Timeline: October 1st, 2022-October 31st, 2024.
The Truth Value Judgment Task: No explicit conjunction in the verbal context.
Across several languages, children have been found to consistently accept disjunctive statements in scenarios where adults find conjunctive statements more adequate (Tieu et al. 2016, 2017). We adopt the methodology used in Tieu et al. (2017) in order to determine whether the effect found in English and Japanese replicates in Romanian. Specifically, we use a Truth Value Judgment Task (Crain & Thornton 1998, 2000) as a guessing game in order to evaluate children’s understanding of inclusive and exclusive disjunction. We find that Romanian children are inclusive with sau-based disjunctions (sau 'or', sau...sau 'either...or') but inclusive and conjunctive with fie...fie 'either...or'.
The Truth Value Judgment Task: With explicit conjunction in the verbal context.
According to the Alternative-based hypothesis (Tieu et al., 2017, a.o.), it might be possible to increase the rate with which children derive exclusive disjunction implicatures by explicitly contrasting disjunctive statements with conjunctive alternatives, like Gǎina a ȋmpins (şi) autobuzul şi avionul (‘The hen pushed (both) the bus and the plane.’), thus activating the <or, and> scale (Horn, 1972). Interestingly, we find that Romanian children are more exclusive in their interpretation of disjunction in the presence of relevant conjunctive questions (such as 'Did the hen push the bus and the plane?') but not in the presence of unrelated conjunctive assertions, i.e. both access to alternatives and relevance are needed to boost exclusivity implicatures.
The Coloring Book Task: With explicit conjunction in the visual context and action required.
We investigate children's handling of disjunction through an act-out task in which participants perform an action in relation to a disjunctive statement. Children’s task is to color the hen who pushed the bus or the plane. On the assumption that the Coloring Book Task is a more accurate methodological tool for testing children’s abilities (as also suggested by Zuckerman et al. 2016, Gerard et al. 2017, 2018, and Bleotu 2020), we expect children to be sensitive to the <or, and> scale (Horn 1972) and show be more adult-like in the Coloring Book Task. In line with our expectations, we do indeed find that many more children are adult-like in this task compared to the Truth Value Judgment Task.
DISJUNCTION AND NEGATION
We look at negative disjunctive statements such as 'Gǎina nu a ȋmpins autobuzul ori avionul' ('The hen pushed neither the bus nor the plane'), which is ambiguous between two readings: (i) a “narrow scope” reading, where the hen pushed neither of the two objects, and (ii) a “wide scope reading”, which generates an exclusivity implicature (Chierchia, 2013), where the hen pushed either the plane or the bus, but not both. Based on previous studies (Goro & Akiba 2004, Crain et al. 2013, Pagliarini et al. 2018, 2021, a.o.) that found that, across many languages, children associate negative disjunctive statements with a neither...nor interpretation, we might expect Romanian children to behave alike. Indeed, we find that, similarly to adults, most children interpret negative disjunctive statements as expressing a strong neither...nor meaning. Interestingly, the children who show this interpretation in negative contexts are very rarely conjunctive with positive disjunctive statements but instead inclusive. These results point to there being no correlation in strong interpretations across negative and positive statements. We argue that this suggests that strong interpretations may have different sources in sentences of different polarity: strengthening in positive statements but scopal in negative statements.
DISJUNCTION AND MODALITY
We investigate experimentally whether Romanian children derive from a sentence such as Gǎina poate sǎ ȋmpingǎ autobuzul sau avionul ("The hen may push the bus or the plane') the free choice inferences Gǎina poate sǎ ȋmpingǎ autobuzul şi gǎina poate sǎ ȋmpingǎ avionul (" The hen may push the bus and the hen may push the plane"). Interestingly, previous studies show that children seem to be able to derive them while not being able to derive scalar implicatures with disjunction (Tieu et al., 2016). We expect a similar performance in the case of Romanian children. We do indeed find that a considerable number of Romanian children derive free choice inferences, while not deriving implicatures. Interestingly, they seem to derive more free choice inferences with the simple disjunction sau than with the complex disjunction fie...fie. In the case of fie...fie, some children show a conjunctive interpretation of modalized disjunctive statements (see Cochard et al., 2024), assigning to fie...fie a conjunctive meaning (Gǎina poate sǎ ȋmpingǎ autobuzul şi avionul 'The hen may push the bus and the plane'), similarly to what happens in plain disjunctive statements.
We investigate experimentally agreement with singular disjuncts in the case of Romanian children and adults. Prior research (e..g. Foppolo & Staub 2020, Himmelreich & Hartmann 2023) has shown variability in whether adults prefer singular or plural agreement when both disjuncts are singular ('The bunny or the beaver is/are smelling the flower'), but the phenomenon has been relatively understudied in child language. We conducted a picture-based forced-choice production experiment and a parallel forced-choice comprehension experiment conducted with Romanian-speaking adults and 5-year-old children. The results reveal that children predominantly select the agreement pattern that matches the number of disjuncts verified in the context, while adults vary more in their response strategies, with some seeming to adhere to syntactic agreement rules, others following the semantic context, and still others seemingly oscillating freely between the two possible agreement patterns. Our findings suggest that agreement with disjunction evolves from a context-sensitive system in childhood to a more rule-based approach in adulthood, though some adults remain sensitive to the number of verified disjuncts. These results support both semantic-pragmatic and non-semantic (syntactic) approaches to agreement with singular disjuncts, and suggest the possibility of multiple grammars.