Sorry, I meant an optical, self-contained method that does not use external references.
Hm. The thermal accelerometer is new to me.
And it looks like those are in the “insanely cheap” category. This 3 axis one is $1.95 in lots of 1,000 and this 2 axis one is 90 cents.
Not certain I understand your question, but if you are asking whether there is a self-contained device that uses no information from outside of itself and is capable of detecting constant velocity motion-well the question has occurred to others. Einstein looked into this, Galileo as well. What they came up with is no-the universe doesn’t work that way. The consequences of this conclusion led directly to the theory of relativity and many other basic physical principals.
I am sure to be ninja’ed on this, but I am so happy to be able to answer a physics question that I had to post! 
So far the thermal accelerometer comes closest to what I seek. Unfortunately the ones available are too slow for my application. 27 Hz. I need up to 2 KHz.
As always I enjoy the variety of information and speculation that occurs in the answers.
I have looked into mems devices again. They have greatly decreased the amount of external circuitry required. Toughened them. So I have ordered 6 types for evaluation.
Thanks for narrowing the field for me.
I think the answer is “sort of”. The point is that if you send a light beam from one end of an accelerating object to the other (in the direction of the acceleration) the receiver is travelling faster than the transmitter was. So you will see a Doppler shift. So, in principle, you should be able to work out your acceleration. The down side is that the implementation is infeasible. The actual change in wavelength is so infinitesimal that there is no possibility of retrieving the effect from under the noise.
Angular acceleration is feasible, because you can wind many kilometres of optical fibre around a spool, and you get a useful delta in wavelength, plus you can easily compare the transmitted and received light in an interferometer, as they can be at the same spot. There is no way of doing anything similar with a linear acceleration, so you are left with either a device many kilometres in length, or one where the signal is many orders of magnitude too small.
I’m sure a real physicist will be along any time soon to put me right.
You said in another thread that you were making a geophone. Have researched MEMS use in other geophones? It seems like that current manufacturers have no problem using them. (Also, sense other products seem to be readily available, are you sure that you have a wheel that needs reinventing?)
You’ll actually see my company doing the tests in some of those documents.
I am not reinventing this wheel. Looking for more mileage, fewer flats.