On the source of children’s conjunctive interpretation of disjunction: Scope, strengthening, or both?

ABSTRACT

In a body of empirical research, it has been observed that young children from across different linguistic communities adhered to a particular type of nonadult interpretation of disjunction: They appear to interpret disjunction conjunctively. Through three experiments with Japanese-speaking preschoolers, we investigate the source of this nonadult behavior. Specifically, we ask whether children’s conjunctive interpretation of disjunction in negative sentences can be reduced to strengthening via implicature. To test this possibility, we presented Japanese children with test sentences in which the crucial disjunctive NP was located in different syntactic positions: accusative-marked object (Experiment 1), nominative-marked subject (Experiment 2), and nominative-marked object (Experiment 3). The results showed that children systematically altered their interpretations of disjunction according to its syntactic position in the test sentence. Importantly, they consistently accepted adultlike disjunctive interpretations of the test sentences in Experiments 2 and 3, but they showed adherence to the conjunctive interpretation in Experiment 1. These behaviors cannot be explained by the strengthening account, suggesting that children’s conjunctive interpretation of disjunction in negative sentences is due to their nonadult scope assignment.

Acknowledgments

We would like to thank Tetsuya Sano and Wataru Sugiura for helping us conduct our experiments. Our sincere gratitude also goes to all the members of Tokyo Psycholinguistic Lab for their valuable comments. We are also grateful to the children and the teachers at Fuchuu Aoi Kindergarten, Matsugaoka Nursery School, Muraoka Nursery School, and Iwasaki Gakuen Shin-Yokohama Nursery School.

Disclosure statement

The named authors have no conflict of interest directly relevant to the content of this article.

Notes

¹ Here and in what follows, following the terminology widely accepted in the literature (e.g., Singh et al. 2016; Tieu et al. 2017), we use the term conjunctive interpretation to describe children’s behavior of understanding nonnegative disjunctive sentences as if they were conjunctive. At the same time, we also use the same term to describe the conjunctive truth condition that children appear to assign to negated disjunctive sentences. Note, however, that under the Strengthening Account (which we will discuss in the reminder of the article), the sources of conjunctive interpretations of disjunction are different for negative sentences and nonnegative sentences: For the former, conjunctive interpretation is an entailment that conforms to one of de Morgan’s law of propositional logic (i.e., $\mathrm{\neg }(p\vee q)=\mathrm{\neg }p\wedge \mathrm{\neg }q$); and for the later, it is an implicature that is computed through recursive exhaustification. We will not incorporate this distinction into our terminology and keep using the term conjunctive interpretation as a theory-neutral description for children’s behavior.

² The result also entails that Japanese children’s nonadult interpretation of ka cannot be attributed to incremental processing and difficulty in revising their initial interpretive commitment (Lidz & Musolino 2002; Musolino & Lidz 2002). If children’s interpretation of ka reflects their processing preference, then … mo … mo should show the same pattern with ka, given that the relative order between ka/ … mo … mo and negation is held constant.

³ Therefore, under Goro’s account, the SSP is a highly specific learning algorithm whose application is restricted to the determination of the default value for the PPI parameter (see Goro 2015 for further discussion). This should be distinguished from more “global” versions of the SSP, which restrict children’s initial interpretive preferences in general (e.g., Crain, Weijina & Conway 1994; Notley et al. 2012; Moscati & Crain 2014, a.o.). In this article, we will follow Goro’s implementation of the SSP.

⁴ The exact acceptance rates vary from 3% (Crain et al. 2013, Mandarin) to 61% (Pagliarini, Crain & Guasti 2018, Italian, Experiment 1). Pagliarini, Crain & Guasti (2018) argue that this is because Italian children acquire the adult [+PPI] setting for disjunction earlier than children acquiring other languages, due to the fact that Italian is a negative concord language that uses a specific linguistic form to express “neither,” the conjunction of two negated propositions (not P AND not Q). On the other hand, Guasti (2018) reported that children acquiring Hungarian, another language with a system of negative concord, showed a consistent adherence to nonadult conjunctive interpretations of negated disjunction. Clearly, more empirical investigations are needed to determine whether there are any real cross-linguistic differences with respect to the timing of acquisition of negated disjunctions and, if there are, to determine the source of the differences.

