At school, (Christ's Hospital,) I enjoyed the inestimable advantage of a very sensible, though at the same time, a very severe master, the Reverend James Bowyer....I learned from him, that poetry, even that of the loftiest and, seemingly, that of the wildest odes, had a logic of its own, as severe as that of science; and more difficult, because more subtle, more complex, and dependent on more, and more fugitive causes. In the truly great poets, he would say, there is a reason assignable, not only for every word, but for the position of every word; and I well remember that, availing himself of the synonymes to the Homer of Didymus, he made us attempt to show, with regard to each, why it would not have answered the same purpose; and wherein consisted the peculiar fitness of the word in the original text.[Samuel Taylor Coleridge, Biographia Literaria, Chapter I]
And the meaning of fine words cannot be made 'obvious', for it is not obvious to any one: least of all to adults, who have stopped listening to the sound because they think they know the meaning. They think argent 'means' silver. But it does not. It and silver have a reference to x or chem. Ag, but in each x is clothed in a totally different phonetic incarnation: x+y or x+z; and these do not have the same meaning, not only because they sound different and so arouse different responses, but also because they are not in fact used when talking about Ag. in the same way. It is better, I think, at any rate to begin with, to hear 'argent' as a sound only (z without x) in a poetic context, than to think 'it only means silver'. There is some chance then that you may like it for itself, and later learn to appreciate the heraldic overtones it has, in addition to its own peculiar sound, which 'silver' has not.[J. R. R. Tolkien, Letter 234 to Jane Neave in The Letters of J. R. R. Tolkien, Carpenter and Tolkien, eds., Harper Collins (London: 1990) p. 310]
There's a lot of poetry here at Siris. There are many reasons for this, but one of them is purely philosophical -- since graduate school I have been irritated by philosophers of language whose conception of language is poorly suited for accounting for the existence and facts of poetry, which is one of the most obvious and important of all linguistic phenomena. There are works in the philosophy of language that lead one to suspect that the authors have never had any more poetry than they had been forced to read by the end of high school. And this is a problem. It is in poetry that one really puts language to the test, as it were, probing its limits and drawing new functions and nuances out of it, turning it on itself and on its head, playing with aspects of it that are often ignored. It is, of course, not only poets who do this -- but it is one reason why there's a never-ending river of poetry here.
Poetry deals, among other things, with finer shades of meaning. As Tolkien notes in the above passage, 'argent' and 'silver' are synonyms, but they are not, in fact, used in the same way. Substituting 'silver' for 'argent' is unlikely to affect the truth value of a statement, but it is likely to affect the "fitness of the word" that Bowyer, as described by Coleridge, attempted to show. Thus, as Tolkien again notes, 'argent' has "heraldic overtones". We would also often recognize 'argent' as being in a higher formal register than the more common 'silver', and as having poetic associations that need not be borne by the more common word.
One way to put this might be to say that 'argent' and 'silver' label closely analogous classes in what have come to be different systems of classification, and part of the difference between the two lies in their suggestion of which approach to classification is in play. We have to say that it is only a suggestion, since we can perfectly well transfer across classifications, as when someone uses overly formal words for comic effect, but our ability to do this depends on there being a difference in the first place.
Philosophy of language for a considerable portion of the past century has tended to be influenced by the Fregean division of meaning into force, sense, and tone, with the lion's share being concerned with sense. The above differences of meaning would usually get classified as differences in tone (coloring or shading, in Frege's own preferred manner of speaking), and that would be the end of it, since in practice 'tone' usually works as nothing more than the wastebin -- you throw in the things that you aren't using and never consider them again. This is problematic, since, as I've noted before, there's no reason to think that the same kinds of things get put into the wastebin each time. Something that can be treated as a mere difference of tone on one interpretation or occasion can, on another interpretation or occasion, be interpreted as a significant difference of sense. Whether or not the difference between 'cur' and 'dog' matters for the truth value of a statement depends in part on context and how one chooses to take the words in that context. The same is true of common derogatory terms, racist epithets, euphemisms, and the like. Again, it's a matter of classification: if you are classifying someone as a Norwegian, that's a different way of classifying them than if you classify them as a Noggie, because the latter includes as part of the classification a negative evaluation that is missing from the more neutral term, one that we can either ignore or take to be important. And doing so affects whether what is being said counts as accurate or inaccurate.
But even when the difference does not affect the truth value of a statement, it's an error to think that the difference is really insignificant. It can still affect the "fitness of words", their appropriateness to their use and context. In communication, there are always many ends, and some words will just fit those ends better than others in a given case. Unlike discussions of sense and reference, however, in which meaning is artifically divided in order to keep all the easy parts of the meaning together and to make it possible to ignore all the subtle parts, taking seriously the fitness or unfitness of words requires careful comparative work and also poetic experimentation.