This article presents a cross-linguistic study of semi-spontaneous data obtained from an experiment conducted uniformly for six languages. It examines how native speakers communicate the changing spatial layouts of toy animals. The analysis of the data focuses on the universal preference for expressing a given constituent before a new one. In terms of grammatical strategies, speakers universally tend to realise the newly introduced or displaced toy animal in a position where it is aligned with a high-level prosodic domain. A constraint to achieve this effect, called Align-Focus-R, is formulated as an optimality-theoretic alignment principle. Language-dependent syntactic and prosodic restrictions may favour or disfavour this tendency. Some languages may reorder their constituents by scrambling, some may use more costly syntactic and prosodic operations, like dislocations, or the insertion of a prosodic boundary. Some use pitch accents, but some do not possess pitch accents in their phonological inventory. A constituent right aligned with a higher-level prosodic domain may be felt prominent, but prominence is only a secondary effect of alignment.