Issues for translators ... the eternal quest. Managing terminology efficiently is essential to complete a language translation that is fit for purpose - and this is becoming even more important as the use of machine translation becomes more and more widespread and translators become "reviewers".Translation quality is a term that seems difficult to define. Quality means different things to different people. A text can "flow" and read well, or not well, or be translated by a professional linguist but use the wrong terminology. Languages, while having a general set of rules, tend to be messy source material unless some kind of controlled language and thinking has been paid during drafting. But this is not often de case. People write with a particular objective in mind and for a particular reason. Language is subjective by nature, and it makes the work of a translator challenging. However, there are certain things that can be done—before even handing off content for translation—to ensure top-notch translation quality. And here we’re talking, of course, about keeping terminological consistency at source. Always naming the same things the same way—that’s terminological consistency. There are many reasons in favor of the consistent use of terminology in our communication—particularly in technical descriptions and instructions. Translation quality is one of those reasons. If often falls down on the translation agency or even the translator himself to provide some kind of terminology consistency when there is none, as Margaret Rogers rightly pointed out in her 2008 article "Consistency in Terminological Choice: Holy Grail or False Prophet?".
What if the source text is not consistent?Translators see the problem as a conondrum with a difficult solutions. There are two basic options open to the translator: maintain the inconsistency of the source text or introduce consistency in the translation where none exists in the source text. This can lead to a lack of consistency in the translated version and thus unfavorable comments from readers, blaming the translator for an inconsistecy that existed there in the original. The conventional wisdom of quality standards (for example in the automotive industry) is to be consistent in both authoring and translating. If the source text is terminologically consistent, then the first option of maintaining inconsistency in the translation falls away. In addition to or in interaction with translation memory, a termbase can be used to support the consistent use of terminology by indicating, for instance, preferred and deprecated terms.
Types of inconsistenciesAt a basic level, terminological inconsistency can be interpreted as the use of different forms for the same referent, that is, synonyms, orthographic variants and geographical variants in the same text or set of related texts, as well as hyponyms. In technical writing, precision of reference, i.e. clarity concerning which objects belong to the class of objects designated by the term (in contrast to vagueness), is prized above elegance of expression, given that the function of most technical documents is informative or instructive: the avoidance of monotony is therefore not a priority. There are also other types of intratextual lexical 'inconsistency' (from another perspective 'variation') which receive less attention than synonymy, such as elliptical forms of nominal compounds which appear in texts in a particular sequence, particularly in English. These have been called "reductions" e.g. small dish earth station ➔ small earth station ➔ small station
ConsistencyWhen consistency of term selection is recommended as good practice in technical translation, 'consistency' could mean one of three things. First, it could mean replicating a consistent pattern of term use in the source text (Table 1, pattern C). It is perhaps in this interpretation that 'consistency' is most commonly and simply understood: what is considered a well-motivated use of one term for the same referent in the source text is retained in the target text. Second, it could mean changing an inconsistent pattern of use to a consistent one, as in Table 2, pattern G. For this to be considered good practice, the inconsistency in the source text would need to be unmotivated, e.g. an apparently arbitrary use of synonyms. In this interpretation, the translator 'corrects' what seems to be a badly written aspect of the source text. There is also a third logical possibility, namely the consistent transfer of a pattern of term choice from the source text to the target text, whether the pattern in the source text is in itself consistent (see Table 1, pattern A) or inconsistent (see Table 2, pattern E). In the latter case, the target text replicates the inconsistent pattern of the source text. In the former case, consistency is replicated (and pattern A is therefore indistinguishable from pattern C). Inconsistency Just as consistency can be variously interpreted, so also can inconsistency. First, a consistent pattern of term use in the source text (Table 1) can be inconsistently translated (pattern D). This could be construed as an inappropriate strategy if it introduces unmotivated variation such as a variety of orthographic forms, regional variants or synonyms in the target. However, it could also be the case that the form of the term in the source language lends itself to consistent use, e.g. a single-word non-compound term, whereas the target language term is a compound which can undergo ellipsis. In such a case, so-called inconsistency in the translation could be well-motivated.
Summing up, the nature of the source-text 'inconsistency' in any particular text is an important consideration in evaluating translation strategies. If the source-text author has used different terms for the same referent (Table 2), the source-text variation may be stylistic (i.e. not well-motivated in the context of technical translation and therefore a candidate for 'correction') or pragmatic as in the case, for example, of a well-motivated sequence of elliptical forms of a compound. In such cases, a similar pattern of variation may emerge in the target text (pattern E), but only assuming that the full target-language term has a similar form and similar reduction patterns. The likelihood of such a situation may vary languagetypologically, such that more closely related languages such as English and German are more likely to share formation and reduction patterns than less closely related languages such as English and Russian.