Context-shift in Farsi, and the Ban Against Illeism

Amir Anvari, Institut Jean-Nicod


It is a well-documented fact that in many languages indexical items (in particular, the first-person pronoun) in reported speech get their meaning, not from the actual context of utterance, but the reported context of speech (see Deal 2017 for comprehensive discussion and references). In such languages an utterance of the sentence corresponding to “The person who A said _ invited me missed the party” by S can mean “The person who A said _ invited A missed the party” rather than “The person who A said _ invited S missed the party”, the latter being the only relevant reading available in languages such as English. The first goal of this talk is to provide evidence that the Tehrani dialect of modern colloquial Farsi (Persian) is a language of the former type. The second goal is to use this claim to study a certain felicity condition, the Ban Against Illeism (BAI), which has not received much attention in the (formal) literature. According to BAI it is infelicitous for the speaker to refer to himself in the third-person, in particular by using his name. This is one reason why the sentence “Dali is contradictory and paradoxical man” amounts to an odd utterance coming from Salvador Dali (interview with Mike Wallace, 1958). On the basis of evidence from Farsi, I will characterise a systematic class of exceptions to BAI: the speaker can refer to himself with his name if his name occurs in a context-shifted environment. Thus in Farsi “The person who A said _ invited her to S’s party missed the party” (which involves no context-shifting since the third-person pronoun is in the third-person) is degraded, but “The person who A said _ invited me to S’s party missed the party” is acceptable under the context-shifted reading “The person who A said _ invited A to S’s party missed the party”. I will use this observation to give a precise characterisation of BAI on the basis of Maximise Presupposition!-type competition (Percus 2006, a.o.) between proper names and indexical pronouns which predicts the observed exceptions. Furthermore, I will suggest a generalisation of BAI to temporal and locative reference as well as reference to individuals. Finally, time permitting, I will turn BAI on its head and speculate that, to the extent that BAI is active in a given socio-linguistic community, it can be used as a fairly robust (but doubtless limited) heuristic to identify indexical items in that language.