Comparing Adium, Fire, and Skype, first you get some logo variation.
The cute little ducky in Adium gets my vote, mainly because it doesn't try to suggest anything obvious about communication and jumps around in your dock (mac users) whenever you get a message. Plus, it's a bird.
What's significantly different about the applications, though, is the way they bundle or fail to bundle consecutive utterances by one speaker. In all three applications, if I say something, you say something, and I say something, we get a visual representation of three utterances in dialogic relationship. Only in Fire, though, is it the case that if I say something, then I say another thing (as I'm often prone to do; hyperlalia does extend to computer mediated communication), then you finally get a word in edgewise, three utterances are presented. So in Fire, this kind of exchange could look like this:
metaspencer: have you seen the ducky?
metaspencer: have you seen it? have you?
you: enough with the ducky already!!!
What's interesting about Adium and Skype is that the first two utterances, when not interceded with an utterance by the other discussant, are bundled together by background color instead of being separated from one another by line break. This grouping makes them appear as if they are part of the same speech bubble, to draw from the visual rhetoric of comic books, or the same speech act, to draw from everyone's dear pals Austin and Searle.
As Fire presents such an exchange, I say one thing, then I say another thing, then you say something. As Adium and Skype present such an exchange, I make one articulatory move composed of two utterances, then you respond. This is a significant difference, I think, because of the way pithy short phrases are normalized in IM discourse. I mean, some people write multi-line messages—possibly proof read them even!—and then click send. But most of us go for just a few words, as that keeps the dialog flowing. With the bundling of multiple utterances by the same speaker in Adium and Skype, a speaker who types one thing, then another, then even another before being responded to seems like less of a conversation hog.
Conversation hog, as you may know, is a highly technical term in conversation analysis.
I bring this up because this is one of the only ways I've seen an application representing computer-mediated-communication in significantly different ways that influence meaning.