In today’s article we will tackle a topic that can be approached from many different angles – translation prices. Naturally, as with any type of service, clients will think the prices are too high, the people providing the services will claim the prices are too low, while the truth is, as always, somewhere in between.
The perspective is also heavily influenced by technological advancements, and we will also briefly look into this aspect. Let’s dive in and decode the translation prices; once and for all!
How are translation prices established?
Let’s start with how the prices are calculated for specific services in the translation industry.
There are several main services:
Translation and review are mainly calculated per word since the majority of content can easily be calculated this way. If a translator can translate around 2.000 new words a day, and their rate is e.g. 0,05 EUR per word, you can easily calculate the gross pay. It is important to remember the ‘new words’ argument, as repetitions are not paid as ‘new words’, and are discounted in most workflows.
Interpreting cannot be charged per word, as we are dealing here with the spoken word, and no-one really knows how many words there will be in advance. Interpreting (both consecutive and simultaneous) is, therefore, charged usually per hour.
Transcription and subtitling are linked to audio and video translation tasks, and it’s also very difficult to calculate the number of source words before the task is done. This is the main reason why these services are charged per hour or minute.
Translation price per word vs translation price per hour. Is this the best way?
There have always been, and still are, debates on how to charge for linguistic services. Should translation also be charged by the hour, or should it be charged the way it is charged now?
We think that charging by ‘new words’ is the fair approach for both the translator and client. This is not because it has been done for so long, but because it makes the most sense.
Machine translation and its influence on translation prices
Machine translation is a very important factor in this discussion. We have written extensively about how MT shapes the translation industry of today, mentioning that the prices for the review of machine-translated content by human native speakers are lower than translation performed by humans from scratch.
The argument is that neural machine translation, which has been used in the past couple of years, has changed the face of the industry. Its quality has reached such levels that the people reviewing machine-translated content became matter experts who rate the quality of the translation.
For these experts, the current rate per word would not make sense, and they should be charged by how long it takes them to actually review the content. Is this a good or bad idea?
The old way vs. the new way
The logic behind the ‘per hour’ approach for post-editing of machine translations (PEMT) is very clear. Its main purpose is to differentiate between the old approach (in which humans were checking human translations) and the new approach (in which machines would perhaps also define which parts need to be post-edited by humans).
With people not really understanding what ‘new words’ are, the hourly rate would be more understandable. However, would it be easy to measure? How would it apply to slower vs. faster reviewers, people taking breaks, and how many breaks could they take in a certain time period?
The question is whether this is really necessary. Let’s say that neural machine translation can actually replace the first step in the traditional translation workflow with the human translator. In a 4-eyes translation workflow, in which you have a reviewer anyway, changing how the reviewer is paid doesn’t really make sense.
How do we calculate translation prices at Text United?
To explain this properly, we have to go back to the ‘new words’ concept. Let’s say that today you need to translate the following:
1. This is the first sentence.
2. It is very important to translate this correctly.
3. Otherwise, the results could be catastrophic.
Prices are calculated as (for example) 0,04 EUR per word, multiplicated by the number of words. All the words are ‘new’, as there is no Translation Memory (a database of all previous translations).
The translation is done and saved to the Translation Memory. Then, next week, you come to us with the following content for translation:
1. This is the new sentence.
2. It is very important to translate this correctly.
3. I need everything ready by tomorrow.
Compared to the previous translation, the sentences can be classified in the following way: the first sentence is a 95%-99% match, as only one word is different, compared to the old translation. This sentence would be heavily discounted.
The second sentence is a 100% match as it can be re-used from the Translation Memory. This sentence would either be heavily discounted or not charged at all.
The third sentence is ‘new words’ as it cannot be re-used from previous translations. This sentence would be charged according to the rate per word (again e.g. 0,04 EUR per word). If we apply the same logic to other workflows, such as post-editing of machine-translated content, then the only thing that changes is the rate per word.
If a human translator would have charged 0,04 EUR per word, the review will most likely cost 0,02 EUR per word. In this way, you get content checked by a human for a lower price, and you can be sure that there will be no errors.
You can get the best translation prices, too
If you choose any of the workflows mentioned above, you will also need project managers and reviewers who are used to this, and proper software to apply it to your digital environment.
The good news is, we have all of the above and more! If you are interested in researching this further and want to discuss how translation prices are formed, especially for your content, we are here for you.
Reach out and let’s talk about how we can accommodate your needs – as always, no strings attached!