...Back to the main page of the program
An exchange between UCL and Paris researchers, with presentations of new experimental and theoretical research on pragmatics and the semantics-pragmatics distinction.
This Workshop will take place Saturday September the 20th in the morning in the same location as all the Ealing lectures.
Richard Breheny & Heather Ferguson
Visual world investigations into the costs of implicature processing
The generation of implicatures is a common process during everyday language comprehension. For example, a sentence such, “Mary ate some of the cake” normally implies that Mary ate some, but not all of the cake. However, it is generally agreed among linguists that the ‘but not all’ implication is a defeasible pragmatic inference [3,4]. A theoretical debate about the more common Scalar Implicatures (SIs) has emerged that contrasts two views of processing. The ‘default’ view  is that the ‘but not all’ SI is generated automatically without attention to context. The contextualist view [1,4] is that even common implicatures are not special and require context to be checked before generation. The issue we examined, therefore, was how scalar terms are processed on-line as the current sentence is unfolding.
Here we report a visual world study investigating the time-course of language-mediated eye movements towards quantity-constrained referents in a scene. The current design resolves confounds from previous attempts to investigate on-line SI processing , which include differing visual salience of the target referents and interference from a secondary task. Sentences like those shown below were presented orally together with a picture showing, amongst other things, an all referent (all broken bottles) and a some referent (some broken plates).
(a) Mark has smashed all of the bottles as he was clearing up after dinner.
(b) Mark has smashed some of the plates as he was clearing up after dinner.
Relative proportions of anticipatory eye-movements (i.e. eye-movements that started reliably before the onset of the target word “bottles” or “plates”) indicated an early bias towards the appropriate quantifier-dependent referent in (a). In contrast, the respective bias in (b) was delayed by a conflict between the some and all referents. This suggests that while the semantically specified quantifier all leads to rapid disambiguation of the referents the weak scalar quantifier some leads to initial ambiguity of the referents. We discuss how these findings relate to current theories of SI processing.
 Breheny, R., Katsos, N. & Williams, J. (2006). Are Generalised Scalar Implicatures Generated by Default?, Cognition,100,434-463.
 Huang, Y. & Snedeker, J. (submitted). From Meaning to Inference: Evidence for the distinction between lexical semantics and scalar implicature in online processing and development.
 Levinson,S. (2000). Presumptive meanings. Cambridge, MA: MIT Press.
 Sperber, D. & Wilson, D. (1986). Relevance: Communication and Cognition. Oxford: Blackwell.
Givenness as Grammatical Marking of Existential presupposition
Semantic entailment is sufficient for licensing deacentuation as a means of grammatical marking of givenness in English (Schwarzschild 1999). In Czech, however, the givenness licensing conditions are stronger: only elements existentially presupposed in the relevant contextual domain count as given. In this talk I will address the question of what it exactly means to be existentially presupposed in the context of givenness.
Discourse particles for beginners
This study addresses the question how and when children learn to deal with the presupposed content of an utterance. Discourse particles such as again, only and too are particularly interesting for such an investigation, because in many languages children seem to master their use extremely early (see, for instance, Nederstigt, 2003 for a corpus study on the production of German auch, and the literature findings mentioned therein on other languages). So far the developmental literature on discourse particles has mainly been directed towards children’s understanding of expressions like too (and its German and Dutch counterparts) in contexts where its focus (on the subject or object of the action) is ambiguous (e.g. see Berger et al., 2007; Bergsma, 2006; Hüttner, et al., 2004). These studies are concerned with relatively old children (typically 4-to-10-year-olds) and tend to show that while children’s production of discourse particles is proficient from very early on, their understanding lags behind until school age (see also Paterson et al., 2003).
The aim of my work is to establish whether very younger children are able to draw the presuppositional inferences associated with the expressions too and again (‘auch’ and ‘nochmal’ in German). From the age of two children are proficient in their use of these expressions, but it is not clear that they fully appreciate their semantic and pragmatic import. 24 3-year-olds and 24 2,5-year-olds participated in this study. Children were presented with two toy characters, one of which performed an action (e.g., dance). The child then heard either the phrase, “Anna wants to dance, too,” or “Anna wants to dance again”, where crucially the name “Anna” hadn’t been used before, and was asked to help Anna do it. Thus, in order to assign the correct referent to “Anna”, pick up the right puppet and make her dance, for instance, the child had to make an inference based on the presupposition carried by either too or again. The performance of 3-year-olds was above chance level for both auch and nochmal, while 2,5-year-olds responded randomly. I will present the findings of this study and discuss the implications they have for the early understanding of presuppositions. A possible explanation for the 2,5-year-old results is that the reference assignment task might in itself be too demanding for them. Therefore, I am currently running a new study with a simpler design to establish whether poor performance is linked to their (mis)understanding of the discourse particles or to the design of the experiment. I will also present preliminary data on this new experiment.
...Back to the main page of the program