I've just found this fascinating study on language and psychosis that found that multilingual psychotic patients can present with either different or less psychotic symptoms depending on the language they use.
It's a 2001 study from The British Journal of Medical Psychology that collected existing case studies from the medical literature and reports on several new examples.
There have been previous accounts of bilingual or 'polyglot' patients who only hear voices in one of their languages, but this seems to be the first study to assess psychotic symptoms using a standardised measure.
This is from the introduction, which outlines some of the curious effects:
Zulueta’s (1984) review article on the implications of bilingualism in the study and treatment of psychiatric disorders showed that certain psychotic fluent bilinguals, who had learnt their second language during or after puberty, could present with different psychotic phenomena depending on which language they used. Most of these patients tended to present as more disturbed in their primary ‘mother tongue’ and as less disturbed in their second language (Castillo, 1970; Hemphill, 1971).
Some patients were thought disordered in one language and less so or not at all in their other language; some complained of having delusions in one language and not in their other language, and some experienced auditory hallucinations in one language and not in another. Moreover, some patients who were fluent bilinguals lost their linguistic competence in their second language during their psychotic illness (Heinemann & Assion, 1996; Hughes, 1981).
The case of Mr Z illustrates the marked change in phenomenology that can be observed in such patients. He was a 30-year-old patient diagnosed as hypomanic with a history of bipolar illnesses. His mother tongue was English, and he had learnt Spanish after puberty. When he spoke in English, he was markedly thought-disordered and complained of hallucinations. On one occasion, whilst being interviewed by his psychiatrist, he addressed her spontaneously in Spanish, knowing that she was a Spanish speaker.
To his surprise, and hers, he discovered that when he spoke in Spanish, he no longer appeared to be thought-disordered. He commented on this difference by observing, in Spanish, that when he spoke in this language, he felt he was ‘sane’, but when he spoke in English, he went ‘mad’ (Zulueta, 1984). This bilingual dialogue took place within the space of half an hour. It would seem that in this case and in others with similar differences in psychotic phenomena across languages, the second language may, in some cases, exert a protective function in terms of psychotic symptoms.