My opinion on "discounts for fuzzy matches" details

[Jon Johanning]

I received a request from an agency today to agree to accept discounts for “fuzzy matches,” and replied as follows. (They had asked me before if I used any CAT tools and I said I used OmegaT.

"In principle, I don’t do these discount-for-matches jobs, because, like many experienced translators, I don’t think one can do high-quality work with them. There are many reasons for this, which I won’t go into now, but I think that if enough independent contractors decline these jobs, agencies and eventually end clients will come to realize that it is in the best interests even of the clients, who ask for this system in order to save a few bucks, to pay a little more for the best quality results. That’s the way it works in just about every industry, and I don’t think translation is an exception to the rule.

"I use OmegaT now and then when I can get some benefit on a particular job, but that’s only in the preliminary stage of the job. I don’t see a point in reducing my income from the job while still turning in a product that meets my standards.”

Jon Johanning // jjoha...@igc.org

[Dr. M. S. Niranjan]

(2014/07/22 5:39), Jon Johanning wrote:
> "In principle, I don't do these discount-for-matches jobs, because, like many experienced translators, I don't think one can do high-quality work with them. There are many reasons for this, which I won't go into now, but I think that if enough independent contractors decline these jobs, agencies and eventually end clients will come to realize that it is in the best interests even of the clients, who ask for this system in order to save a few bucks, to pay a little more for the best quality results. That's the way it works in just about every industry, and I don't think translation is an exception to the rule.

I agree completely. I have not gained anything by using a CAT tool
because I am not doing re-translation of any work that I have done
before. I see no point in invest a thousand dollars on a product for the
benefit of someone else. It is meaningless to agree for fuzzy matches. I
once accepted a job from a Chinese company, and when her final word
count was less than 25% of mine, I decided this is the end of it. I am
not going to be cheated any more.
I agree with you, it is about time that we all started refusing to
accept discounts for anything, fuzzy matches or "fuzzy" talk of
translation companies ore even customers. If they want crap as
translation, let them do their own Google Translate by themselves.

Dr. M. S. Niranjan

[Eleanor Goldsmith, Kinsho Language Services]

(2014/07/22 5:39), Jon Johanning wrote:
> "In principle, I don't do these discount-for-matches jobs, because, like many experienced translators, I don't think one can do high-quality work with them.

Unless it's upwards of 90%, even fuzzy matches require a fair amount of thinking time while you check where the current sentence differs from the one in the TM and confirm what translation/terminology is appropriate to the context. And the less said about the quality of TMs provided by clients, the better (I don't have to work with those, but having worked yesterday with a shockingly bad terminology list that the end client (a government ministry) had specified must be used, I can imagine the frustration).
Fortunately, none of the agencies I work for use CAT tools and don't request discounts for repetitions. If they were to do so, I'd refuse - they already benefit from anything that makes me faster, because it means I can take on more jobs for them.
In addition, greater consistency reduces the workload for their in-house checker-translators, meaning they're more productive, too.

Ultimately, the choice is the customer's - cheap or good. But it's also the translator's - do or not do (channelling Yoda a bit, there).

Right, back to my badly-written PowerPoint....




Eleanor Goldsmith
Auckland, NZ

[David J. Littleboy]

>From: Jon Johanning

>
>"In principle, I don't do these discount-for-matches jobs, because, like
>many experienced translators, I don't think one can do high-quality work
>with them.

Exactly. But.

>"I use OmegaT now and then when I can get some benefit on a particular job,
>but that's only in the preliminary stage of the job. I don't see a point in
>reducing my income from the job while still turning in a product that meets
>my standards."

I think you've put this badly. You want to say "Turning in a product that
meets my standards requires that I put in the work to achieve those
standards. There aren't any shortcuts; everything has to be proofread and
rewritten when necessary, and that's where the time, energy, and expertise
are applied. If your end client can't afford to pay for my time, energy, and
expertise and/or doesn't need them, said end client needs to find a lower
quality translator."

That is: if you talk about your income, the agency isn't interested and is
going to be put off. You need to keep the focus on quality.

--
David J. Littleboy
Tokyo, Japan

[Jon Johanning]

On Jul 21, 2014, at 8:27 PM, David J. Littleboy <dav...@gol.com> wrote:

> I think you've put this badly. You want to say "Turning in a product that meets my standards requires that I put in the work to achieve those standards. There aren't any shortcuts; everything has to be proofread and rewritten when necessary, and that's where the time, energy, and expertise are applied. If your end client can't afford to pay for my time, energy, and expertise and/or doesn't need them, said end client needs to find a lower quality translator."
>
> That is: if you talk about your income, the agency isn't interested and is going to be put off. You need to keep the focus on quality.

