Intersubjective impersonals

In this paper, I will offer a corpus-driven analysis of impersonal constructions, focusing on two strategies which allow for keeping the agent rather vague or, altogether, off stage in Spanish, namely the indefinite pronoun uno (‘one’) (1) and the reflexive impersonal construction with se (2). I will offer a detailed account of factors that influence this alternation of impersonal constructions. On the basis of this analysis, I will argue for describing the difference in terms of intersubjectivity (based on Benveniste 1966, Lyons 1994, Traugott 2003). (1) El problema es querer reconocer que uno es celoso. The problem is want.INF recognize.INF that one is jealous ‘The problem is to want to recognize that one is jealous’. (2) Qué bien se ve por aquí. How well REFL sees by here ‘How well one see from over here’. The difference between the impersonal construction with se and the indefinite pronoun uno ‘one’ has principally been attributed to two factors. On the one hand, pronominal verbs, which already imply a clitic form from the se paradigm, cannot receive an impersonal reading through the use of se. From a purely formal perspective, a subject uno is then the only way for them to acquire an impersonal reading. Since alternation is not then possible for these verbs, they will be excluded from the analysis. On the other hand, from a semantic-pragmatic perspective, uno is traditionally considered a strategy that includes the speaker, as opposed to se (see among others Gómez Torrego 1992). Counter examples may be found, though, such as (3). In addition, constructions with se frequently allow for speaker involvement. These two factors force us to look for a more satisfactory explanation of the alternation between uno and se-constructions. (3) Si uno decide ser adicto, es su problema. if one decides be.INF addict, is his problem ‘If one decides to become an addict, it’s his problem’. (uttered by a non-addict journalist) The hypothesis that the semantic-pragmatic difference between both constructions can be accurately described in terms of intersubjective effect, will be tested through the (manual) analysis of a 250 examples sample of each construction, taken from the CORLEC corpus. Multivariate statistics are employed to investigate the results of this manual analysis. Among others the following factors are taken into account: genre/register, tense, aspect, the presence of a specific person reference form in the discursive context or of scope restriction through adjuncts, and the presence of modal auxiliaries. I will show that a clustering of factors points to the uno construction being a more intersubjective strategy and the se-construction a more neutral one. In addition, corpus-driven analysis permits the identification of the importance of each factor in the choice for one of these constructions