EDIT: I may have screwed up some of this explanation. I'll probably make changes soon.

There are researchers in my department studying topics related to linguistic frames of reference. From the Wikipedia link:

A frame of reference is a coordinate system used to identify location of an object. In languages different frames of reference can be used. They are:

Intrinsic frame of reference

Intrinsic frame of reference is a binary spatial relation in which the location of an object is defined in relation to a part of another object (its side, back, front, etc.). For instance, saying "The cat is in front of the house" means that the cat is at that part of the house we call its front, the side of the house that faces the street and has an entrance and maybe a porch.

Absolute frame of reference

Absolute frame of reference is also a binary system in which the location of an object is defined in relation to arbitrary fixed bearings (for instance, cardinal directions (North, South, East, West)). For instance, saying "The cat is to the south of the house" the location of the cat is described independently of the position of the speaker or of any part of the house (as in intrinsic frame of reference). Apart from cardinal directions such fixed bearings as seacoast, upriver/downriver, and uphill/downhill/across are used.

Relative frame of reference

Relative frame of reference is a ternary system. The location of an object is expressed in relation to both the viewpoint of the perceiver and position of another object. Thus saying "The cat is to the left of the house" we refer to three points of reference: the cat, the house, and the perceiver himself.

This terminology is widely used, and I believe coined by Stephen C. Levinson.

But I don't like it. "Intrinsic" sounds like "innate". An "absolute" frame of reference really isn't absolute at all; it's relative to geography. And "relative" makes it sound like it stands in contrast to absolute (e.g. relative and absolute morality), when it really isn't used that way.

Also, it seems to me that the three cases fall into two general categories: viewer-centered and object-centered.

A viewer-centered frame of reference is based on the perspective of the perceiver. An object-centered frame of reference would position the 3D axis on something else: a house, a dog, another person, or in the case of what Levinson calls "absolute", the earth.

Think about the frames of reference of a sea captain. How do terms like "aft" and "starboard" fit into Levinson's frames of reference? Are they intrinsic (to the boat)? Or absolute? This case doesn't seem to fit well into Levinson's framework.

Anyway, Levinson's research group also makes claims about the language and cognition of certain groups, for example, saying that speakers of Tzeltal (a Mayan language) and Guugu Yimithirr (an Australian aboriginal one) strongly prefer the absolute frame of reference over the relative (or viewer-centered), and that this ends up altering their cognition. I believe there is a general claim that all language groups use the intrinsic (or object-based) frame of reference.

Right off the bat this makes me skeptical. What they call the "intrinsic" frame of reference basically seems like an extension of the relative frame of reference, and it seems more cognitively demanding. Let's say you had a doll and a ball next to each other on the table in front of you, and the doll was facing you:

Using the absolute frame of reference, with the axis centered on the earth, you'd say "The ball is to the west of the doll."

Using the relative frame of reference, you'd say "The ball is to the left of the doll."

To use the intrinsic frame of reference and say "The ball is to the doll's right" requires you to assume the perspective of the doll.

Subjectively, this seems to me to require using the relative frame of reference, but then performing a mental rotation onto another perspective. I just don't see how any speaker could easily use an intrinsic frame of reference without mastery of the relative frame of reference. How could you be able to easily say "The ball is to the doll's right", but not "The ball is to my left"?

If it is the case that speakers of certain language can easily adopt some external object or person's perspective and describe spatial relationships in those terms, but not their own, I would like to hear an explanation of how this might be possible, and see a proposed model. It smells fishy to me. I'm open to the possibility that this is actually what's going on, but I'd really like to hear an explanation of the underlying mechanisms of the phenomena. I could buy that in some circumstances, where people live in a fixed environment with salient landmarks, that the absolute frame might be more cognitively efficient than using the relative frame. But I just don't see how speakers could naturally use the intrinsic while strongly dispreferring or not even having a relative frame of reference.