Exhaustivity, linguistic prediction and integration: it-clefts and 'only' sentences

Abstract

This research investigates sentences with the adverb ‘only’/‘só’ and it-clefts/clivadas totais, in English and in Brazilian Portuguese (BP). There is consistent evidence that these focus markers (Kiss, 1998; Paterson; Liversedge; Underwood, 1999; Liversedge et al, 2002; Zimmerman, 2004; Filik et al, 2009; Carlson, 2015) convey exhaustivity through different processes (Horn, 1981; Rooth, 1992; Krifka, 1999; Drenhaus; Zimmerman; Vasishth, 2011; Carlson, 2014; Lignani; Fonseca, 2023). Exhaustivity is the semantic effect whereby a focused element in a sentence excludes all other alternatives to make that sentence true (Rooth, 1992; Krifka, 1999). Two types of experiments were designed to compare exhaustivity violations for clefts and ‘only’-structures: a) three Likert tests (Likert, 1932) – AJTs (in English, tasks 1 and 2; in BP, task 3); and b) two eye-tracking tasks (one English and one in BP). Our hypotheses were that: 1) exhaustivity violations in ‘only’-sentences/'só’-sentences' are less acceptable than in it-clefts/clivadas totais; 2) clefts are more sensitive to context integration than sentences with the adverb. In the AJTs, there were sets of stimuli in four conditions: exhaustive/non-exhaustive ‘only’-sentences (‘Yesterday, only Anna brewed coffee and pastries too’/‘Yesterday only Anna brewed coffee and Olivia did too’); and exhaustive/non-exhaustive clefts (‘Yesterday it was Anna who brewed coffee and pastries too’/ ‘Yesterday it was Anna who brewed coffee and Olivia did too’). In the eye-tracking experiment, we included context-sentences, displayed on previous screens (‘Anna and Olivia run a busy bakery in London’). Results show that: 1) ‘only’-sentences were less accepted and elicited more regression-based measures than clefts; 2) non-exhaustivity decreased the acceptability of all sentences, and reduced the reading times for adverb-sentences. Our findings corroborate the idea that exhaustivity relates to the truth values in ‘only’-sentences (Rooth, 1992; Kriffa, 1999), whereas to pragmatic in clefts (Drenhaus; Zimmerman; Vasishth, 2010; Onea, 2019). We argue that exhaustivity in clefts engage prediction-by-production (Pickering; Gambi, 2018), in which comprehenders simulate the speaker’s communicative intention (Destruel et al., 2015; Destruel; Tunnicliff, 2020). However, exhaustivity in ‘only’-sentences is part of the core semantic content, unleashing the fast, automatic process of prediction-by-association based on associations between words or structures (Pickering; Gambi, 2018).