Deconstructing SE-Constructions: Number Agreement and Postsyntactic Variation

Abstract

Most analyses of non-paradigmatic SE derive their agreement patterns structurally, forcing a passive/impersonal distinction against all evidence. Instead, we uniformly analyze them as regular sentences where the T-agreeing subject is se itself, an argumental clitic pronoun, with [person] but no number ϕ-features, and show that the overt argument, which has object properties, does not genuinely agree in syntax. We reveal a new asymmetry between postverbal and preverbal/null arguments, which conceals two postsyntactic processes with very distinctive properties: morphological clitic Mutation into number agreement, and T’s Number Harmony with a close DP, not ruled by syntax or morphology.

Person matters in impersonality

“…In any case, this is yet one more restriction that differentiates Basque impersonals from passives in other languages, such as English and Spanish (Ormazabal & Romero 2019, 2020). In Spanish analytic passives, the event can be understood to be initiated by a nonvolitional causer (e.g., a fire), as can be seen in (22a).…”

“…In fact, Spanish analytic passives contrast with reflexive‐marked passives in this respect. The implicit subject of reflexive‐marked passives, such as (22b), must be an animate agent, like in Basque impersonals (Mendikoetxea 1999, Sánchez 2002, Ormazabal & Romero 2019, 2020).…”

“…A similar restriction has been observed for Spanish reflexive‐marked passives. Ormazabal & Romero 2020 argues that se in reflexive‐marked passives cannot license reflexive pronouns like su ‘his/her’ because su is specified for person features whereas se is not. Similarly, the Basque reflexive anaphor bere burua = his/her head ‘himself/herself’, which varies depending on the person features of the antecedent ( nire burua = my head ‘myself’ for first‐person antecedents, zure burua = your head ‘yourself’ for second‐person antecedents), cannot be bound by the implicit subject of impersonals because it requires a binder with the same person specification, whereas the implicit subject has an unspecified person feature: it can be interpreted as first person, second person, indefinite, or generic, like the person pronoun se in Spanish (Mendikoetxea 1999).…”

“…Additionally, being a person pronoun but unspecified for person, the implicit external argument must be interpreted as an animate agent but cannot license adpositional phrases with agentive interpretation that have a specific person feature (first, second, or third person). In all these respects, Basque impersonals are similar to reflexive‐marked impersonals/passives in Spanish (Mendikoetxea 1999, Sánchez 2002, Ormazabal & Romero 2019, 2020).…”

“…On the other hand, the person restriction on the internal argument—also attested in Spanish reflexive‐marked passives/impersonals (Mendikoetxea 2008, MacDonald 2017, Ormazabal & Romero 2019, 2020)—is a consequence of the defectiveness of the version of the Voice projection that is involved. Assuming the typology of Voice proposed in Schäfer 2008b and later developed in Alexiadou et al 2015, we propose that the defective Voice in impersonals is semantically transitive but lacks a D feature.…”