This paper investigates cose (pl. of cosa ‘thing’) as a general extender (GE) and marker of non-exhaustivity from Latin to contemporary Italian. The study employs three corpora: CODIT, LIP/VoLIP and KIParla. We show that the frequency of cose-GEs dropped in 16th c., when they started being perceived as colloquial. In Old Italian, cose-GEs already expressed non-exhaustivity in list constructions and, until late 17th c., were frequently specified by a nominal modifier, which however was uninformative to identify the category. Contemporary spoken Italian results confirm the role of spoken language in developing structures encoding non-exhaustivity, also in a dialogical sense. Moreover, recent data show an increase in frequencies of cose-GEs. Finally, we found more variability compared to the structure usually identified for GEs.