Purpose: to find out how many professional translators have really been involved in MT postediting projects and formal MT evaluation campaigns

This is a multivote poll so that you can choose several options. The different options reflect various categories.
For example: 
* number of projects actively participated in (one/two and stopped participating versus several projects, possibly regular involvement)
* level of postediting (rapid, full high-quality, or somewhere in-between)
* dictionary building (was conducted or not for the projects)
* MT evaluation only (postediting was not included)
* no experience in MT postediting or evaluation

Are there really only 4 people on this site who have some experience in MT postediting or no experience in it?

Jeff

Originally written by Harry Bornemann on November 14, 2008 2:30 AM I am not sure  -  I often use Personal Translator 2006 Pro for En-De pretranslations, because it is great for colour and country names and many short simple sentences.  This means I can keep 5% of the pretranslated segments, as well as the first 1 to 3 words of another 5%, and have to delete the remaining 90%.  Would you call this postediting? I also tried building a dictionary first, but I found that it is much faster to use search and replace for some words in the translated text, even if I have to adjust some of them grammatically.

Thanks Harry for your feedback.  Well, it seems like you have helped to coin a new term in this field. I think it should be called Partial Postediting (PP).

The product you mention is by Linguatec.  I've not used that specific one.

As for dictionary building, it depends on the features available in the product. For PROMT, I can analyze the texts, extract the terminology candidates, create an Excel based glossary including the part of speech, and then upload it into the PROMT dictionary and validate the entries with any supplementary info.  This speeds up the dictionary building process.  My 2006 project case study here on TC indicates all the steps, time-stamped, on the dictionary building process for identifying 1200 terminology candidates, and only selecting approximately 450 to be coded (which would result in the same transaltion quality as the 1200), and then uploading and coding them.  Results of the translated content with the application of the specific dictionary were also provided.

The PROMT Expert version has a terminology extractor which displays all terms in context. This combined with a few other features help make it a dictionary building module that can be worthwhile to use.

Jeff