Checking translations for inconsistency - a tool for the editor

Published in the Procedings from AMTA-96, Montreal, 1996. CHECKING TRANSLATIONS FOR INCONSISTENCY – A TOOL FOR THE EDITOR Magnus Merkel NLPLAB Department of Computer and Information Science Linköping University Sweden Email: magme@ida.liu.se Abstract Being consistent in technical translations is difficult. Using translation memory software could be one remedy, but the effectiveness of these tools rests on the assumption that most repeated source sentences correspond isomorphically to repeated target sentences. With the help of a discrepancy tool to identify inconsistencies between source and target texts, data from translations made manually and with the aid of translation memories are presented. Apart from being used as a verification tool at the postediting phase of a translation project, the discrepancy tool will measure the relative efficiency of the use of translation memory tools, as well as identify possible shortcomings with the source text. 1 . Introduction When large quantities of technical texts are being translated manually, it is very difficult to produce consistent translations of recurrent stretches of text, such as paragraphs, sentences and phrases. This can have many different reasons, for example, several translators work on different sections of the same document simultaneously, the source text is not final and may be changed at a later stage, and it may be too time-consuming or practically impossible to identify recurrent units in the source text manually. Individual translators making up a translation team will also have individual criteria for choosing a certain translation or even choosing from a set of possible translations.1 One suggested remedy to the problem of consistency in translation is to use tools based on translation memories. When using such systems the translators translate the text interactively with a computer tool that stores and retrieves all identical source sentences with their corresponding translations, which will guide the translator towards consistent translations. It is also possible to reuse old translations stored as translation memories of previous versions of handbooks and thereby minimizing the chances of producing variant translations of the same source segments. The quality of the translation memories that are being put to use in a translation project becomes crucial for the quality of the new target text. Translation memories can be produced either by using a translation memory- based translation tool interactively or by aligning a source text with its corresponding translation with some kind of alignment tool. Translation memory tools are used to the best advantage when the following conditions hold: 1 In this paper the consistency problem concerns consistent translations of sentences, not consistent terminology, although these two are related. An example of a translation checker for terminological (lexical) consistency can be found in Macklovitch (1994). 1. The source text is highly repetitive internally 2. The source text is a new version of a text that has already been translated and exists as a translation memory (i.e., external repetition). 3. Repeated sentences (or segments) of source text are in principle transferable to corresponding repeated sentences in the target text. The first two conditions can quite easily be checked …

Content truncated to 500 words