An approach to analyzing meaning and expression in spatial language developed by Chris Sinha and Tania Kuteva. This approach starts from the rejection of the assumption that spatial meaning (for instance, path, source, goal and so forth) is ‘packaged’ by selecting single classes of lexical items.

It is often, in fact, assumed that fully specified spatial meaning is expressed universally by closed class grammatical classes such as prepositions. According to the distributed spatial semantics approach, this view is problematic. For instance, the semantic value associated with the English lexical item in, as evidenced by examples such as the following, varies: the coffee in the bowl versus the spoon in the cup. While coffee is enclosed by the cup, the spoon is not wholly enclosed.

The distributed spatial semantics approach holds that to assume that variation in meaning of this kind is the result of distinct senses being conventionally associated with a form such as in would lead to the polysemy fallacy. In contrast, Sinha and Kuteva propose that spatial meaning itself is distributed over several grammatical classes, not only those conventionally considered to have locative meaning such as prepositions.

This leads to the general proposition that spatial meaning is characterized by many-to-many mappings between spatial meaning and the particular forms employed to encode the given spatial meaning. In this approach, the mappings may be many-to-one (conflation), one-to-one (compositional) or one-to-many (distribution). Additionally, Sinha and Kuteva argue that there is a typological distinction in the languages of the world in the way in which spatial meaning is distributed. Some languages, for instance Japanese, frequently repeat the same spatial information in different places in the utterance, sometimes using the same morpheme in different grammatical roles. This they refer to as ‘overt’ distributed spatial semantics.

Other languages, including English, tend to compress the spatial meanings, omitting explicit encoding of some parts of it, as in for example the boy jumped the fence, where the preposition is optional. This is referred to as ‘covert’ distributed spatial semantics.