Is good, fast and cheap really impossible?

The Project Management Triangle (also referred to as the Triple Constraint or Iron Triangle) tells us that it is not possible to carry out projects quickly, cheaply and at the same time, of a high quality. According to the theory, one side of the project management triangle cannot be changed without affecting another.

I don’t want to challenge this fundamental assumption, but I can safely say that we at STAR Servicios Lingüísticos do our best to make the impossible possible, and that at times we achieve a balance of the three constraints that leaves our clients happy. And happy clients are our ultimate raison d'être. Here is just one example of many where we were able to achieve this!

The challenge

We recently delivered a medium-sized translation project (about 80,000 words) to a new client from the construction machinery sector. The original challenge (remember the three constraints) posed by our client was to translate a highly technical publication from Spanish into both English and French in a relatively short time, and it goes without saying, for as little money as possible.

The key: good source material

At the outset, we received professionally layouted Adobe InDesign files with well-structured and concise technical product information. The client also provided us with the same documentation from the previous year as reference material, also in InDesign.

CAT makes its entrance

We imported the files into Transit NXT, our preferred translation tool, and saw that there was a significant number of internal repetitions or IRs (segments that are repeated several times throughout the publication), and a high number of perfect matches (PMs). Perfect matches in Transit are segments that have a 100% equivalent segment in the reference material, or segments that contain only numbers and therefore do not need to be translated. Internal repetitions and perfect matches are translated automatically or semi-automatically and are therefore charged at a considerably lower rate than new words. That is where we identified the first big money and time saver for our client.

A closer look at the IRs and PMs revealed that there were many codes in the text consisting of strings of numbers and letters that did not need to be translated. We therefore adapted the import filter in Transit NXT to protect all of those untranslatable strings. The import filter is used to convert the InDesign Files (or any other file format) into the internal Transit format, XML tagged Unicode text, a transparent and lightweight format. With our fine-tuned filter, we managed to write-protect around 7000 words, thus deducting them from the word count and further reducing the translation cost for our client.

Leveraging reference material

We then quality-checked and aligned the reference material sent by the client in order to convert it into a Transit translation memory. Once we had imported the reference material into Transit NXT, we reimported the publication and leveraged it using our translation memory. In so doing, we were able to pretranslate approximately 40% of the new publication automatically. 30% of the remaining untranslated segments were so-called fuzzy matches of over 70%, i.e. segments that bear a resemblance of more than 70% to segments in the translation memory. Fuzzy matches are also cheaper than new words.

Lots of small savings make big savings

Altogether, the ability of our computer-assisted translation technology to identify internal repetitions, the usage of leveraging techniques such as perfect and fuzzy matching and the ability to write-protect untranslatable strings, resulted in savings of around 56% compared to the cost of the translation without computer assisted translation technology. As a result, the average price per word was a mere 44% of the new word price. It is difficult to see how machine translation with post-editing could match that price.

More good news for our client

The cost savings were not the only good news for our client. Thanks to the high degree of leveraging and reuse, the total turnaround time for the whole project from kick-off to delivery of the translation was reduced to 20 working days, with two translators and one proofreader working closely as a team.

Takeaways

The right CAT technology, used to its full potential, can easily surpass MT + Post-editing, with regards to price and speed, especially when the material to be translated is highly specific technical documentation. Another lesson learned is that, at least in terms of translation projects, the quality of the source material is a decisive factor. If the source material is of a high quality, the three constraints of the project triangle will give you less of a headache.

Clients like the one in our example who have understood the value of their linguistic assets, are still a relatively rare species here in Spain. However, with the growth of the export industry and the need to sell products and services in other areas, such clients are growing in number. The idea that high-quality multilingual product information is key to international success is gaining ground. That’s definitely good news, for the Spanish economy as a whole and especially for technical writers and translators.