In modern linguistics where natural spoken language has been the main, if not the sole, target of research inquiry, one area in which orthography becomes as yet at issue is phonology. This should be the case because (a) writing systems encode phonological structure, syllables in particular, as shown in Gnanadesikan (2011), accordingly (b) the orthographical knowledge of a language may merge into the knowledge of that spoken language, and finally (c) once the knowledge at the two levels merge in the language user’s brain, it is hard to separate the two levels. Strictly speaking, then, it is not possible to retrieve phonological evidence for purely spoken language from members of orthography-rich communities. This paper nonetheless demonstrates that there are atypical occasions when orthography-free phonological evidence can be observed. Specifically, it reports on a remarkable performance in a Japanese word-reversing ludling (Latin ludus ‘game’ + lingua ‘language’), which was demonstrated by an individual born with a genetic disorder called Williams syndrome. The paper contends that written evidence, of syllables in particular, may constitute evidence for their linguistic reality but only at the time of the orthography invention. Purely phonological evidence that is free of orthographic and prescriptive influences is vital and ideal in that it encodes and reveals far more dynamic phonological structures of the language.