A few things to remember when looking at CFD diagrams like this one from the Wikipedia article:

1. Whether the fluid is water or air makes no difference at moderately low air speeds and at very low water speeds. In both cases, the fluid can be considered non-compressible, and from looking at the diagram, there would be no way of knowing whether we are seeing airflow or water flow, if we did not know the scale -- i.e., the flow speed relative to the size of the object around which the fluid must flow. (More correctly, we need to know the Reynolds number, which also involves viscosity -- but that is a complicating fact that does not aid understanding of the basic principles, I think.)

2. The very low pressures associated with cavitation in water flow are not likely to occur at these water speeds.

3. If you imagine yourself a CFD machine, you could consider the fluid to be a bunch of thoroughly lubricated BBs. For each BB, your job as a human CFD machine is to determine in which direction and at what speed each BB will move.

4. Common sense tells you that the BBs approaching from the left in the diagram will "stack up" and generally be affected by the object long before they reach the object.

5. We know that each BB will move in response to a pressure differential -- in other words, it will tend to move from regions of high pressure to regions of low pressure.

6. We also know that Coanda effect works -- in other words flow tends to cling to surfaces, following the surface curvature.

7. If you suspend disbelief for a second, and simply accept the speeds shown, we know that high speeds mean low pressure, so any BBs even remotely close to the high velocity in the middle of the throat are likely to flow toward that generally lower pressure.

8. We also know the the BB's have mass, so that there is a competing tendency for the BBs to simply go straight.

9. Given all that, the flow looks "about right" if you ignore the flow lines (fourth from the top and fourth from the bottom) where there is a serious kink in the flow. If, for a second, you ignore that pair of flow lines, then you can see that the top to bottom dimension at the left of the diagram of the flow lines that go into the throat is about twice the dimension of the throat. That means that the areas involved are in the ratio of 4:1. So the flow through the throat should be on the order of 4 times as fast as the flow leading into the throat from the area of free stream velocity (which is out a little beyond the left side of the diagram).

10. So, as drawn, (and even if the venturi were constructed of continuous solid walls instead of vanes) you'd expect it to work about as the article describes. The vanes can enhance the flow, by accelerating flow around each slot (working like slotted flaps).

11. The peculiar flow around the leading edge of the first vane, where the local angle of attack is nearly 90 degrees, is simply the result of the CFD program answering each BBs question "Which way should I go?" and it bases that answer on momentum (go straight) vs pressure differential (turn and go to the area of low pressure.) For each BB alone, that is a question that a human could answer on one side of a napkin. Given the interdependence of BB motion, and the huge numbers of BBs, combined with the three dimensional nature of flow (of which we are seeing just one slice) you can see that a human would have to fill up thousand of napkins with calculations to come anywhere close to the actuals flows involved.

12. If we zoomed into the diagram to be able to see the flow around the leading edge of the first airfoil, we'd expect to see a quite large low pressure peak just behind the leading edge on the generally low pressure side (i.e., the side facing the inside of the throat.) That very low pressure zone would help encourage BBs to make the turn into the throat.

13. If we made the "back" (exit side) of the venturi smaller, then the region of high pressure (low speed) outside the venturi would be smaller, and so less flow would go through the throat, and more would simply bypass the venturi entirely. Make the back end small enough , and you get increasingly close to a straight pipe, which would have no venturi effect at all.

This is the truth, the whole truth, and nothing but the truth (provided you ignore the parts I made up , misrepresented, over-simplified, over-complicated, or left out).