The recovery of pieces of information which are not linguistically expressed is a constant feature of the process of language comprehension. In the processing literature, such missing information is generally referred to as ‘gaps’.Usually, one resolves gaps by finding ‘fillers’ in either the sentence or the context. For instance, in Peter seemed to be upset, Peter is really the subject of being upset but appears as surface subject of seems. Sometimes constituents move, leaving gaps behind. Various Romance languages such as Spanish or Italian have a grammatical particle se/si which, as it is extremely ambiguous, licenses different sorts of gaps. In Spanish, se can encode at least reflexive, impersonal, and passive meanings. In an eyetracking experiment we contrast reflexive structures containing post-verbal subjects, with impersonal structures with no subjects (GAP se vendó apresuradamente el corredor / ‘the runner bandaged himself hurriedly’ vs GAP se vendó apresuradamente al corridor / ‘(someone) bandaged the runner hurriedly’ ). And in a second manipulation we contrast the presence of an extra argument with se-passives (GAP se vendó el tobillo el corredor / ‘the runner bandaged his ankle’ vs GAP se vendó el tobillo al corridor / ‘the runner´s ankle was bandaged’). Our comparisons involve contrasting standard transitive structures with non-standard word order (post-verbal subject and a preverbal subject gap) against inherently complex and less habitual structures such as impersonals (with no subject) or se-passives (with subjects in canonical object position). We evaluate the Minimal Chain Principle (de Vincenzi, 1991), according to which displacement is costly because it entails complex (derivational) ‘chains’ that must be undone before phrasal packaging can commence. We show the Minimal Chain Principle to be essentially correct when contrasting more complex but more frequent structures with less complex but less frequent structures. A noteworthy feature of this research is that the gaps appear before the fillers in the structures we analyze.