What's behind some (but not all) implicaturesLuca Surian
, Francesca Panzeri
, Francesca Foppolo
, Greta Mazzaggio
, 2018, published scientific conference contribution abstract
Abstract: Several studies investigated children’s derivation of pragmatic inferences by testing different items in different languages, populations and tasks (Skordos & Papafragou, 2016). In general, pre-schoolers have difficulties in the computation of the scalar implicature (SI) related to some, while a better performance has been documented in the case of non scalar or ad hoc implicatures (AIs), even in younger kids (Horowitz, Schneider and Frank, 2017; Katsos & Bishop, 2011; Stiller, Goodman & Frank, 2015). Children’s difficulty has been accounted for by different hypotheses: children are more tolerant of pragmatic violations than adults ('tolerance account', Katsos & Bishop, 2011); children do not (always) recognize what is conversationally relevant ('relevance account', Skordos & Papafragou, 2016); children have difficulties in lexicalizing the scale and/or retrieving the lexical alternatives ('lexicalist account', Barner et al., 2011; Foppolo et al., 2012; Tieu et al., 2015). Yet, the source and nature of children’s difficulty is still unknown, as well as the interplay of different factors and their impact on different inference types.Different theoretical accounts make different predictions for SIs and AIs. In principle, no difference is expected between implicature types within a pragmatic approach (like the tolerance or the relevance accounts), provided that children’s non adult-like behavior relies in a principle of pragmatic tolerance or in a failure in accessing or recognizing relevant alternatives, that would equally affect all kinds of implicature. Under lexicalist approaches, on the other hand, a difference between AIs and SIs is expected: while in the case of AIs the alternatives depend solely on context, in SIs the set of alternatives is a feature of the language that relies on the lexical representation of scales. The crucial difference is in the access to the alternatives, which depends on a linguistic representation and a lexical retrieval mechanism in the case of scalar quantifiers, while it is purely contextually determined in the case of ad hoc scales. In our study, we compared AIs and SIs by means of a Picture Selection Task modelled after Surian & Job (1987) and Stiller, Goodman & Frank (2015). Participants had to find the target (among 4 pictures) by following instructions.
The tasks included 4 implicatures of each type, interspersed with control sentences. To understand the developmental factors beyond children’s performance, children were also administered tests of cognitive and linguistic development (Raven’s Progressive Matrices; BVL for lexicon and morphosyntax; the first four tasks of Wellman & Liu 2004 to test for Theory of Mind, ToM: Diverse Desires-Diverse Beliefs-Knowledge Access-Content False Belief). We tested 141 children aged 3 to 9 (75 in kindergarten: 3;10-6;0, M = 61 months; 66 in primary school: 6;10-9;2, M=90 months). Our findings add an additional piece to the understanding of children’s failure and success with scalar inferencing. In particular, we show that younger children succeed with AIs but have difficulties with SIs. We also found a significant role of linguistic (i.e., morpho-syntactic) abilities for both type of implicatures and a contribution of ToM: KA predicts implicatures derivation, and DB seems to play a role in SIs.The overall picture is rather puzzling: focusing on pre-schoolers, their ability to derive both types of implicatures seems to depend on common mechanisms, such as morpho-syntactic skills and the ability to recognize that another person can know something only if she has access to relevant information. Nevertheless, by controlling for task effects, we confirmed that SIs are harder than AIs. This finding is not easily explained within a pragmatic approach: if children were, in general, more logical or more tolerant than adults, they would be equally so with any kind of implicature; at the same time, if they were not sensitive to informativeness or unable to recognize relevance, they would fail with all pragmatic inferences. A lexical account instead predicts that, beyond general mechanisms common to SIs and AIs, the derivation of SIs requires one further step, that takes time to be acquired or automatized, namely: the lexicalization of the relevant scales. This might be the real source of the observed difference between these types of inferences, although further research is needed to fully capture its impact on children’s performance.
Found in: osebi
Keywords: scalar implicatures, adhoc implicatures, typically developing children, experimental pragmatics
Published: 21.09.2021; Views: 183; Downloads: 6
Fulltext (0,00 KB)
This document has many files! More...