Machine translation for manufacturing & technical content
Our client, a global leader in the construction industry, manufactures machinery and equipment for worldwide use. We’ve been working with them for a number of years, providing translation services for their user guides, operating manuals and other technical content. In one year, we received 127 million English words for translation, across 52 language combinations.
After we had been working together for some time and fully understood our client’s individual position, goals and direction, we explored more ways to ensure our client remained competitive on a global scale. With budget cuts within the company, and an increased need to improve turnaround times, we turned to language technology tools to improve the process.
As we had been working in partnership for several years, the client had built up a large bank of translation memories, helping us to reduce the 127 million words received to just 10.5 million words actually requiring new translation. These valuable linguistic resources were then also used to build customised machine translation (MT) engines in their most requested languages, including: French, Latin American Spanish, Brazilian Portuguese, Italian, Russian, Dutch, Indonesian, Polish, German, Swedish and Ukrainian. These extensive translation memories meant that the level of usable translated output created by the MT engines was high. All of our MT engines are run through vigorous testing phases before full use with the client, as this ensures we can guarantee effectiveness before live launch.
We have also trained our linguists to effectively post-edit – a critical factor in the success of machine translation workflows.
Through the use of technology, 50% of the words received for translation were processed by MT engines, enabling the customer to significantly reduce their translation costs, with a saving of 30% on the new word rate.
This has resulted in their budget being able to go further; translating more material and reaching new audiences.
Whilst the cost has been reduced, Capita TI has ensured that quality is not affected, by closely monitoring the engine output and post-editing performance. All languages currently using a machine translation workflow have seen an increase in their quality scores in comparison to their human translation counterparts. We’ve continually exceeded our client’s quality expectations, proven by being the top-rated vendor for quality for the 6th year running (as verified by independent and internal linguist teams), highlighting that machine translation can provide excellent service and value for money.
The customer has also seen a reduced turnaround time per project. Currently we are realising up to 50% time improvement versus human translation, significantly improving the time to market and therefore reducing overall operating costs, not only those linked to localisation. We re-tune our MT engines on a regular basis, in order to ensure continual improvements.
We have already started exploring the application of neural machine translation, to see whether this can yield better results and bring more benefits than the traditional statistical method.
Cost and quality remain a priority for the client, however one key focus this year will be on improving the turnaround time of translations. We’ve provided consultation and advice on how the client can implement automation and software on their side, in order to streamline and centralise the process.
Using machine translation to translate content for British manufacturer
Delivery multilingual technical manuals for luxury yacht brand
Find out more about who we help