How Adults and Children Interpret Disjunction under Negation in Dutch, French, Hungarian and Italian: A Cross-Linguistic Comparison

ABSTRACT

In English, a sentence like “The cat didn’t eat the carrot or the pepper” typically receives a “neither” interpretation; in Japanese it receives a “not this or not that” interpretation. These two interpretations are in a subset/superset relation, such that the “neither” interpretation (strong reading) asymmetrically entails the “not this or not that” interpretation (weak reading). This asymmetrical entailment raises a learnability problem. According to the Semantic Subset Principle, all language learners, regardless of the language they are exposed to, start by assigning the strong reading, since this interpretation makes such sentences true in the narrowest range of circumstances.). If the “neither” interpretation is children’s initial hypothesis, then children acquiring a superset language will be able to revise their initial hypothesis on the basis of positive evidence.

The aim of the present study is to test an additional account proposed by Pagliarini, Crain, Guasti (2018) as a possible explanation for the earlier convergence to the adult grammar by Italian children. The hypothesis tested here is that the presence of a lexical form such as recursive né that unambiguously conveys a “neither” meaning, would lead children to converge earlier to the adult grammar due to a blocking effect of the recursive né form in the inventory of negated disjunction forms in a language. We compared data from Italian (taken from Pagliarini, Crain, Guasti, 2018), French, Hungarian and Dutch. Dutch was tested as baseline language. French and Hungarian have – similarly to Italian – a lexical form that unambiguously expresses the “neither” interpretation (ni ni and sem sem, respectively). Our results did not support this hypothesis however, and are discussed in the light of language-specific particularities of the syntax and semantics of negation.

Acknowledgments

We would like to thank all the children who participated in the study, their teachers and their parents. The Dutch part of the project was done in close collaboration with Jack Hoeksema. We gratefully acknowledge his input as well as the feedback from the Acquisition Lab participants at the Center for Cognition and Language Groningen. We are very grateful to Florent Roul and Caroline Plet, who helped collecting the French data, and Mieke Slim and Annika van Wijk for collecting the Dutch data. Thanks also to Dorothy An, for providing the pictures used in the experiment.

The work of L. Pintér and B. Surányi was supported by grant no. KH 130558 of the Hungarian National Research, Development and Innovation Fund.

We gratefully acknowledge support from the Netherlands Organization for Scientific Research NWO (GraMALL, Grasping Meaning across Languages and Learners, 2014-2017, PI A. van Hout).”

Authors’ contributions

E.P. and M.T.G. conceived the project, E.P designed the experiment, A.v.H. O.L., L.P., B.S. adapted the material to the respective languages; A.v. H, O.L., E.P., L.P. collected the data, E.P. analyzed the data, E.P. wrote and revised all the sections of the paper, A.v. H., O.L., L.P., B.S., S.C and M.T.G. revised various sections of the paper.

Disclosure statement

The authors declare that they have no competing interests.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

¹ Throughout the paper, when referring to the expression “negative sentences with disjunction”, we mean ‘negated sentences containing disjunction in the verb phrase’.

² Throughout the paper, when we are using the “not this or not that” label for the interpretation, the sentence can be true in three circumstances {A-AND-notB, notA-AND-B, notA-AND-notB}.

³ More accurately, the relevant rule of classical logic simply does not apply to negative sentences with disjunction in Mandarin Chinese and Japanese, because disjunction does not reside within the scope of negation at the level of semantic interpretation.

⁴ Importantly, an expression that is –PPI is not a Negative Polarity Item (NPI). Rather, it is an expression that can be interpreted in the local scope of negation and other downward entailing operators.

⁵ The polarity sensitivity of disjunction is also neutralized when negation appears outside the clause that contains disjunction, such that negation takes wide scope. This is characteristic of Positive Polarity Items more generally (see Crain, 2012).

⁶ In Turkish disjunction is expressed by two distinct morphemes, ya ya da and veya. Both expressions are tolerated with both case-marked and not case-marked disjunction phrases. The results by Geçkin et al. (2015) showed that Turkish-speaking children assigned a “neither” interpretation to negative sentences with disjunction, regardless of case marking, whereas adults assigned a “neither” interpretation only to sentences in which the disjunctive phrase was not case-marked; when the disjunction phrase was case-marked, adults assigned a “not this or not that” interpretation.

⁷ The predictions made by the SSP are not different from those that can be derived from Dalrymple et al.’s (1994) Strongest Meaning Hypothesis (SMH) (Dalrymple et al., 1998, 1994). The SMH is a principle governing adult language, initially proposed as specific to the interpretation of reciprocals but it has been – and can be – generalized beyond that. As such, the SMH is a principle of preferential interpretation proposed independently of the developmental SSP. The two bear an obvious similarity, and therefore provide indirect support for each other, since we can see essentially the same mechanism at play both in the processing of meaning by adults and in the acquisition of meanings in children.

⁸ Two experiments were run: in the first experiment, children ranging in age from 3;5 to 6;0 were tested. In the second experiment, children ranging in age from 4;10 to 5;7 were tested. Both experiments revealed similar results.

⁹ However, when né né is in preverbal position, it does not occur with negation, as Italian is a non-strict negative concord language (e.g., Giannakidou, 2006; Zanuttini, 1997; Zeijlstra, 2004).

¹⁰ PRT stands for verbal particle.

¹¹ Van Bergen and De Swart (2010) found in a corpus study that definite NPs do not scramble that much, but their data involved all sorts of adverbial elements, including adverbs in higher positions than negation. For negation, we maintain the position that scrambling is preferred.

¹² Hoeksema (2014) provides corpus evidence which shows that Dutch main clauses with negation hardly ever have scrambling of the PP to a position in front of negation (0% in cases where the PP is followed by a finite verb, 8% in cases where there is a nonfinite verb following the PP). Note that the Dutch test sentences like (16) all involved sentences with finite verbs.

¹³ As suggested by an anonymous reviewer, the performance of the Italian children might be just due to sampling bias, given our modest data set. The bimodal distribution of responses together with the high rate of accuracy on the filler sentences, suggest that children understood the task. Nevertheless, further replications with bigger sample size are needed in order to be able to generalize to other Italian-speaking children. Indeed, our recent new set of data (which was presented at 46° Incontro di Grammatica Generativa as Silleresi, S., Pagliarini, E. & Guasti, M. T. When the interpretation of disjuncted negative sentences varies between two languages: a study on Italian-English bilingual children) replicates the same pattern of results presented in Pagliarini, Crain, Guasti (2018).

¹⁴ PRT stands for verbal particle. The verbal particles in the Hungarian examples in this paper have a telicizing function and are independent of Negative Concord.

¹⁵ DIM stands for diminutive affix.

¹⁶ One possible way to conceptualize this situation is to say that OR is +PPI in Turkish. When nouns are accusative marked, they are definite and OR can scope out (along with the NPs it occurs with). When nouns are not accusative case marked they are not definite. Wide scope is not possible, because these NPs have to be licensed by negation. A similar situation occurs in Italian. When one of the disjuncts includes a NPI, the “not this or not that” reading is no longer available (the “neither” reading becomes available), as the disjunctive phrase must be licensed by the negation for the NPI to be licensed. This is seen in (i):

1. Da quel giorno Mario non lesse alcun giornale o settimanale. From that day Mario NEG read any journal or weekly publication.

This explanation is different from that in Geçkin et al. (2015)

¹⁷ It is also possible that Turkish-speaking children may not have been sensitive to case-marking (definiteness) in addition to the fact that they may be puzzled by the existence of two patterns depending on case marking. Further investigation is needed to establish whether Turkish-speaking children are indeed sensitive to case marking.