With effort, most literate persons can conjure more or less vague visual mental images of the written form of words they are hearing, an ability afforded by the links between sounds, meaning, and letters. However, as first reported by Francis Galton, persons with ticker-tape synesthesia (TTS) automatically perceive in their mind's eye accurate and vivid images of the written form of all utterances which they are hearing. We propose that TTS results from an atypical setup of the brain reading system, with an increased top-down influence of phonology on orthography. As a first descriptive step towards a deeper understanding of TTS, we identified 26 persons with TTS. Participants had to answer to a questionnaire aiming to describe the phenomenology of TTS along multiple dimensions, including visual and temporal features, triggering stimuli, voluntary control, interference with language processing, etc. We also assessed the synesthetic percepts elicited experimentally by auditory stimuli such as non-speech sounds, pseudowords, and words with various types of correspondence between sounds and letters. We discuss the potential cerebral substrates of those features, argue that TTS may provide a unique window in the mechanisms of written language processing and acquisition, and propose an agenda for future research.