Pragmatisch Redeneren met Schalen, Spatiale Termen en Incompatibiliteit

If some inferences require cognitive effort, could that mean that all of them do? The scalar term some has long fascinated academics, as it can be interpreted either in a broad, semantic (or logical) way, as some and possibly all, or in a more narrow, pragmatic way, as some but not all. The pragmatic reading is believed to involve the generation of a scalar implicature—implicit meaning derived from an utterance that goes beyond its explicit linguistic content, in this case specifically in relation to a scalar term. The derivation of such implicatures has been the subject of much debate, and the disagreement between the two dominant frameworks—the literal-first and the default accounts—has not convincingly been settled. The loading and measuring of Working Memory (WM) has become an important experimentation tool for testing the contrasting hypotheses with regard to the cognitive effort of implicature generation, which are made by the two main groups of theories. In Chapter 1 of this dissertation, we present a comprehensive systematic review that we conducted on the role of WM in implicature derivation. Our findings show that scalar implicatures require cognitive effort to derive, and that this process is sensitive to changes in the available WM resources. This seems to support the literal-first hypothesis, at least with regard to the word some. Notably though, the reported effects are relatively weak and capricious, and there seems to be considerable heterogeneity in the behavior of scalar terms other than some—an observation in line with the notion of scalar diversity. In Chapter 2, we expanded these findings to a different, yet related class of expressions. Spatial terms, like left, right, above, and below are extremely prominent in our daily lives, and their interpretation is generally automatized to a very high degree. They allow us to swiftly communicate the position of various objects and entities in space, and reasoning with purely linguistic spatial information often allows us to infer the (approximate) location of specific objects without any visual cues whatsoever. The process of interpreting spatial terms is one that relies heavily on pragmatics. Just like with some, a statement like to the left of can be understood in a broad, semantic sense as anywhere to the left of or in a more narrow pragmatic way, as directly to the left of. A key difference, however, is that such spatial scales are non-bounded—they do not have a natural end point in our language, like all in the case of some—which is a characteristic found to be a driver of scalar diversity. We hypothesized that factors like an additional object in between the two referents, and people's available WM resources would influence their interpretation of spatial expressions. Two sets of experiments were conducted in order to investigate both the horizontal (left vs. right) and the vertical dimension (above vs. below). Participants performed either a simple sentence verification task, or a dual-task where WM was burdened to different degrees. Overall, we observed relatively low pragmatic derivation rates in all four experiments. However, we found that the presence of a distractor object between the two referents increased the likelihood of a pragmatic interpretation of a spatial statement. Notably, an effect of Working Memory was only observed for terms related to the vertical dimension. For Chapter 3 of this dissertation we carried out a comprehensive investigation on the pragmatics of reasoning with incompatibility statements. The notion of incompatibility is fundamental to human cognition and ever present in our language. However, to date, only three empirical papers have been published on this topic, two on adults, and one on children. All three studies found evidence for semantic content effects on human reasoning with incompatibility. In this project, we replicated and extended these findings, by also accounting for pragmatic influences on these inference processes. We proposed an alternative formalization of incompatibility, which makes use of modal logic, instead of the classical, extensional propositional method, and performed a set of experiments in order to evaluate the merits of our alternative approach. In order to maximize its replication value, our empirical investigation included both child and adult samples, and made use of (modified versions of) the experimental items used by the previous studies. Our findings were not fully consistent with the existing literature, and we posit they are best explained through a combination of pragmatic influences and WM involvement, which both play a role in people's selection of a reasoning strategy. Additionally, we found that the response options that are provided in a verbal reasoning task significantly impact participants' performance, and researchers should therefore be cautious in the interpretation of any data resulting from experiments using this paradigm.