The present paper reports the results of an acceptability judgement test designed to test two hypotheses: first, that the Russian 3pl impersonal construction is affected by agent prominence, i.e. verbs implying a more prominent agent are more acceptable than verbs implying an agent with fewer agentivity features; second, that movement as an agentivity feature should be conceptualized as a gradual category, thus making a fast-moving referent a more prominent agent than one moving slowly. The former hypothesis could be verified, the latter falsified.

The article is written in English.