Abstract Presuppositions are contents of utterances that must be mutually known by the interlocutors, i.e. the speaker and the addressees, in order for an utterance of a sentence with the presupposition trigger to be judged to be acceptable. However, a substantial subset of contents traditionally taken to be presuppositions does not exhibit this property and can instead be used to inform the addressee about the supposedly presuppositional content. Classical analyses of these so‐called ‘informative presuppositions’ nevertheless assume that they are associated with a common ground requirement but one that can be accommodated. According to an alternative characterization, ‘informative presuppositions’ are not associated with a common ground requirement but are informative like Potts' conventional implicatures; such a characterization does not need to make reference to accommodation. The debate about the characterization of ‘informative presuppositions’ has been based almost exclusively on judgments about English provided by native speaker researchers. This paper explores how results from research with theoretically untrained speakers (including one‐on‐one elicitation and experiments) as well as corpus studies can bear on this debate. I argue that such results, though tenuous, support the alternative characterization at least for some ‘informative presuppositions’ and emphasize the merit of bringing data collected through a wide range of methodologies to bear on theoretical research on meaning.