Update on the function of Broca's area: we still don't know much.
It seems like everyone is studying Broca's area for any number of functions. The most recent is the Science paper that David commented on a while back. That paper argued that Broca's area supports sequential processing of lexical, grammatical, and phonological information. Like other claims about Broca's area, this paper has sparked a debate. For example, see the e-comment by Matt Goldrick et al.
But Broca's area is also a prime suspect in the hunt for mirror neurons, speech perception, syntactic movement, hierarchical structure processing, semantic integration, working memory, and cognitive control. What's interesting (or unfortunate) is that some of these debates go on without reference to others, like they are living in a parallel universe. E.g., you never hear the mirror neuron folks talking about cognitive control and vise versa.
New rule: when speculating about the function of Broca's area you have to at least mention the range of other ideas/data. Maybe this will promote cross-universe interaction.
New rule #2: don't use the term "Broca's area" without further specification. We all know that Broca's area is composed of at least two subregions that seem to do different things. Please specify.