On testing for de se and de re construals across languages

Abstract

Putatively de se expressions such as shifted indexicals and logophoric pronouns have be- come an object of considerable interest among formal semanticists and philosophers of language over the past 20 years. Fieldwork on understudied languages such as Amharic, Zazaki, Nez Perce, Ewe and a number of sign languages has played a crucial role in this research, resulting in a large inventory of pronouns and anaphors that have been described as unambiguously de se. Yet the question of how to go about eliciting data from a native speaker consultant about whether some pronoun or anaphor is obligatorily read de se has seldom been explicitly discussed. In Pearson (2013, 2015) I argued that the claim that logophoric pronouns are unambiguously de se is incorrect, at least for the Niger-Congo language Ewe. In this paper, I discuss the methodological difficulties that arise with elicitation work in this domain and describe how I have approached them in my research. There are two main issues: (i) the unusual or implausible nature of the ‘mistaken identity’ scenarios that must be presented to the consultant and (ii) the tendency of mistaken identity scenarios to systematically induce a bias against the de re reading. The difficulty presented by the first issue has been overestimated, while the second one has to my knowledge not been discussed before. I propose that (ii) is related to a broader obser- vation about truth value judgment tasks: when asked to give a truth value judgement for a sentence S relative to a scenario, linguistically naïve consultants’ judgments are likely to be conditioned by an assumption that every feature of that scenario is relevant. I argue that this tendency leads to a bias in favour of de se over de re construals and propose some strategies for overcoming the difficulty that this presents for diagnosing de re readings.