This paper investigates Japanese-speaking children's (in)sensitivity to information strength when interpreting sentences containing the focus particle datte 'even.' It examines whether or not their sensitivity is affected by the Question Under Discussion (QUD) and the Felicity Judgment (FJ) task based on the Processing Limitation Hypothesis (PLH; cf. Chierchia et al. 2001 cf. Chierchia et al. , 2004.Because datte 'even' is not a scalar term, it does not give rise to scalar implicatures (SIs) by constituting a part of scales. 'Even' evokes conventional implicatures. However, an 'even' sentence in context -in addition to the conventional implicatures -evokes a special case of SI calculated from contextdependent scales, which is pragmatic inferences induced by 'even's conventional implicatures (i.e., the word's semantic/pragmatic import). The relevant scale (generally) concerns the NP element focused by 'even' and a set of alternatives. Because sentences without 'even' do not evoke any SI, I assume that the relevant implicatures are conventional in nature.Three experiments were done to examine whether Japanese-speaking children are really (in)capable of calculating conventional implicatures derived from datte sentences. It was found that (i) Children are insensitive to pragmatic anomalies of single "infelicitous" sentences (Experiments 1 and 2 Zondervan (2007Zondervan ( , 2009, the wh-focus/QUD way of asking questions does not facilitate the children's performance with datte sentences (Experiment 2). (iii) The FJ task improved children's performance (Experiment 3), thus supporting the PLH, which was applied to SI computation cf. Reinhart 1999 cf. Reinhart , 2006, and to the computation of implicatures involved in datte sentences. (iv) Children are unable to compute implicatures based on context-specific pragmatic scales Brought to you by | Simon Fraser University Authenticated Download Date | 6/1/15 3:15 AM 106 M. Ito derived from datte sentences, when a single "underinformative" statement is given. This partly contradicts previous findings on SI based on contextdependent pragmatic scales (Papafragou and Tantalou 2004).
). This inability to detect pragmatic infelicity is consistent with findings about their ability to compute SIs. On the other hand, the adults showed sensitivity to conventional implicatures in all three experiments, which may be taken as evidence that the computation of implicatures derived from datte sentences differs from that of SIs. (ii) Unlike the results reported for SI inThe results indicate that children's insensitivity to the pragmatic infelicity of datte sentences stems from the processing load induced by building and maintaining alternative representations, as reported for SI computation.