Like many industries, the translation industry has a lot of specialist jargon. Translators, and translation companies love spouting it, and sometimes forget that it can be confusing. Here are a few favourites, and our explanation, some of the definitions shouldn’t be taken too seriously:

Machine Translation (can be abbreviated as MT):

This is when the machine does the work. Essentially two databases of words are matched, and the original language replaced by the new language. Unfortunately, it isn’t that simple. You can tell this if you have ever tried to “translate” a webpage using a free tool like Google. The problem is that you get drivel out. It may be OK, if you just want to get some general gist, but no more. If a translation agency says that they use machine translation, you should run a mile.

Computer Assisted Translation (or CAT):

Though it sounds the same as machine translation, this is quite different. The translator works with a programme which commits phrases to memory as they work. If they then translate a very similar text again, they are prompted by matches from the memory.

Unlike machine translation, this matches larger units, such as complete sentences. It helps the translator, as they can build a glossary and ensure they use consistent terminology. For documents with a high level of repetition, it means that they can focus on the new text. They then simply proofread the matching sections. This gives productivity gains for the translator, and is cheaper for the client.

The very important difference from machine translation, is that the text is always human authored and edited. There are current moves in the industry to integrate machine translation into CAT. We are concerned by this. The underlying philosophy is that the machine produces text which is then improved by a human proofreader. We think this is fundamentally unsound. It is always harder to improve a poor text, than to take a well written one and tweak it. Some MT is so bad, that it really wouldn’t be any use at all.


This is the process of adapting a text for a local audience, usually in a different country. Some localisation issues are very obvious. For example, if your text includes a phone number, is it in the correct format for the target country? The aim being that someone can read the document, and use the phone number in the target country. A result may be that you get phone calls that you can’t deal with. Will your switchboard in London be able to respond to phone enquiries in Spanish or Russian?

Localisation should also take account of culture. An example might be a translator being asked to translate a food product containing alcohol for an Islamic country. If they simply translate the text, they aren’t helping their client. The product will be stopped at port, and the client needs to be aware of this potential problem. Here correct localisation, may mean selling a different product range.


This is the spoken task of expressing the meaning of one language in a second language. Interpreting is quite a different skill to written translation. The interpreter needs to be able to listen in one language, and express in the other. This can be simultaneous (i.e. delivered as the other person is speaking), or consecutive (delivered after the original speaker has spoken).


This is the new buzz word in the industry. The concept being that an original text might need major alteration to be presented in a new language. A good example might be a marketing text. Here the translator is looking to reflect a correct writing style in the target language. They therefore have to do more than just reproduce the original text in the target language. The problem with this is that there is an extent to which all good translators do this. They aren’t giving a literal word for word translation into the target language. Rather, they should actually be thinking about how the text reads and sounds in the target language.