Refining the distributional hypothesis: A role for time and context in semantic representation


Distributional models of semantics assume that the meaning of a given word is a function of the contexts in which it occurs. In line with this, prior research suggests that a word’s semantic representation can be manipulated – pushed toward a target meaning, for example – by situating that word in distributional contexts frequented by the target. Left open to question is the role that order plays in the distributional construction of meaning. Learning occurs in time, and it can produce asymmetric outcomes depending on the order in which information is presented. Discriminative learning models predict that systematically manipulating a word’s preceding context should more strongly influence its meaning than should varying what follows. We find support for this hypothesis in three experiments in which we manipulated subjects’ contextual experience with novel and marginally familiar words, while varying the locus of manipulation.

Back to Table of Contents