dc.typeContributo a convegnoIT
File
Original bundle
Ora in mostra 1 - 1 di 1
Immagine di anteprima
Nome:
Brelstaff-Ciclo2012DiSeminariInterniCRS4.pdf
Dimensione:
1.4 MB
Formato:
Adobe Portable Document Format
Descrizione:
Download
License bundle
Ora in mostra 1 - 1 di 1
Immagine di anteprima
Nome:
license.txt
Dimensione:
2.06 KB
Formato:
Item-specific license agreed upon to submission
Descrizione:
Download
collections
CRS4 Contributo a convegno