Since Sun-Joo Shin's groundbreaking study (2002), Peirce's existential graphs have attracted much attention as a way of writing logic that seems profoundly different from our usual logical calculi. In particular, Shin argued that existential graphs enjoy a distinctive property that marks them out as "diagrammatic": they are "multiply readable," in the sense that there are several different, equally legitimate ways to translate one and the same graph into a standard logical language. Stenning (2000) and Bellucci and Pietarinen (2016) have retorted that similar phenomena of multiple readability can arise for sentential notations as well. Focusing on the simplest kinds of existential graphs, called alpha graphs (AGs), this paper argues that multiple readability does point to important features of AGs, but that both Shin and her critics have misdiagnosed its source. As a preliminary, and because the existing literature often glosses over such issues, we show that despite their non-linearity, AGs are uniquely parsable and allow for inductive definitions. Extending earlier discussions, we then show that that in principle, all propositional calculi are multiply readable, just like AGs: contrary to what has been suggested in the literature, multiple readability is linked neither to nonlinearity nor to AGs' dearth of connectives. However, we argue that in practice, AGs are more amenable to multiple readability than our usual notations, because the patterns that one needs to recognize to multiply translate an AG form what we call complex symbols, whose structural properties make it easy to perceive and process them as units. Nevertheless, we show that such complex symbols, though largely absent from our usual notations, are not inherently diagrammatic and can be found in seemingly sentential languages. Hence, while ultimately vindicating Shin's idea of multiple readability, our analysis traces it to a different source and thus severs its link with diagrammaticity.