All together now: Disentangling semantics and pragmatics with together in child and adult language

ABSTRACT

The way in which an event is packaged linguistically can be informative about the number of participants in the event and the nature of their participation. At times, however, a sentence is ambiguous, and pragmatic information weighs in to favor one interpretation over another. Whereas adults may readily know how to pick up on such cues to meaning, children—who are generally naïve to such pragmatic nuances—may diverge and access a broader range of interpretations or one disfavored by adults. A number of cases come to us from a now well-established body of research on scalar implicatures and scopal ambiguity. Here, we complement this previous work with a previously uninvestigated example of the semantic-pragmatic divide in language development arising from the interpretation of sentences with pluralities and together. Sentences such as Two boys lifted a block (together) allow for either a Collective or a Distributive interpretation (one pushing event vs. two spatiotemporally coordinated events). We show experimentally that children allow both interpretations in sentences with together, whereas adults rule out the Distributive interpretation without further contextual motivation. However, children appear to be guided by their semantics in the readings they access, since they do not allow readings that are semantically barred. We argue that they are unaware of the pragmatic information adults have at their fingertips, such as the conversational implicatures arising from the presence of a modifier, the probability of its occurrence being used to signal a particular interpretation among a set of alternatives, and knowledge of the possible lexical alternatives.

Acknowledgments

This research would not have been possible without the participation of the children and staff at Pennington Presbyterian Nursery School, Pine Grove Cooperative Nursery School, and the Douglass Psychology Child Study Center. We relied upon a host of multitalented undergraduate research assistants to assist in the creation of the stimuli and running participants, including in particular Hannah Baker and Ariana Kalkstein. For fruitful discussions on this topic we thank audiences at BUCLD 2011 and Northwestern University. Finally, we are indebted to two anonymous reviewers for their invaluable insight and careful, constructive feedback.

Funding

We gratefully acknowledge funding provided by NIH grant HD-057699 and a postdoctoral fellowship from the Rutgers Center for Cognitive Science to K.S., NSF grant BCS 0545067 to J.M., and funding from the Aresty Research Center at Rutgers University–New Brunswick.

Notes

¹ We restrict our attention to the post-VP use of this modifier for the following reason. Referencing Lasersohn (1990) and earlier work by Bennett (1974), Schwarzschild (1994) argues that the post-NP and post-VP uses of together (e.g., John and Mary together lifted a block vs. John and Mary lifted a block together) should be treated distinctly, pointing out that differences in syntactic position are correlated with differences in meaning (see §3.1 and §3.3, and pp. 245–247). For example, post-VP together attaches to the right of the VP and modifies the event, giving rise to multiple interpretations, while post-NP together attaches to the left of the VP and requires the predicate in question to apply to the plurality (e.g., John and Mary), giving rise to an obligatorily collective reading.

² We employed a puppet rather than a live female experimenter, given children’s success in other experimental paradigms, such as the TVJT (Crain & Thornton 1998), in which they interact with a puppet.

³ An anonymous reviewer observes that an additional possible influence of the sentences accompanying the Collective contexts is that they might make children aware that the indefinite (e.g., a book) could have narrow scope relative to the subject and/or that each occurrence of the indefinite in the utterance makes reference to a new entity in the discourse. If this is so, then these sentences might also further facilitate children’s accessing the Distributive reading, which they were already eager to accept, and remind them of the temporal overlap that should hold among subevents in the Distributive context.

⁴ All conditions presented in Figure 3, along with those represented by “previous results,” were run in succession as part of a unified research project, with the same representative population of children.

⁵ In Condition 1, five of the children accepted 75% or 100% of the time, while five of the children accepted at 50% or 0% of the time. This is in contrast to Condition 2, in which only two children accepted 75% of the time, one accepted 50% of the time, and the rest accepted 0% of the time.

⁶ This astute linguist-in-training shares with Schwarzschild (1994) the observation that post-VP together contrasts with separately.

⁷ Abbreviated as “Sm Pl, Df Tm” in the legends of Figures 6 and 7.

⁸ One participant described the “book reading” event in the distributive context as The girls pretended to read. This item was not counted because the participant did not use the verb read as the main verb in the sentence. Thus, the total number for this context is 159, rather than 160.

Additional information

Funding

We gratefully acknowledge funding provided by NIH grant HD-057699 and a postdoctoral fellowship from the Rutgers Center for Cognitive Science to K.S., NSF grant BCS 0545067 to J.M., and funding from the Aresty Research Center at Rutgers University–New Brunswick.