Background: Over recent years a number of studies have suggested that phonological and deep dyslexia are not separate acquired dyslexias but actually reflect different points along a single continuum. Behaviourally this continuum was originally defined in terms of the graded presence/absence of the "symptoms" of deep dyslexia. Aims: Given that orthography → phonology translation is core to phonological-deep dyslexia, it becomes critical to measure the degree and nature of any such direct phonological activation in these patients. Nonword reading accuracy has been the most commonly used measure of the status of direct orthographic → phonological activation but this measure can floor out in many phonological-deep dyslexic patients. In such circumstances, however, researchers have found evidence for some residual O→P function using reading tasks that do not require overt production-and these tasks might provide an important set of additional measures to test for a gradation of performance at the severer end of the phonological-deep dyslexic continuum. Methods & Procedures: In the present study three tasks of this type (spoken-to-written nonword-matching, a novel word-matching task, and pseudohomophone reading) were developed and tested in a case-series of phonological-deep dyslexia (and made available in the appendices to this paper). Outcomes & Results: As suggested from past studies, even patients with severely impaired overt nonword reading exhibited above-chance performance on the matching task and a pseudohomophone effect. The accuracy on these tasks varied in a graded manner across the case-series. Conclusions: Two factors are key to this dyslexia continuum: the severity of phonological impairment and also the degree of interaction between semantic and impaired phonological representations, indicating that semantic representations become more central to reading in the face of phonological impairment. © 2011 Psychology Press.