Different rate for English words in the source text? - When translating into English Thread poster: Katalin Horváth McClure
|
|---|
I have recently came across an agency with a very "unusual" way of counting the text to be translated. I was wondering what you all think about it.
The source text is Japanese (but it could be another language, too), and the target is English. The field is technical, computer related, and the source text includes English words, such as commands to type in, numeric data, etc. These strings are part of continuos text, most of the time explanations about what those commands are doing, or how ... See more I have recently came across an agency with a very "unusual" way of counting the text to be translated. I was wondering what you all think about it.
The source text is Japanese (but it could be another language, too), and the target is English. The field is technical, computer related, and the source text includes English words, such as commands to type in, numeric data, etc. These strings are part of continuos text, most of the time explanations about what those commands are doing, or how to use their parameters. I am writing this just to make it clear that I am NOT talking about a data-table with standalone alphanumeric strings.
The agency counts the source text with its own proprietary software, which gives a count in WORDS, and not characters. A bit of explanation here for those that don't work with Asian languages:
Normally, translations from Japanese to English are paid by the source character count, or by the target English word count (target word count is usually used when it is not practical to obtain a source character count, for ex. the source is a hardcopy).
The rate is obviously different in these two cases, but most translators in this language pair use a 2.5 multiplier between the rates. The reason is that on average, the number of target English words is 2.5 times less, than the number of source Japanese characters.
Now, here is the trick. MS Word and other software can give a "Word" count of a Japanese text, in addition to the character count. This means that every Japanese character is counted as 1 word, and every alphanumeric string is counted as 1 word. So, if you have the string "print" in the source, it would be 5 characters, but it would be only 1 word.
The agency is paying my "per character" rate for each WORD. This means, that they pay less (approx. 2.5 times less) for strings that are written using alphanumeric characters, compared to text written using Japanese characters.
I found this pretty upsetting, it feels like they are trying to skim off the last possible layer from the translator's pay. (I don't even want to mention the fact that they pay 50% flat rate for fuzzy matches, and anything above 50% match is considered fuzzy, and they pay zero for repetitions and full matches but they do want you to mention if you find anything wrong with those - free proofreading, huh?)
I bet if the source language was using latin characters, such as German, they would not count the text this way, simply because they would not be able to, right? (Words are words, that's it.)
Yeah, one may argue, that those strings don't require translation in the strict meaning, but they are part of a continuous text, requiring understanding, proper placement in the target text, and sometimes spelling corrections. In addition, with this agency, no CAT tool is used during the translation process, the source is provided as a two-column table, so there is no automatic insertion of fuzzy matches, no automatic skipping of non-translatable section, etc. All manual work.
I would like to know whether it is only me thinking that this is an unfair practice to count text and pay for translations.
I would appreciate your comments.
Thanks
Katalin ▲ Collapse
| | | |
heikeb 
Ireland
Local time: 16:58
Member (2003)
English to German
+ ...
| look at the bottom line | May 27, 2005 |
Well, they have to count the source somehow. It doesn't make sense to apply two different counting methods for a "bilingual" text.
What they count - words, characters, lines or pages - doesn't really make any difference, as long as the amount of money you will receive in the end is acceptable for you. If you think the final amount will not compensate your work adequately, don't accept the job.
But I definitely wouldn't consider it unfair.
| | | |
| This is precisely my point | May 27, 2005 |
Heike Behl, Ph.D. wrote:
It doesn't make sense to apply two different counting methods for a "bilingual" text.
This is precisely my point. There should be no double standard.
They are applying two methods: paying per source word(=target word) for alphanumeric characters, and paying per character for Japanese characters. But they are paying the same rate!
Normally the per target word rate is 2.5 time higher than the "per source character" rate. In other words, the amount you get "per target word" is lower than what you expect based on your per source character rates.
If you think the final amount will not compensate your work adequately, don't accept the job.
This makes perfect sense, however, this counting method makes it impossible to give a rate quote without risking loosing out at the end. In other words, if the agency gives you a Word count for the source text, you really don't know how much text you have to work with, until you see the actual text. (If they work with character count, this gives you the upper limit in terms of the work quantity, and you can't loose out.) In a lot of cases, you don't get to see the entire text, only after you accepted the job.
And, the text is not "bilingual". It just has sentences like "Next, click on the PREVIEW button" - where everything except PREVIEW is in Japanese.