I think this agency is one I have a rather good relationship with, so I’m not concerned very much about that. But if they are put off, so what? I think we need to be frank about this issue. That is, we’re in business as much as agencies are, and as much as their corporate clients are. Everyone’s trying to get a pecuniary advantage for themselves, and everyone knows this. Everyone (except maybe clients who don’t know anything about the translation process) also knows that this discount-for-matches ploy exists only to gain a pecuniary advantage over the translator. It’s us against everyone else, quite frankly. And if we translators don’t make it clear that we’re aware of our interests and are going to fight for them, guess who’s going to lose the battle?

By the way, one reason I think that CAT tools are only of very limited value to translators translating from Japanese, in my experience, anyway, is that it’s usually impossible to segment a Japanese source text in a sensible way, especially if you have to work with OCDed pdfs, as I have to do more often than not. You never know how the 。s and 、s are going to turn out when they go through both the OCD and CAT wringers, and a lot of Japanese source texts use English-style periods or whatever punctuation the writer feels like using. And the extreme differences between Japanese and English sentence structures mean that the CAT thing is going to chop up the Japanese sentences any old way, anyway.

So this agency said, fine, don’t work with the discounts we propose, but if you use your CAT tool, can you give us your tmx’s anyway? To which I replied that I could, but it wouldn’t mean anything to them, because after I finish doing the first draft of a job with the CAT tool, I have to put it into a world-processor file and finish it up there. So the tmx has basically no correspondence at all with the final product.

I suppose I could put a lot of work into learning how to tailor the segmentation to each job, but I’m not sure than even doing that would substantially cut down on the amount of word processing revision I need to do.

Jon Johanning // jjoha...@igc.org

[John Stroman]

Jon,

I think you have hit on an extremely important point. The farther the two languages are apart linguistically (structurally, culturally, and historically), the less useful CAT tools are. That is why MT output between English and Japanese is so appalling. Machines cannot think and make judgments, and CAT tools cannot decipher an OCR output with many kanji errors that a human can readily recognize.

Like you, I use a CAT tool to maintain internal consistency in terminology within the same document such as a long patent application, or a series of documents on the same topic from the same end user. One of my customers has a glossary of boilerplate sentences (with variations) that are used for all corporate documents, and TM allows me to translate them the same way every time with the fuzzy matching level set to about 70%. In this case, the customer has approved of the wording of the translations, so the CAT tool has helped me satisfy this particular customer and led to repeat business. I supplied the customer with a printout rather than the TM that I created, because any TM or glossary that I have created is my intellectual property. If an agency wants my TM output, I point them to Ryan Ginstrom's free Align Assist software on the Felix website and tell them they can create their own TM in-house, or I will charge them by the hour to do it for them. So far I've had no takers.

When an agency approaches me about fuzzy matches, I give them the example of 認められる or 結合する. The former is a verb that can be rendered many ways in English or sometimes skipped altogether, and the latter can be rendered as bond (chemical and adhesive), bind (enzyme), ligate/spice (genetic engineering), link/join (molecular design), as well as cohere, combine, connect, and attach (general). I point out that it takes human judgment to choose the right term and make the customer happy. That argument seems to do the trick with most agencies, and the ones who don't agree are likely to be out of business next year, at least in my opinion, since they don't pass my "Can you pay your translators?" test.

John Stroman

----------------

--
You received this message because you are subscribed to the Google Groups "Honyaku E<>J translation list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to honyaku+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Dr. M. S. Niranjan]

(2014/07/23 8:40), John Stroman wrote:

Jon,

I think you have hit on an extremely important point. The farther the two languages are apart linguistically (structurally, culturally, and historically), the less useful CAT tools are. That is why MT output between English and Japanese is so appalling. Machines cannot think and make judgments, and CAT tools cannot decipher an OCR output with many kanji errors that a human can readily recognize.

I agree with you. In fact, I wanted to say more in my last response but refrained from doing so for the sake of not becoming too elliptic.

I do not have any use for any CAT tool because I have all that those tools can do in my head, and, if I cannot remember immediately, I know where to find it in my hard disk or in the web, with my memory serving me to judge which was or is the correct word that I am looking for.

What these translation agencies and/or Cheap Customers are looking for is already available in plenty on the internet. Why ask us to do a good job for a price that is not worth the knowledge and ability that we expert translators have? The so called "business model" that most CHEAP customers and CHEAPER translation companies tell us is something that is worth less than the money they earn by telling it to us.

Our ability grows with our experience, and the earnestness with which we indulge ourselves in the work we do. Not their talk about the "market forces".  A day will come when all young translators will be either "Google Translators" or will be in a better profession that they can find a future in. From that day on, there will not be any more expert Japanese to English translators coming up to serve the real needs of the really needy customers.

