Abstract Two Twitter-based corpus studies are reported to account for the increasing preference in The Netherlands for the stigmatized subject use of the object pronoun hun ‘them.’ Twitter data were collected to obtain a sufficient number of hun -tokens, but also to investigate the validity of two hypotheses on the preference for hun , this is, that subject- hun is a contrast profiler which thrives in contexts of evaluation and qualification, and that subject- hun is propelled by its dynamic social meaning, being a tool for nonposh and streetwise self-stylization. Although the latter is not normally a predictor included in regression analyses of constructional choice, it turns out that expressively spruced up tweets with vivid contrast profiling are the prime biotope of subject- hun . Along the way, this paper reviews the potential of Twitter data for the reconciliation of macro-big-data analysis with micro-sociolinguistic focus, but it also reports and attempts to remedy three concerns.