If the source was in German, and they would want to pay you 2.5 times less for words that will stay the same in the target English text, wouldn't you think that is a bit stingy?
[Edited at 2005-05-27 02:39]
[Edited at 2005-05-27 02:46]
| | | |
heikeb
Ireland
Local time: 16:58
Member (2003)
English to German
+ ...
| How else would you count? | May 27, 2005 |
[quote]Katalin Horvath McClure wrote:
This is precisely my point. There should be no double standard.
They are applying two methods: paying per source word(=target word) for alphanumeric characters, and paying per character for Japanese characters.
[quote]
The problem is that Japanese characters are not comparable to English characters.
One could rather argue that a Japanese character (kanji) is more like a word than a character when compared to non-Asian languages; a single character in German, for instance, doesn't have any meaning, whereas Japanese characters already carry meaning (even if some of them need to be used in combination with other characters to convey the desired meaning one could look at them as compounds).
How would you count "per word" in Japanese automatically? So for me, they are using the same counting standard for both Japanese and English. I don't think that counting each individual character of the English words and charging for them the same rate as for Japanese characters would be fair - even if these were words that needed to be translated.
Katalin Horvath McClure wrote:
This makes perfect sense, however, this counting method makes it impossible to give a rate quote without risking loosing out at the end. In other words, if the agency gives you a Word count for the source text, you really don't know how much text you have to work with, until you see the actual text. (If they work with character count, this gives you the upper limit in terms of the work quantity, and you can't loose out.) In a lot of cases, you don't get to see the entire text, only after you accepted the job.
Since the majority of the text is Japanese and you don't actually need to translated the English terms, the overall "word" count (i.e. number of English words and number of Japanese characters combines) should give you a pretty good indication of what amount of work to expect. How would counting English characters ("hidden" among the number of Japanese characters) help you to better estimate the amount of work?
Katalin Horvath McClure wrote:
If the source was in German, and they would want to pay you 2.5 times less for words that will stay the same in the target English text, wouldn't you think that is a bit stingy?
[Edited at 2005-05-27 02:39]
[Edited at 2005-05-27 02:46]
They don't "want" to pay you less for English words; this is a coincidental by-product of having to count a source text that combines two languages that apply the term "character" to very different concepts. Therefore, I don't see anything stingy in this approach. The alternative would be to count English words manually and apply a different rate for them - not very efficient. Do you have any other method for counting the embedded English words quickly and efficiently?
The issue is not that your client wants to cheat you out of your well-earned money, but that they need to have an efficient way to count the source text to put a price on the translation. We all know how arbitrary this price can be. I once quoted my standard rate for a technical text only to find out later that most of the text consisted of a list of technical equipment, i.e. no articles, no prepositions, no function words etc. Of course, it took me much longer than a "normal" text. Would it be fair to accuse the client of wanting to trick me? No, they just applied an automatic word count, nothing more.
Therefore, again, if you don't like the overall price, ask for a higher rate or don't accept the job - or ask them if they would consider paying per target word. But, quite frankly, I don't see how the presence of English words justifies a higher rate. If the rate they quote you is within in the range of the standard rate for Japanese texts, I don't think you have anything to complain about.
Think about the times you actually get a break from a client, a text very similar to one you have done before, a high number of repetitions (without Trados discounts), a text so easy you can translate 800 words an hour without even thinking... Are you playing it fair in these instances and offering to charge less? - I don't think so. It all evens out in the long run.
| | |
|
|
|
| No CAT for CAT count is more upsetting | May 27, 2005 |
The agency is paying my "per character" rate for each WORD. This means, that they pay less (approx. 2.5 times less) for strings that are written using alphanumeric characters, compared to text written using Japanese characters.
I found this pretty upsetting, it feels like they are trying to skim off the last possible layer from the translator's pay.
I can understand you are being upset here, but surely there are not that many of the 'untranslatable' commands in the text, and as was said before - it's a way of counting the text.
I find this a lot more upsetting:
pay 50% flat rate for fuzzy matches, and anything above 50% match is considered fuzzy, and they pay zero for repetitions and full matches but they do want you to mention if you find anything wrong with those. In addition, with this agency, no CAT tool is used during the translation process, the source is provided as a two-column table, so there is no automatic insertion of fuzzy matches, no automatic skipping of non-translatable section, etc. All manual work.
