Ziegeler discusses the problem that statements of past ability or potentiality sometimes metonymically evoke the actuality but also sometimes the non-occurrence (counterfactuality) of the event expressed in the infinitival complement clause. She challenges Levinson's (1995, 2000) view that an utterance like 'John could solve the problem' implicates 'John solved the problem' on the basis of the second Gricean quantity maxim (Q2), and that 'John had the ability to solve the problem' conveys the complementary implicature that John did not solve the problem. The latter is supposed to be an M-implicature, an inference that, according to Levinson, applies to the more marked (periphrastic) member of a manner set . M-implicatures are not metonymic inferences in the prototypical sense, since they are not content-to-content relations but associate a comparatively marked form with the negation of the content that is assigned to the unmarked member of the set. Ziegeler provides empirical evidence against Levinson's analysis, which seems to be based on made-up examples. She questions the tacit assumption that 'could' and 'had the ability' are synonymous concepts that contrast in "prolixity" in the same way as pairs like 'drink/beverage' or 'house/residence'. Among other things, she shows that 'could' is hardly ever used in present-day English in connection with single past events and that the more marked "alternative" of 'could', the periphrastic 'was/were able to', does not produce an implicature of non-actuality (via M-implicature) as predicted by Levinson's model - on the contrary, the latter has a strong suggestive force of actuality. Ziegeler arrives at the conclusion that the opposite directionality of the metonymies POTENTIALITY FOR ACTUALITY vs. POTENTIALITY FOR NON-ACTUALITY is due to principles that "appear to be founded in pragmatics and the notion of scalar relationship between items." (Klaus-Uwe Panther and Linda Thornburg)