⁵ This is the group called “Child Free Choice (CFC)” in Singh et al. (2016:328, Table 6).

⁶ Epistemic modals like can typically give rise to Free Choice inferences with disjunction (when disjunction is not in the scope of negation). In contrast to English can, however, the Japanese potential morpheme -rare- lacks epistemic interpretation and exclusively expresses dynamic modality that relates to the ability or capacity of the subject (Kaufmann & Tamura 2017). Thus, adding the potential morpheme to a disjunctive sentence does not give rise to a Free Choice inference. For example, the following sentence does not allow the inference that John could eat sushi AND John could eat pasta; ka receives a normal disjunctive interpretation just as in sentences without -rare-.

Jon-wa sushi ka pasuta-o tabe-rare-taJohn-_TOP sushi or pasta-_ACC eat-can-_PAST‘John could eat sushi or pasta’

Therefore, we conclude that the use of the potential morpheme in the test sentences does not introduce a potential confounding factor that relates to Free Choice (and conjunctive) inferences to our experiment.

⁷ Notice that in the pair of sentences in (17), we use -re- instead of -rare- as the potential morphology. This is because the potential morpheme -rare- is now undergoing a language change (Fukushima 2004; Ito & Mester 2004), as younger generations have become more familiar with the newer, shortened form -re-, even though it is often considered “wrong” in prescriptive grammar. The -re- form can be found in very early child spontaneous speech, as well as in parental input. Observe the following examples from CHILDES database (MacWhinney 2000):

a. *MOT: ake-re-masu ka? open-can-_PRES Q “Can you open this?” *CHI: ake-re-ma: tsu@u open-can-PRES ‘I can open this’ (Asato, 2;08)

b. *CHI: datte ne gohan tabe-re-nai ja(n) because SFP food eat-can-NEG TAG ‘Because he cannot eat, can he?’ (Nanami, 2;07)

c. *MOT: ori-re-ru ? descend-can-_PRES ‘Can you come down?’ *CHI: ori-re-ta

descend-can-PAST ‘I could come down’ (Tai, 1;07)

These early spontaneous uses of the -re- form in the correct adultlike way suggest that children at our target age range (around age 5) have no problem with this shortened form.

⁸ The experiment is a revised version of a preliminary study reported in Shimada (2014). We modified the number of test items and conditions in accordance with the other experiments in the current study and administrated the experiment with completely new groups of participants.

⁹ This acceptance rate seems slightly lower than the corresponding rate from Experiment 1. We currently have no good explanation for this. The incorrect responses (i.e., rejecting the … mo … mo sentences in the black cross condition) appeared to have occurred rather randomly in this experiment, with no concrete sign of systematic misinterpretation of the relevant test sentences.

¹⁰ This assumption predicts that English or, a [–PPI] disjunction, should not show any scope ambiguities, especially when it appears in the object position of a negative sentence. This may appear to be too strong at first glance. Jing (2008) showed that, for example, with some manipulation of the experimental contexts, English-speaking adults and children accepted the disjunctive interpretation of negated or to a certain extent (acceptance rates were at around 50%). However, Goro (2007) argued, following Schwarz’s (1999) reduction theory, that the apparent disjunctive interpretation of English or is derived from a conjunction of two clauses and subsequent ellipsis:

[John didn’t eat sushi] or [John didn’t eat pasta]

Under this analysis, the disjunction or is interpreted outside the scope of negation but not through some covert scope-shifting operation: Its surface position is outside the scope of negation. Given the reduction analysis, the existence of apparent wide-scope interpretation of or can be made compatible with (32b).

Additional information

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.