Doubled clitics are pronouns

Abstract

Controversy and uncertainty have plagued the question of whether “object markers” (OMs) are object pronouns cliticized to the verb or realizations of object agreement. Using data from Amharic, we address this question from a new perspective. Specifically, we claim that Amharic OMs should be analyzed as clitics because they are unable to double nominals that are quantified, anaphoric, or contain a variable bound by a quantifier. These restrictions can be derived from familiar principles of grammar—the Weak Crossover condition and the Binding theory—if and only if the OM is taken to be a pronoun (D) adjoined to v at LF. We then explain in terms of syntactic structure why these OMs can double even nonreferential DPs in experiencer subject constructions, whereas they cannot double theme arguments in unaccusatives or passives. Finally, we consider whether our analysis provides a diagnostic that distinguishes pronominal clitics from agreement morphemes across languages. OMs in Greek, Bulgarian, and Spanish seem to work broadly like Amharic, confirming that they are pronominal clitics. In contrast, OMs in Burushaski and Sambaa behave like agreement, even though certain other putative diagnostics suggest they might be clitics. We thus confirm that both object clitics and object agreement exist.