I am totally breaking my own rules and double-posting today because I think these ideas are complimentary enough that they would suffer from a split across the weekend. This won't make a huge amount of sense without the post prior to it, so if you haven't read that yet, start there.
Feng Shui's skill system was one that I really came to appreciate as I started digging into other games. It had a very clever element to it that let a reasonably short skill list feel suitably broad yet cinematic. The idea was simple: each skill actually represented three things. The first was the skill itself, the things you would expect to fall under the skill. Drive let you drive things, Guns let you shoot guns and so on.
The second was your knowledge about the skill. That is, your Guns skill also encompassed your knowledge about guns, ballistics and all things gun related. You might not be a scientist, but you could answer serious metallurgical questions if they had to do with bullet composition or gun barrels.
The third was that it encompassed how connected you were within the social network surrounding the skill. That is, your Guns skill represented how well you knew and could find gun makers, gun smugglers, black market gun dealers, the location fo the nearest gun show and so on.
I adopted this idea in a number of games, adding a 4th element: perception. Your Guns skill might not help you spot a footprint, but it would let you recreate a firefight from evidence or spot a sniper. It worked decently, better in soem contexts than others.
Recently, Brennan Taylor has taken this idea and nitro-injected it for his game Bulldogs. In Bulldogs, he has structured his skills explicitly in terms of these broad categories of action. It's pretty slick.
Anyway, I like this idea a lot, and it's one of the best ways I know to distinguish between broad an narrow skills. A narrow skill just does a thing, a broad skill has a whole array of associated things (knowledge, connections, perception and the like) with it.
Now, let's take this back to the skill system as we've proposed so far, with it's increasing narrowness of scope. It's foundation - the culture skill - is one that explicitly depends on context. By creating the culture skill, you are implicitly creating that culture in your setting. You are saying things about the culture based on what the skill does. This is pretty potent, and I intend to use that potency over the course of the game.
Specifically, I intend to make it a necessary part of advancement. That is, I am going to add an extra step between "Soldier 2d6" and "Musketeer 3d6", and that step is the explicit creation of context.
That sounds fancy, but in practice it's much simpler - it requires taking the skill, which is fairly abstract, and concretely nailing it to the setting with specifics. In this case, the specifics might be what military the soldier is serving with, such as "Sargent of the Army of the Republic 2d6" or "The Queen's Guard 2d6".
This difference is easy to point to in the fiction, but it also has the mechanical impact of turning the skill from a narrow one into a broad one (see, there was a reason for that whole preamble). It also now opens the gateway to buying a specific skill at the next die-step up, which also indicates the context within which the character's skill is exceptional.
Doing this as advancement is simple enough - the context should be something that evolves out of play, but doing this as a part of character creation offers an extra bonus: Players may _create_ these setting elements as part of character creation. In effect, character creation can become setting creation.
Obviously, this isn't required. The GM can have a list of contexts to pick from if so desired. Heck, I'd suggest having such a list as a starting point, then letting players come up with exceptions, or drive you to come up with something off the top of your head when they really need a context for the best bakers in the kingdom.