Some time ago, after reading the millionth online discussion of the evils of CAT discount schemes and how to counteract them, I decided to add a more quantitative tone to the discussion. I do not share the blanket objections that some have to discount schemes of any sort; if I have an easy text to do that consists of 50% repetition, I am open to discussing the rate for the repeated content. However, I do find some of the CAT schemes proposed by agencies to be beyond ridiculous, and anyone who is serious about paying nothing or a few percent for matches and repeats will be greeted with a hearty, spontaneous laugh for starters.
Clearly, in many cases there is a need to look closely at proposed CAT schemes and consider alternative schemes or word price increases to achieve fair overall compensation. With that in mind, I created an Excel spreadsheet I call the Target Price Defense Tool to help translators (or other service providers) evaluate alternatives. It is available here.
Update September 16, 2012: After realizing recently that the tool has been unavailable for a long time after a domain change, I reviewed it and decided to add it to the growing Sodrat Suite for Translation Productivity, part of an Open Source resistance movement to the abusive complexities of ill-conceived technology in the translation profession.
Quite. There are instances where it's absolutely reasonable to grant a discount, but only where repetitions are true repetitions. Only too often those repetitions identified by the agency based on its analysis are not real repetitions, but ones that still require a major rewrite, if not retranslation.
ReplyDeleteThe problem with such analyses is that they capture repetitive patterns at the word level only; that the entire segment (e.g., sentence) is not really a match of any sort is, of course, ignored by the software.
Proof positive also of the fact that a machine cannot "appreciate" language.
Not sure what you mean, Werner. Repetitions are measured at the segment level, not at the word level. But you're right that many people don't understand the difference between repetitions and 100% matches from past work. To measure repetitions and be very sure you can always run your analysis with no TM or an empty one :-)
Delete