How can you apply CAT 'rates' for no-CAT work?? Unless you find a way to use a CAT tool anyway, how are you going to remember you translated this or that before? They must have done the word count in a CAT tool... I do not understand. I would not accept these conditions. FWIW, Anjo
| | | |
| Efficiency and invoicing | May 27, 2005 |
It strikes me that the time wasted trying to calculate a dual word/character-count system (using the source language) would far exceed the slight loss in profit on the words that aren't going to be translated anyway.
In your position I would either insist on a destination word count (meaning you are charging the full rate for words which you won't be translating but will probably have to move around and/or retype), or accept that it's all the luck of the draw anyway and accept the 4... See more It strikes me that the time wasted trying to calculate a dual word/character-count system (using the source language) would far exceed the slight loss in profit on the words that aren't going to be translated anyway.
In your position I would either insist on a destination word count (meaning you are charging the full rate for words which you won't be translating but will probably have to move around and/or retype), or accept that it's all the luck of the draw anyway and accept the 40% rate (which might still give you a higher total, depending on how "wordy" the source is).
In my view nobody benefits from overdoing the "precision" of the calculation. It's all swings and roundabouts! ▲ Collapse
| | | |
Heike Behl, Ph.D. wrote:
How would you count "per word" in Japanese automatically?
I would not, there is no way. Currently, there is no way to automatically parse Japanese text and correctly identify where one word ends and where another begins - and there are other tricks with the grammar as well, which makes the concept of "word" fuzzy in Japanese.
So for me, they are using the same counting standard for both Japanese and English. I don't think that counting each individual character of the English words and charging for them the same rate as for Japanese characters would be fair - even if these were words that needed to be translated.
I think everybody sets up their rates based on their experience with the amount of source text related to the amount of translation work. If you sometimes need to quote based on source count and in other cases based on target count, you want to establish a rate system where no matter which way you quote, you would end up with approximately the same sum, right? I don't know if there is a significant difference between the number of German source words and the number of English target words when you translate from German to English, but if there is, your "per source" and "per target" rates will be slightly different, right?
Based on my experience having translated many similar technical texts is that the ratio between the number of source characters (yes, this includes alphanumerics counted character by character) and the number of target words (where everything is alphanumeric and counted by word) is 2.5. This is how I have set up my source rate, which I quote "per source character" and my (2.5 times higher) "target word rate".
Most translators in this language pair do it the same way.
All of my other clients ask for quotes and pay "per source character", no matter how many alphanumeric strings are in the source text. By the way, Trados ONLY COUNTS characters in double-byte languages, it doesn't even give you a word count (because word is an unclear concept, and everybody in the industry uses character count - well, not everybody, as we see.)
This client was the first one that took my "per character" rate and applied it "per word", resulting (in this particular case - over 25% difference at the end).
The only way to prevent this is to come up with a new "per source word" rate, which would involve running some statistics on past translated texts. It is doable, but I don't want to waste the time, when the standard counting method is not this one anyway.
On another note, I don't think you answered my question:
Katalin Horvath McClure wrote:
If the source was in German, and they would want to pay you 2.5 times less for words that will stay the same in the target English text, wouldn't you think that is a bit stingy?
They don't "want" to pay you less for English words
I am not sure what they "want" or "don't want", but they do pay less at the end.
this is a coincidental by-product of having to count a source text that combines two languages that apply the term "character" to very different concepts.
That's why I asked the same question for German-English translations, to set aside the character concept. Have you ever met a client that paid less for words that showed up unchanged in the target text? What would you think if they did? (Let's assume they CAN count them automatically.)
The issue is not that your client wants to cheat you out of your well-earned money,
I am not accusing anyone of cheating. I am just questioning the fairness of this, given the fact that translation rates are usually established based on a different counting method. (See longer explanation above.)
Think about the times you actually get a break from a client, a text very similar to one you have done before, a high number of repetitions (without Trados discounts), a text so easy you can translate 800 words an hour without even thinking... Are you playing it fair in these instances and offering to charge less? - I don't think so.
Believe it or not, yes, sometimes I did charge less. On a number of occassions, I switched to an hourly rate, when I felt it was unfair to charge based on the text count, because I could do it very quickly. Needless to say, the clients appreciated it and they keep coming back.
All of my clients that have repetitive or similar texts coming up time to time, do use Trados, and do require discounts, so I don't "get a break" very often.
That's why ideas like this "new" counting makes me feel that somebody is trying to peel another layer of clothes off of me, when I am already practically naked.
| | | |
Login to reply/comment 
Translation news
