Nuclear transitivity in English and Spanish

This paper presents a contrastive overview of nuclear transitivity in English and Spanish from a systemic-functional perspective. The study attempts to achieve two main goals. Firstly, we investigate the usefulness of the transitive/ergative distinction developed by Davidse (1992) for material processes in English, when applied to different process types in both English and Spanish. Secondly, we attempt to provide a fine-grained specification of these linguistic resources which might form the basis for computational treatment in the applied context of Multilingual Generation (MLG), the automatic production of texts in various languages from a single underlying data source.

We first review the specifications provided for English in the computational grammar Nigel (Mann 1983; Matthiessen 1988; Matthiessen and Bateman 1991), and in the extensive reference grammar developed by Matthiessen (1995), showing that, though useful for practical generation purposes, they conflate the notions of agency and causation under the same system. These specifications thus blur the transitive/ergative distinction, which is also fundamental to observe how semantically related verbs may behave differently in English and in Spanish (Lavid and Arús 1998a, 1998b; Arús and Lavid 2001). We propose, instead, a model of nuclear transitivity consisting of three simultaneous systems: a system of agency (concerned with the presence or absence of the feature Agent), a system of process type (concerned with the semantic type of process involved), and a system of causation (concerned with the variable of instigation). This latter system establishes the paradigmatic distinction between transitive and ergative processes which we claim to be fundamental for an accurate contrastive description of English and Spanish. It is expected that the proposed model, which captures the common semantic potential of nuclear transitivity in English and Spanish, will prove useful as the linguistic basis for a more detailed computational specification in the context of MLG.