This paper discusses Ferdinand de Saussure’s participation in a case study of somnambulism with glossolalia from 1896 to 1899. The subject of that study, a medium named Catherine-Élise Müller, claimed to speak in languages she had never learned. Saussure produced a theory of how she assembled her glossolalia in a way that sounded like Sanskrit. That theory of glossolalia assemblage met belated controversy when Tzvetan Todorov criticised Saussure for overlooking a ‘symbol’ at operation in Müller’s glossolalia. I show that Saussure’s theory of Müller’s glossolalia has broader explanatory power than the symbolic aspect favoured by Todorov, and that what Todorov perceives as a complicated ‘symbol’ can be better explained with simple physiology.