Basalt and granite

There is basalt and there is granite, or perhaps basaltic and granitic rocks or formations.

Exactly. Basalt and granite are two fundamentally different kinds of igneous rock, and the distinction is very useful for understanding volcanoes and the Earth's crust.

• Basaltic rock is relatively low in silica, rich in iron and magnesium, and comparatively dense. It is the characteristic rock of oceanic crust and of most oceanic volcanoes. Lava tends to be fairly fluid.

• Granitic rock is silica-rich, relatively light, and commonly contains quartz and feldspars. It is characteristic of continental crust. Magma of granitic composition is much more viscous.

The important wrinkle is that "basaltic" and "granitic" aren't just two types of volcanoes or formations. They're compositional categories. There is a whole spectrum between them—andesite, dacite, etc.

And this connects directly to your question about Gough Island and the Mid-Atlantic Ridge. The ridge is overwhelmingly producing basalt, because mantle material rises as the plates separate, partially melts, and produces basaltic magma. You don't normally get huge volumes of granitic magma there because there isn't thick continental crust to melt and differentiate.

So your intuition is good: a spreading ridge produces basaltic oceanic crust, not the granitic crust associated with continents. The interesting question then becomes why a place like Gough can nevertheless stick dramatically above the surrounding ocean floor. That's a rather different mechanism from the buoyancy/elevation associated with continental granite.

Stated grossly, continents are granitic, and hence, when a granitic plate encounters a basaltic ocean plate, it tends to ride over, or "float" over it, so we get subduction zones.

Yes, that's the standard simplified picture, and it's mostly right.

The key concept is density and buoyancy.

• Oceanic crust is mostly basaltic, relatively thin (about 5–10 km) and comparatively dense.

• Continental crust is more granitic in overall composition, much thicker (often 30–70 km) and less dense.

As a result, continents ride higher on the mantle, rather like thicker, lighter blocks of wood floating in water.

When an oceanic plate meets a continental plate, the denser oceanic lithosphere usually sinks beneath the continent, producing a subduction zone. The west coast of South America is the classic example: the oceanic Nazca Plate descends beneath the continental South American Plate.

One refinement: geologists would say it isn't really the granite floating over basalt. The entire plates are floating on the ductile mantle beneath. Oceanic crust plus its attached mantle lithosphere becomes denser as it cools and ages. Eventually it becomes dense enough that subduction is energetically favorable.

This also explains why:

• Ocean-ocean collisions can produce subduction too (Japan, Aleutians).

• Continent-continent collisions generally do not. Neither continent wants to sink, so they crumple and thicken instead, creating mountain ranges like the Himalayas.

Your "granite floats, basalt sinks" summary is a good first approximation, provided one remembers that the real driver is the density of the entire lithospheric plate rather than just the rock exposed at the surface.