Acquisition of overt and covert and: support for the semantic subset principle

ABSTRACT

The goal of this paper is to tease apart two approaches to the source of children’s consistent scope assignment in negative sentences containing logical connectives; the Semantic Subset Principle and the Semantic Subset Maxim. Previous developmental work has observed that four- to six-year-old children across languages have difficulty with disjunctive interpretations in these sentences and assign conjunctive interpretations. The results of our experiment however show that Japanese children can access the disjunctive interpretations when conjunctions are elided. This finding supports the idea that children are guided by the Semantic Subset Principle when determining the default value of any parameter associated with a logical connective.

Acknowledgments

We would like to express our gratitude to William Snyder, Diane Lillo-Martin, Jeffley Lidz, Kamil Ud Deen, and Tetuya Sano for helpful feedback and discussion. We also thank the audience at GALA 14, LAWNE 2019 at MIT, and the members of Acqui Lab at the University of Connecticut for their insightful comments and questions. We are also grateful to Wataru Sugiura for helping us conduct our experiment. We would also like to acknowledge the children and the teachers at Iwasaki Gakuen Shin-Yokohama Nursery School, Muraoka Nursery School, Fuchuu Aoi Kindergarten, and Matsugaoka Nursery School.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ It is also reported that children often interpret disjunctions in affirmative sentences conjunctively (conjunctive inference; e.g., Paris 1973, Singh et al. 2016). We leave it open whether the conjunctive inference attested in these studies is the same phenomenon as the one we discuss here, and we focus on negative sentences containing logical connectives. See Pagliarini, Crain & Guasti (2018) and Shimada & Goro (2021) for relevant discussion.

² It is reported that contextual manipulations such as controlling the question under discussion do not override children’s dispreferences for disjunctive interpretations (Notley, Zhou & Crain 2016).

³ Several studies adopt the LF-copy approach to argument ellipsis (e.g., Oku 1998; Saito 2007; Sakamoto 2019) rather than Deletion, but the choice between the deletion approach and the LF-copy approach does not matter here. The LF-copy approach states that the LF form of the antecedent is copied onto the ellipsis site. Under the LF-copy approach, the cancelation of the PPI properties can be captured by postulating that the PPI feature of the antecedent gets checked and deleted before it is copied (see Saito 2007 for a similar idea for case features).

⁴ It should be noted here that this prediction cannot be tested with Japanese +PPI disjunction although it is known that the PPI property of Japanese disjunction -ka is also cancelled under argument ellipsis, as shown in (i) (Funakoshi 2013). The elided disjunction in (i) takes scope under negation resulting in a conjunctive interpretation (i.e., Mary did not eat the pepper AND Mary did not eat the carrot).

Table

As we have seen in Section 1, previous studies revealed that 3- to 6-year-old children incorrectly allow an overt disjunction to take scope under clause-mate negation in languages like Japanese. It is thus expected that children consistently access the conjunctive interpretation (i.e., neg > or) in negative sentences with a disjunction whether or not it is elided (see An 2014; Gao et al. 2019, 2021). The contrast between overt and covert disjunction thus cannot be used to tease apart the SSP and the SSM, and hence we focus on ellipsis of conjunction rather than disjunction in this study.

⁵ One may wonder whether (7b) with the disjunctive interpretation is pragmatically felicitous in the situation where the outcome of the eating event can be seen. It is known that some sentences with disjunctive interpretations are used to express a speaker’s uncertainty about what actually happened. Thus, for example, (i) with the disjunctive interpretation (i.e., neg > and) is infelicitous with regular intonation when the speaker knows what exactly the elephant ate and did not eat.

• (i) The elephant didn’t eat both the pepper and the carrot.

• In order to make the use of (i) with the disjunctive interpretation felicitous, the speaker must be uncertain about what the elephant did not eat. Other sentences with disjunctive interpretations however do not require such a pragmatic context. (ii) is felicitous even when the speaker knows what exactly the elephant ate and did not eat.

• (ii) It is not the case that the elephant ate both the pepper and the carrot. (neg > and)

• Similarly, as one of the reviewers pointed out, the VP-ellipsis counterpart of (i) does not call for the pragmatic requirement. The answer in (iii) is pragmatically felicitous even when the answerer is certain about what the elephant ate and did not eat.

• (iii) Did the elephant eat both the pepper and the carrot? – No, he didn’t eat both the pepper and the carrot.

• Like (iii), the disjunctive interpretation in (7b) does not require such a pragmatic context. A full discussion of the contrast between disjunctive interpretations with and without the pragmatic requirement is beyond the scope of this paper. What is important here is that (7b) with the disjunctive interpretation is not infelicitous in the situation where the puppet can see the outcome of the eating event.

⁶ This judgment can be confirmed with an overt counterpart of each interpretation of (8).

⁷ One might consider that their performance provides a piece of supporting evidence to the SSM since they seemed to access the subset interpretation (i.e., and > neg) in (7b). One of the reviewers also points out that (7b) with the indefinite pro analysis (Hoji 1998) also gives rise to a subset interpretation (neg > indef pro), which is false in the given situation. We are however not sure whether they took (7b) as a scope-ambiguous sentence or not since they may have interpreted (7b) intransitively or with a definite null pronoun. Indeed, as reported in Matsuo (2007), children who do not apply ellipsis tend to interpret a null object sentence intransitively (i.e., focusing on whether or not the event of a verb actually happened). In either way, the truth-value of (7b) is false without yielding any scope-interactions. It is thus not sure that the 8 children’s performances on (7b) actually support the SSM approach.

⁸ One of the reviewers pointed out the possibility that the contrast between overt conjunction and convert conjunction in the current study is due to pragmatic felicity. As noted in fn 5, some disjunctive interpretations require certain pragmatic requirement but other disjunctive interpretations such as the one in (7b) do not. Recall that we did not employ the Uncertainty Mode of the TVJT, which could make some disjunctive interpretations pragmatically felicitous, in this experiment. It is thus possible to consider that the children in our experiment rejected to access the disjunctive interpretation (i.e., neg > and) in (7a) because they thought that it is infelicitous. We agree that this is a possible account for children’s performance of (7a), but it does not necessarily undermine the main point of our finding, that is, children were able to access the disjunctive interpretation in (7b), which was pragmatically felicitous without the Uncertainty Mode, as noted in fn 5. It is also unlikely that the children in our experiment did not assign the disjunctive interpretation to (7a) because of its infelicitous use if we take the results of previous studies into consideration. For instance, Goro & Akiba (2004) and Shimada & Goro (2021) adopted the Uncertainty Mode of the TVJT in their experiment, and both studies reported that children clearly rejected to access disjunctive interpretations in negative sentences with an overt conjunction like (7a). We therefore consider that the result of (7a) reflects children’s knowledge of Japanese +PPI conjunction rather than their disfavor of pragmatically infelicitous disjunctive interpretations.