A popular topic for discussion in the translation world is how prices should be calculated. Some translators charge by the page or by the line, but most prefer to calculate their rates according to the number of words in a document. However, when providing a quotation for a customer, it is important to specify whether this means the number of words in the original, or source document, or whether it refers to the translated, or target text.
Like many of my colleagues, I base my prices on the number of source words. The customer’s text is usually (although not always!) a finished product and the number of words is fixed. On the other hand, at the time of quotation, the translation has not yet been written and its exact length unknown. Even after submission, the client may request certain changes, which could alter the word count of the finished product. Therefore, to base a quotation on the translated document would mean that the actual price would not be known until the end of the project. Also, it would open up the possibility for dispute and malpractice. An unscrupulous translator could be unnecessarily verbose in order to increase the fee, or he or she might feel obliged to use the minimum possible number of words and the translation quality may suffer as a result.
Since I began translating, I have always been aware that my translations usually turn out to be slightly longer than the originals. This is easily explained when working from Russian into English, as Russian is a much more compact language. There are no articles, no present tense of the verb ‘to be’, and prefixes or case endings give added meaning without the need for extra words as in English. For example the Russian ‘Я – переводчик’ (2 words) would be translated into English as ‘I am a translator’ (4 words). For translations from French into English, however, the reason is less clear. Nevertheless, I do find myself occasionally translating culture-specific references with an explanatory phrase, rather than just one word. ‘Payer l’octroi’, for instance, became ‘to pay the octroi tax’, making the meaning clearer for an English-speaking reader.
Although I knew that my translations were generally longer than the originals, when I was asked how much longer they were, I realised I had no idea. I hazarded a guess at approximately 20% for Russian translations, and slightly less for French, perhaps 10%, but I was far from certain. So I decided to find out. I took a selection of French and Russian texts, both literary and commercial, which I had translated into English, and compared the word counts of the original texts with the translated versions. I was surprised to find that my translations from Russian into English were an average of 26% longer than the source document, with literary translations being longer than commercial ones, whereas translations from French into English were an average of just 2% longer than the originals, with little difference between various types of text.
Given more time, and an aptitude for statistical analysis, these findings could make a fascinating study. Looking at the difference between my two source languages, I find myself wondering whether other languages follow similar trends. German, with its long compound nouns, might produce some interesting results, and what about non-European languages, such as Chinese or Arabic? Is the word count reduced when translating the opposite way, e.g. from English into Russian? And would the results vary from one translator to another depending on individual styles? This might make an excellent topic for discussion at some future translators’ gathering.
Although my French-speaking customers are unlikely to be pleased about a 2% price increase, their Russian counterparts would certainly be less than impressed by a 26% rise. In a supermarket-style price comparison, the discrepancy between the languages does not seem particularly fair, especially as the end product is in English in both cases. So, in the interests of fairness and simplicity, I’ll carry on calculating my rates according to the original text.