Charging for Japanese to English translation

Some notes for translators and their clients

• Do you count words written in non-Japanese characters, such as English names?
  Some clients claim that they are paying only for J->E translation, and as no work is involved in words like these they should be excluded from any character count. There are some counter-arguments though ... (a) In other languages such words are always included; (b) There IS work involved, as the translator needs to type the word out, check the spelling and layout, see how it fits into the rest of the sentence, and check whether the original writer might have made a mistake -- particularly important with numbers; (c) If a client claims certain items are not part of the (paid) translation, the translator could in turn claim there is no need to include them at all, no matter how illegible this renders the translation.

• Do you count punctuation marks?
  Again, some clients are loath to include these in a character count, and counter-argument (a) above does not apply this time. However, counter-arguments (b) and (c) do still apply, and although it may be true that punctuation was not an integral part of the Japanese language many hundreds of years ago, nowadays there are many occasions where one comma etc. will totally change the meaning of a sentence. Furthermore, given the regular arrangement of most printed Japanese text, it is far more difficult to leave out the various punctuation marks when counting than to include them. Finally, the convention in Japan is to use manuscript paper (20 columns x 20 characters) and calculate or state the length of a document by multiplying the number of pages by 400; spaces, punctuation marks etc. all take up the same space as other characters, and are included in the count. Let us call this counter-argument (d).

• Do you count spaces and empty lines?
  This is the most difficult one to justify, as counter-arguments (a), (b) and (c) would not tend to apply other than in exceptional circumstances. However, counter-argument (d) does still apply and for long documents laid out in a typical Japanese style (i.e. square blocks of text) it is still easier to accept it than to try and subtract all the spaces etc. from the total count. In such a document, the amount of white space is in any case going to be only a small proportion of the total.

How to count characters -- three ideal cases

1. Documents that are laid out in Japanese style, such as newspaper articles or patents, are very easy to work out -- all characters are the same size (unlike proportional fonts in English or other European languages) and all lines are normally the same length, so all one needs to do is calculate e.g. 25 characters per line x 20 lines per paragraph x 2 paragraphs per column x 2 columns per page = 2000 characters per page. If there are large quantities of English words or numerals (e.g. tables), and the client can supply these electronically (e.g. on disk) so that there is no need for re-typing or checking, a translator would probably agree to deduct an appropriate amount.
2. If the client can supply the original Japanese text in electronic format (on disk, by modem, or via the Internet), it may be possible to do a character count using a Japanese word processing programme. In such a case, the total number of characters might come down a little (some areas of white space are likely to be achieved via formatting commands), but the time saved counting characters and the convenience of having a text to type over would normally be more than adequate compensation.
3. For a long document, it can take a very long time to count all the characters. Yet as the resulting translation is almost certain to be produced in a word processing package, it is easy to count the number of English words produced. Wouldn't it be nice if there were some way to correlate the two? Actually, there is and I sometimes use it. Having compared several documents with my translations of them, it turns out that the ratio of Japanese characters to English words is normally between 2.1 and 2.3, although texts with many Katakana words (e.g. lists of chemicals, video game titles etc.) often end up over 3. I therefore take the English word count, multiply by 2, round up, and use that as the Japanese character count. In theory, texts with a high proportion of Katakana should be multiplied by 3 but unless the Katakana words are particularly difficult I tend to ignore this. One important point ... the above ratios apply to my own style of translation, and the corresponding ratios for other translators who are more terse or verbose than I may be quite different -- in effect, each translator ought to work out their own ratio and revise it regularly.

Agreeing on a rate

Examples

You will note that the text consists of 213 full-size Japanese characters, 22 punctuation marks and other symbols, 32 numerals (of which 4 are half-size), and 21 English letters (in 4 words) -- a total of 288 characters. Now let us apply a grid as in manuscript paper:

It is evident that in normal Japanese practice this text would be reckoned as 400 characters, the "missing" 112 characters being white space. (Note: this is rather a high proportion due to the artificial nature of this example.) The conclusion is that if a client only paid for (in the worst case) the full-size Japanese characters, the translator would receive around 53% of the normal rate (and should therefore quote the client a rate 89% higher than usual to receive the normal amount)!

In practice, I think experienced translators and experienced translation agencies will both appreciate the points raised here, and both be accustomed to compromising to a certain extent. Much depends on how desperate each side is for the work! Yet the question of how to count characters is definitely worth some thought, particularly when dealing with a new client, and I hope this page helps to serve as reference material.

Finally ...