Effects of Processing Limits on Computing Scalar Implicatures: Evidence from Child English and Child Mandarin

ABSTRACT

Relatively late mastery of scalar implicatures has been suggested to correlate with children’s immature processing capacities, such as their limited working memory. Yet, many studies that tested for a link between children’s working memory and their computation of scalar implicatures have failed to find any correlation. One possible reason is that the children in these studies showed little individual variation in computing scalar implicatures. The present study therefore tested for an effect of working memory in younger children, who still clearly displayed difficulties with scalar implicatures and showed greater individual variation. Subjects were 4- to 7-year-old children acquiring either English or Mandarin Chinese. A covered-box task was used to investigate the computation of scalar implicatures associated with ‘some’. A digit span task was used to measure their working memory. Neither the Mandarin-speaking children nor the English-speaking children computed scalar implicatures at an adult-like level. Moreover, a significant correlation was observed between children’s computation of scalar implicatures and their digit spans. These results have provided new support for a processing approach to the observed late mastery of scalar implicatures.

Acknowledgments

I would like to thank Prof. William Snyder, Prof. Diane Lillo-Martin, Prof. Magdalena Kaufmann, Prof. Željko Bošković, and the members of the University of Connecticut Language Acquisition Group (“Acqui-Lab”) for helpful discussion at all stages of this project. I am grateful to UConn K.I.D.S. and Child Development Labs at UConn for their help with recruitment; to parents and children for participating in this project. I also thank the audiences at GALA 14 for their many helpful comments. Last but not least, I would like to express my gratitude for the insightful comments and feedback provided by two anonymous reviewers and Editor Toby Mintz.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

The data that support the findings of this study are available at https://figshare.com/s/44f8f9a71d4b3fec509f

Notes

¹ The slides used for the test of scalar implicatures can be found at https://figshare.com/s/d5eaa41608cbfee9c6ac. There are four different lists of orders.

² The nine excluded children (4;02, 4;03, 4;07, 5;02, 5;02, 5;05, 5;11, 6;03, 7;04) all failed the control condition. As an anonymous reviewer pointed out, most of the children who were excluded were younger (seven out of the nine children were younger than age six). In the control condition, they chose the hidden picture but not the visible “some but not all” picture. When asked why the “some but not all” scenario was wrong, they explained that Peppa Pig failed to take all the referents or there were still some left. It seems that they interpreted ‘some’ as ‘all’. It was unclear why these children displayed this pattern. One possibility is that these children had not yet fully developed the basic semantics of yixie ‘some’. This issue will be left for future research.

³ As an anonymous reviewer pointed out, given that every participant saw six items of the critical LU condition, within-subject variability should also be considered. Therefore, the data were submitted to mixed effect logistic regression.

⁴ Please find the variances for random effects and the correlation of fixed effects for all the mixed effect regression models in Appendix 2.

⁵ The four excluded children (4;02, 4;08, 5;03, 5;05) all failed the control condition. Like those excluded Mandarin-speaking children, they chose the hidden picture but not the visible “some but not all” picture. In their justifications, they would say something like “Oh, there were still three left”. These four children seemed to interpret “some” as “all”.

⁶ As suggested by an anonymous reviewer, since the sample size of English-speaking children (n = 36) was smaller than that of Mandarin-speaking children (n = 45), it would be useful to run analyses that combine the English and Mandarin data sets to maximize power and add “language” as an additional predictor.

⁷ Note that this does not necessarily contradict the processing account argued for here. As discussed above, children’s computation of scalar implicatures can be influenced by many different factors, such as the knowledge of core semantics, pragmatic reasoning abilities, and general processing capacities. This is not an all-or-nothing issue.