According to Francis Heylighen:
“There is an interesting new method to measure how happy or sad people feel by analysing the “feeling” words that they use in their texts (blogs, song texts, presidential addresses, …).
Common words are given a “valence”, i.e. an average value on a scale ranging from most happy (e.g. “love”, “hope”, “proud”, “hug”…) to least happy (e.g. “death”, “lonely”, “rape”, “afraid”, …). The average valence of a text is calculated as an average (weighted by frequency) of the valence of the words it contains.
By analysing millions of blogs it is possible to measure the average valence on any given day, or in specific groups, like countries, age groups, or genders. These valences can fluctuate quite a bit, being high on days like Christmas or the Obama election, and low on days like anniversaries of 9/11 or the death of Michael Jackson. They also tend to be lower for teenagers and pensioners, and higher in middle age.”