Our first duty is to convey the same message as the original. Like it or not, the reader is interested in knowing what the author said, not in what we think. It is well known that all translations involve a degree of distortion, that a translation always adds something to the original and subtracts something from it. A good professional translator should minimize distortion, the opinion of several academics notwithstanding.

To improve or not to improve, that is the question. We do not see much harm in patching up a badly written text, and, to be honest, more than once we have pretended not to see "principle" written where Mr. Fowler would claim "principal" should be used. But redoing the whole thing in our fashion is surely out of bounds.

The point at issue here, however, is the message, the content. If the original is a nasty piece of work full of racist prejudice, those characteristics should also be seen in the translation. Watering the text down to make it more palatable is outright lying. What the author said may be a bunch of lies, but it is true that the author said them and readers are entitled to know that and take such measures as may be appropriate in their opinion.

The same goes for the colleague who deleted a few sentences from a baby-care manual because she did not agree with them. She should have attempted to write her own book instead.

Ambiguity is our enemy. Murphy's first law of translation says that if a phrase can be understood in two different ways, readers (including editors and professional critics) will always interpret it the way that makes the sentence read like a mistranslation.

Once, many years ago when Danilo was a young translator, an editor found a phrase that was not to his liking and "corrected" it. In so doing, the editor entirely distorted the meaning. Danilo was very angry, but in fact it was his mistake, not the editor's; the phrase was ambiguous, allowed two readings—and the editor, as Murphy would have it, followed the wrong path and made the wrong correction. It is quite possible that many readers would have mentally done the same.

Fortunately, the error was found and corrected in time and had no further consequences, except making Danilo afraid of ambiguity.

Since our job is to transmit someone else's message as accurately as possible, the only reason to present a translation with more than one interpretation is to reflect ambiguity found in our source text, a task far more difficult than most of us would believe.

This is the tricky step. When the first draft of these 7 Steps was published in Danilo's blog, this was the point that drew fire from readers. They claimed that all translations should be grammatically correct. Yes, we agree, up to a point, that is. In certain cases, grammatically correct text will simply fail the rule set out in Step 1, because it will not convey the message. More often than not this will happen in literary texts, where violations of the grammatical norm may be used to convey a message. Otherwise, translations should conform to grammatical norm.

What we mean here is that once a translation conveys the message and is unambiguous, the time has come to bow to the purists, to those people who keep telling you that it is wrong to use a preposition to end a sentence with.

To the extent possible, that is after the three above conditions are met, a translation should reflect the style of the original. Don DeLillo and Ernst Hemingway should read differently in translation—as they do in the original. It seems to be the fashion among a certain type of publishing house to impose a manual of style that should better be called a Procrustean bed. The outcome is a perfectly correct and readable text which also is absolutely flat, tasteless and bloodless. It does not matter who you are reading; the style is always the same.

The problem is, of course, what type of style can be used in the target language to reflect the author's. Oh, well, we will deal with this some other time. Perhaps.

The concept of flowing is very fluid, so to say. Very subjective. What flows well for you may not flow well in the opinion of the next reader. Nobody said translation is an exact science, you know. Except for those people who believe in machine translation, but that's another story.

It's horrible to read a translation that does not flow well, a bumpy text that reads like a translation. We know that some people advocate exactly this type of work, translations that are hard to read and that smack of the syntax of the original. Frankly, we do not care a hoot. Our clients love translations that flow well and so do we.

Some people consider translating a fine opportunity to demonstrate their sapience and vocabulary. They think they are justified in using any word that can be found in the largest dictionary of their language—and that the reader will have that dictionary handy and use it all the time.

This does not mean that you should limit your vocabulary to the three thousand most frequently used words of your language. It means that some words suit Immanuel Kant but not the blockbusters people read in airport waiting rooms.