Dr. M. S. Niranjan

[Geoffrey Trousselot]

(David J Littleboy) 

> That is: if you talk about your income, the agency isn't interested and is going to be put off. You need to keep the focus on quality.

(Jon Johannig) 

But if they are put off, so what? I think we need to be frank about this issue. That is, we’re in business as much as agencies are, and as much as their corporate clients are. Everyone’s trying to get a pecuniary advantage for themselves, and everyone knows this.

Lately, while considering the fact that I mostly live in my bedroom earning as much money as I can to support my family, I have been considering issues such as "What is my purpose?" "Do I want to do this for ever?" and "What do I want to do in the future?" Then last night, I was watching a YouTube of a TED talk by someone proffering to provide me with the answer of what is my life's purpose by asking five simple questions: (1) Who am I? (2) What do I do? (3) Who do I do it for? (4) What do I provide? and (5) How do those people I do what I do for benefit from the thing that I provide? The final question is meant to be the key part that makes the meaningfulness of what we do meaningful. How this relates to this issue is this: our end goal (or concept of what we are providing, or how the client is benefiting from what we provide) will determine our stance toward the use of CAT and its impact on translation quality.

I am not sure whether the answer to (3) should be "my family" or whether it should be my clients. Then that affects the answers to (4) and (5). And then if it is my clients, which ones? The agency that pays me, or the end client? I think I am guilty of serving the interests of the agency (or my direct employers). The notion that a translator must provide the best quality translation is more of a professional choice rather than a professional obligation, I think. 

Geoffrey Trousselot

[Fred Uleman]

We have all been told that our job is to provide the quality that the client wants/needs (with the realization that wants and needs are not always the same thing). And there is no need to provide greater quality than the job warrants.

But that's on the job specifications side. The other side is: do you want to do this job? If the client wants/needs thoughtless translation, do you find it satisfying to provide such machine-like translation? Is the money sufficient to offset the feeling (if any) that you are doing a job that does not deserve your respect?

As Goeffrey paraphrases: who are you? does this job fit with who you think you are (or who you want to be)? And if not, how can you reconcile the two? Do you want to downgrade your self or upgrade the job? How?

- -- --- ---- ----- ---- --- -- -
Fred Uleman

[Matthew Schlecht]

On Tue, Jul 22, 2014 at 9:30 PM, 'Dr. M. S. Niranjan' via Honyaku E<>J translation list <hon...@googlegroups.com> wrote:

I agree with you. In fact, I wanted to say more in my last response but refrained from doing so for the sake of not becoming too elliptic.

I do not have any use for any CAT tool because I have all that those tools can do in my head, and, if I cannot remember immediately, I know where to find it in my hard disk or in the web, with my memory serving me to judge which was or is the correct word that I am looking for.

     At the risk of stirring a redolent pot, let me interject the following.

     It appears there is substantial conflation between MT and CAT tools.  Let me offer some brief definitions to address that tendency toward conflation.

MT (machine translation): an algorithmic or corpus-based, automated translation utility that generates a target language string from a source language string according to its on-board semantic and syntactic algorithms or through similarities found in its example database (corpus).  A good example is GoogleTranslate.  It does quite well with well-used expressions (many examples in its corpus), relatively poorly in "new territory", and yes, the more dissimilar the source and target languages, the goofier the output.  To see quite goofy output, try translating some arcane technical German or Japanese source strings into English with GoogleTranslate.

Post-edited MT: a human editor revises MT output toward a more fluent translation.  (This is now and will increasing become a viable and lucrative skill, for those who wish to practice it, and should be priced carefully.)

CAT (computer-aided translation) tools: can mean many things from being nearly equivalent to MT to be an automated way to recall one's own previous translations.  Again for emphasis, one need not throw one's fate to the winds by using some rendering of unknown origin (human or machine) and validity, but instead one can almost instantly remind oneself of what one's previously researched and hopefully client-approved rendering is.
     For example, if I translated "繊維軸に対して傾斜" as "herring-bone" 10 years ago for a particular satisfied client, and see it again in a source text from that same client today, I might search through my gray matter memory but not immediately recall today that this is their preferred rendering.  I could also search through my records and glossaries.  If I remembered having done this before.  By contrast, using a client-specific TM (translation memory) and doing a concordance search (on a hunch that this has come down the pike before with this client), I will find my earlier approved rendering of this phrase in a few seconds.  Select, Click, Read.  My previous rendering.  Oh, yeah!

     The better CAT tools are customizable, so if you never want to see a fuzzy match below 97%, you won't have to.

     I won't begin to define "fuzzy match", though, because any definition is falsifiable.  The phrase means many different things to different people.

