For me it is different: I see the differences easier when I have them side by side, as in a bilingual printout. Then I wonder why the 'target' language version is different and whether it is something that is part of a pattern or just a dubious formation in the original (which always is a possibility!), and ideally I should then look it up in a dictionary or check it through Google, but in practice this is the exception - I normally just notice it and continue. But being primed in this way will make it easier to spot the same thing later on.
But there is one snag to this: the text I use for intensive study are short, and they can't
possibly contain all the formulaic formulations I need, so in practice I also have to rely on the the unstructured and inefficient sucking-up process during extensive reading - however my impression is that I don't suck many valuable phrases and structures up before I already am a relative proficient reader.
I like to search for the
logic behind different ways to express yourself in different languages, and an important part of the learning process is to formulate explicit hypotheses concerning the different mechanisms and test them out - but as I have written in another thread you will ultimately have to feel grammar as something semantic, not as some kind of mathematics.
And idiomatic expressions? There is almost alway some logic behind them, although it may have become almost opaque with time. That's why I separate translations from a weak into a strong language from translation the other way. When the target is a strong language you don't have to pretend that you are learning it, and here I 'think' not only idiomatic expressions in the form of hyperliteral translations, but as far as possible also syntactical patterns. However the other way round, from a strong language into a weak one you need to collect "ways to say things" almost as words, or in other words: for everything you might want to say you need something you can use in the weak language, and it takes time and diligence to build that store house piece by piece.
jeff_lindqvist wrote:Dict.cc used to be (!) a good resource when I was working on my German many years ago.
To me the formulation here (with its exclamation sign!) indicates that dict.cc isn't a good source any more. With "was" (the simple past tense) it might still be a good source. Or maybe it really has deteriorated so much that the compound past tense is necessary? I don't use it so I don't know ..