There is no linguistic theory which states this, because the underlying ideas are too vague to submit to theorization. However, you might be able to fill in the gaps. Length requires a standard of measurement, and there are multiple possible standards – number of letters, sounds, phonemes, words, morphemes, time in milliseconds to pronounce (given some limitations on the speakers task).
There is no measurement for simplicity. For example, there is no reasonable way to compare the simplicity of "the" and "cat". Usually when people speak of language being "simple", they are speaking from the perspective of a perception task, but sometimes they mean "hard to pronounce" (this is a rationale that is popularly given behind flapping of /t/ in "winter", that it's "simpler" to pronounce the word with a flap rather than with [t]). Structural complexity can be measured in many ways, such as number of transformations required to derive a string. One could also define "simplicity" based on frequency of occurrence (in some corpus). That kind of metric could lead you deem words like "deem" as complex, and to count the "Had I but known, I would have...." construction to be complex.
I don't know what "median distance between the tokens in which the readers have to relate in order to comprehend the text" means. For example, it is unclear how the notion of "token" is relevant (unless you are dealing with spoken language), and what scale you are using to compute "distance" (distance between what and what?).
Since you seem to be ultimately interested in sentence comprehension, you would probably be most interested in psycholinguistics.