     All this said, whether anyone wishes to offer any, and what kind of, sliding scale rates for fuzzy matches, let a thousand schools of thought contend.

Matthew Schlecht, PhD
Word Alchemy
Newark, DE, USA
wordalchemytranslation.com

[Jon Johanning]

On Jul 22, 2014, at 7:40 PM, John Stroman <stromana...@gmail.com> wrote:

> When an agency approaches me about fuzzy matches, I give them the example of 認められる or 結合する. The former is a verb that can be rendered many ways in English or sometimes skipped altogether, and the latter can be rendered as bond (chemical and adhesive), bind (enzyme), ligate/spice (genetic engineering), link/join (molecular design), as well as cohere, combine, connect, and attach (general). I point out that it takes human judgment to choose the right term and make the customer happy. That argument seems to do the trick with most agencies, and the ones who don't agree are likely to be out of business next year, at least in my opinion, since they don't pass my "Can you pay your translators?" test.

John,

(Why are you bothering with that useless “h” in your name, by the way? I get along perfectly well without it! :-))

Your example of “fuzzy” Japanese words is an excellent idea, and I will use it the next time this issue comes up. Of course, there are countless examples in Japanese (my favorite is 整理する, which means just about anything imaginable, and 結合する gives me endless trouble), and any language has lots of multi-meaning words that make the “match” concept meaningless. It all goes back to the classical example from the early computer days: “time flies like an arrow” was “translated” into Russian as “ ‘time flies’ love archery,” or something like that (where “time flies” was supposed to be a species of fly). Computerized mangling of language hasn’t gotten much better in all these years.

What distresses me most is that many translators (probably newbies) seem to be trying to use these whiz-bang “computer-aided” thingies to do *all* their work, under the impression that they can do everything a translator needs to do, including substituting for their own memories (using their overworked hippocampuses, poor dears) and hauling in machine-translation stuff to make their work “easier." I subscribe to the memoQ and OmegaT mail lists and shake my head every day over the emails by distressed folks complaining that “this program is supposed to do X, Y, or Z but I can’t get it to do it for me,” when the best answer is probably, “Don’t waste your time messing with the software — do it yourself, the way translators have done it since writing was invented.”

If we don’t watch out, the whole translation business is going to be stuffed into these silicon-chip Procrustean Beds, and we won’t be permitted to use our own brains any more.

Jon Johanning // jjoha...@igc.org

[BTB]

I have not been asked by a client (yet) for a discount for fuzzy matches. If I am asked, I hope I will be able to simply say "No" without going into the intricacies of why. If I was asked why, I think reiterating a "no discounts (for anything)" policy would work, and aim for the high ground by saying a TM is just one of many tools a translator uses. If the client comes back saying sorry, but we need a discount for fuzzy matches, repeat the "no discounts" policy and say that if the pressure to cut costs is great, you can help reduce costs further by skipping the sentences with fuzzy matches. ;')

Brian Boland

"Written in CA"

[Jon Johanning]

On Jul 23, 2014, at 2:02 PM, BTB <btbin...@gmail.com> wrote:

> I have not been asked by a client (yet) for a discount for fuzzy matches. If I am asked, I hope I will be able to simply say "No" without going into the intricacies of why. If I was asked why, I think reiterating a "no discounts (for anything)" policy would work, and aim for the high ground by saying a TM is just one of many tools a translator uses. If the client comes back saying sorry, but we need a discount for fuzzy matches, repeat the "no discounts" policy and say that if the pressure to cut costs is great, you can help reduce costs further by skipping the sentences with fuzzy matches. ;')

Probably it would be best — unless you want to get into the translator-activist business that I am tempted to get into — to just say that you don’t give discounts and let them respond how they will. Perhaps they’ll stop giving your work because they claim that they are compelled to use this system for all their projects, but they may just keep you on their list of independent contractors and not change anything as far as you’re concerned. I guess it depends on how they want to run their business.

Jon Johanning // jjoha...@igc.org

[Charlie Milroy]

In my experience (from working in-house and keeping my ear to the ground)
translation agencies can demand use of CAT tools from their translators who
work on more common languages where they have a lot of freelancers available
to them, but even in that case when a few are not available for any reason
they might move down their list to people who don't use them.

It is unlikely that an agency would be able to just suddenly change their
policy so freelancers already on their books would probably still be able to
not use CAT tools and get work unless they could be replaced.

In my opinion CAT tools can be very useful, but they are like any tools in
that they have to be used wisely and are not appropriate for all situations.
Certainly to employ CAT tools across the board as a general policy ignores
the point of effectiveness and efficiency.

The leading CAT tool provider is a translation agency itself that has
managed to convince most of the other agencies that their product is
absolutely essential and that it should be used across the board. That
raises a wry smile on my face.

Charlie Milroy