Sachchidanand: That is a really interesting article, and it does provide information about the "smile" emoticon, but there is a really big variety of them, and they comunicate different emotions; I wonder about the psychological responses they trigger, and the psychological process that makes us use them
I would think their purpose often, is to make things unambiguous. In face to face interaction we constantly receives feedback which provides a multitude of information and intonation is very closely linked to meaning. In text we cannot monitor feedback and we are as communicators (though there are individual differences), aware that we can use the same words to transmit many different meanings. Emoticons reduce ambiguity in many instances and provide a kind of 'intonation' which in speech would make the utterance less ambiguous.
One major reason is that electronic communications are written in a style that has been shown to be perceived as impolite, sarcastic, or having other negative tones. People don't start messages with "excuse me please but could you..." types of introductions as they would in speech, they state their points at length in a conversational tone, but do not allow for an actual dialog where turn-taking happens more rapidly, there are more "I agree with your point, but..." while spending much more content on the "but", again without the recipient being able to engage in conversation.
Many people do not have much need to write at length in their life, and are not used to constructing long messages as various professionals are. People who grew up before email was common communicated mostly by speech for much of their lives. Younger people find electronic communications to be as normal as a television. But in either case email, text messages, and website commentary function as a conversation without speech, but it lacks much the feedback that indicates respect or agreement for the other person.
Similar things happen with phone conversations due to missing facial expressions and uncertainty if one person is done speaking or is just pausing for a brief thought before continuing, etc. So people tend to have to explain away perceptions of negativity due to this lack of information, and often go through lengthy rituals of confirming things or ending a conversation, which would not be so long or ritualized in face to face conversations.
So emoticons help reduce these perceptions of negativity. Yes, they can be fake or sarcastic, etc, but people also lie and express overt or covert sarcasm in face to face or telephone conversations. The major reason for their use is to substitute for missing elements of conversations due to the format of the message.
Because of this, how much impact an emoticon can have depends on the same things in other conversations - how much do you know about the person? how honest are they? are they known for being sarcastic?
And this results in emoticons from strangers or persons who are not well know being more difficult to evaluate in the same way that any other speech from them would be. So context has an impact - is this person trying to sell me something? is this discussion happening on a website known for relatively intellectual discussion (NY Times) or one know for humorous comments (the Onion's AV Club)?
Many people also just think that they are cute and fun to use.
Here is a guide to smileys/emoticons, but it's in Dutch. I'm sure there are English versions, but I am feeling lazy and have the Dutch page bookmarked. Japan also has developed a system of emoticons that differs from NA/EU usage.
http://www.wimdaniels.nl/sms/pages/smileys.html
There may be interesting results with regard to universal/cross-cultural emotion recognition to see how people from non-computerized or non-literary cultures interpret emoticons.