This isn't my area of expertise, so I hope someone will correct me on the physics if I'm wrong: if an object is travelling along a plane at constant velocity, there will be no acceleration to measure. There will have been measurable acceleration to get to that speed, so you should be able to estimate velocity and direction based on PAST data - but there's nothing to measure otherwise, unless the velocity changes. However, if you add a solid-state gyroscope to the equation, when you are moving against the against the axis of the gyro, there is always something to measure.
I don't think I've seen a Freescale accelerometer & gyro in a single device - the only combined sensor I know of is the FXOS8700CQ, which incorporates an accelerometer with a magnetometer for 6-axis sensing.
So - if you were to grab a Freescale I2C accelerometer, you could also hang something like ST's L3G4200D of the bus too, and work with the combined data. Or use the FXOS8700CQ plus the L3G4200D for 9-axis sensing!
You can get mounted gyros from the likes of Sparkfun and SeeedStudio for prototyping purposes.
Hope this helps!
EDIT: this may help you, but it's way too mathematical for me! https://community.freescale.com/community/the-embedded-beat/blog/2012/03/29/degrees-of-freedom-vs-ax...
