Notes from Risk Analysis in Translation Process
. Anthony Pym's talk was interesting. He opened with a disconcerting but very true statement:"Not much of translation theory is really helpful to translators"
According to Pym, all translations have zones of low risk and high risk. In the zones of low risk, the translator usually works fast, in the zones of high risk, the translation process is slowed down: thinking is required because depending on the choice made by the translator there is a possibility of not fulfilling the purpose of the translation, as defined in the Skopos Theory
. Translation risk is based on a failure condition, part of the text is highly inductive to translator putting her foot in her mouth, i.e., failing to deliver the skopos. If the skopos calls for precision, translator must be accurate which sometimes entails adding information not included in the original. If skopos calls for gist, translator may generalize and collapse part of the information contained in the original. All translators are familiar with this: you stumble on a part of the text that doesn't contain crucial information but presents a translation problem (terminology for example). If there is trust between you and your client, it's safe and perhaps preferable to generalize or use a descriptive solution. Bottomline: if working on a low risk area of the text, don't go out of your way to translate a term or sentence.
The risky parts of a translation are not necessarily difficult. My example: front matter on a manual, customer support phone number. If you're translating the manual for a Brazilian audience, this phone number has to be the local phone number. If you don't provide it, the reader will not be able to get in touch with customer support and the skopos will not be reached. Moreover, this leads to an important point to remember:"Translation always requires more information than that's on the text"
How to get more information? There is the obvious: dictionaries, glossaries, discussion lists, experts etc. Anthony Pym favors parallel texts
as a reference source. In addition to this, the client should be used as an information source. Some of the information that will help determine the skopos, for example, age group of readership, type of publication, is indeed only available to the client. Terminological queries should only be directed to the client only after all other information sources have been exhausted. (I don't really agree with this. In my experience, the earlier the client gets involved in clarification of issues in the source text, the better the results. Some of my agencies distribute regular Q&As files throughout the translation and editing phases, with questions being asked by translators and answered by the end client.)
To manage risk in the translation process, the best strategy is adhering to a job specification, as proposed to Daniel Goudec (not sure if this is the correct spelling). Goudec argues that technical translation is the easiest thing in the world, provided translator, client and agency have mutual agreement on the job specification. Goudec also says that the translation workflow should be distributed as follows: 60% of time for pre-translation, that is defining the job specification, 20% of time for translation and 20% of time for editing (which leaves no time for quality assurance, which in my view requires another 10%).
Here's a checklist for defining your job specification, that is things that should be provided by the client.Starting materials
- Material (all files received?)
- Images and graphics (must be received to avoid translation risk)
- Client glossaries (received? include creation of glossary in specification?)
- Previous translations (received? useful for reference)
- Phone number of in-country experts (touchy issue)
- Style guide (not mentioned by Pym, but highly desirable)Function
( or questions you should always ask your client)
- What is this translation for? Information? Publicity? Confidential?
- Who is the reader ? Expert? Layman?
- Level required (quality has a price, sometimes client only needs gist translation of 200 pages and will come back saying 'This five pages are interesting. Please have them translated decently)
- How many revisions do you need? (Editing, proofreading, expert revision?)High Risk = High PayTask
- File format (the mermaid notes: file format may be different, for example, client wants Frame Maker or a fully functional dynamic website, but translator my work on s-tagged . RTF files or Catalyst or what have you)
- Payment terms
- Additional costs estimate (Any faxing or fedexing involved?)
Finally, print everything out, put a translation services contract on top and have it signed by client and translator.
Risk usually arises when there is a rich point, as defined by Michael H. Agar
, points where cultures make contact but equivalence is hard to find. For example Tu/Vous, du/Sie, in different cultures usage differs and it takes years for a non-native speaker to realize when one or the other is required.
The translator's reaction to risk often consists of adopting an avoidance strategy, which may be ok, according to the skopos.
In cases where the original skopos is to indicate alterity (think for example of a Discovery Channel documentary about foreign lands or a book about exotic music instruments), it's best not to translate the exotic term, for example, candomblé, cavaquinho, entrudo etc. Translation of such terms, such as for example shrovetide for entrudo, if the context is Brazil or Portugal, would actually be detrimental to the skopos.
The localization industry is far more advanced than translation industry in assessing risks and establishing risk management procedures. (Absolutely right, most of what Pym said just makes part of my everyday routine as a localizer. I rarely work on a project that doesn't have an extremely detailed instructions sheet and many times, a scope file).
Objections to the exposé above:
- The general expectation on Western culture is that translators should translate everything they can, leaving no parts of the text out. Image of translator often evokes distrust, adopting this model could accentuate this further.
- In other cultures, clients are not pro-active, so job specification may not work at all.
- Only works if there is trust between translator and client. Doesn't take into account competing vendors' quality assessments and other low blows commonly used among agencies/translators to steal each other's clients.