Search

In the registered study, we will test whether logical representations specify semantic features of lexicalised universal quantifiers. To gain insight into this question, we will conduct a French structural priming experiment that tests whether the universal quantifiers 'chaque' (‘every’) and 'tous' (‘all’) are represented alike at the logical form (Feiman & Snedeker, 2016; Slim, Lauwers & Hartsuiker, 2023). The experiment tests priming of logical representations in the interpretation of sentences like 'Chaque requin n'a pas attaqué le surfeur' (“Every shark did not attack the surfer”). These sentences contain a universal quantifier and a negator. These operators both take scope, and therefore this sentence corresponds to two possible logical representations: One in which the quantifier takes scope over the negator (the universal-wide reading, in which none of the sharks attacked the surfer), and one in which the negator takes scope over the universal quantifier (the negation-wide reading, in which some but not all sharks attacked the surfer). We will influence the interpretation of these sentences with a priming manipulation. We will vary the quantifier in the prime sentences between 'chaque' (‘every’) and 'tous' (‘all’). The target sentences always contain 'chaque'. If logical representations are sensitive to quantifier-specific lexical content (such as distributivity or scope-taking mechanisms; e.g., Beghelli & Stowell, 1997; Champollion, 2015), we would expect that priming only emerges in case prime and target share the same quantifier words. If, on the other hand, logical representations abstract away from the combinatorial properties of specific quantifiers, priming is expected to emerge both between and within quantifiers. Here, we will register our experimental plan. The analysis plan will be added later (before we will start our analyses).

Human adults have the ability to reason and talk about possible outcomes of non-actual events. One type of linguistic expression that encodes possibility is conditionals (“If/When A, (then) C”). In adult speech, “when” conditional clauses are often used by speakers to talk about antecedent scenarios A that they think are going to happen, while “if” conditional clauses are often used to talk about antecedent scenarios A that they are uncertain about. The goal of the present experiment is to test children’s comprehension of “if” and “when” by asking them to match “if” or “when” statements with events that are either certain or uncertain. We manipulate the (un)certainty of an outcome using images of sticker machines, inspired by gumball machines (Degen & Tanenhaus, 2015). On each trial, one machine only has stickers of one color (e.g. blue) and so is certain to produce a sticker of that color, while the other machine has two different colors of stickers, so which it will produce is uncertain. We show children a character in front of each machine, and ask them which character says “If I get a blue sticker…” and which says “When I get a blue sticker…”, to see if they can tell which character is certain about the sticker they’ll get, and which is uncertain.

To make our design complete, in E3 we test for ME in a focus condition (see other two registrations)

When children hear a novel term in the context of two potential referents--one familiar or already-named, and one novel--they tend to assume that the novel word picks out the unfamiliar referent, a pattern typically described as the “Mutual Exclusivity” (ME) effect. In a typical demonstration of ME (Markman and Wachtel 1988 et seq.), children are presented with two objects -- one of them is novel and the other has a familiar label (e.g. a car). Next, they are asked, e.g. to “Show me the dax!”; children as young as 17 months of age (Halberda 2003) reliably choose the as-yet unlabeled object. While there are several competing explanations for why children (and adults) in these tasks treat dax and car as being mutually exclusive in reference, all of them invoke a bias to avoid applying two labels to the same object. This study explores an alternative hypothesis that the exclusivity inference, and in turn the choice of referent, depends on the information structure of the carrier sentences. Information structure is a cover term for a set of pre-theoretic concepts (focus, givenness, topicality) that have to do with discourse coherence. We take a fairly standard (in Linguistics) assumption that information structure is an aspect of the syntactic representation of sentences, with specific rules for intonation and semantic interpretation. Relevantly here, we assume the existence of two distinct markers that serve to relate a sentence, and the information expressed by it, to the surrounding discourse (Kratzer & Selkirk, Buring, Schwarzschild): Givenness (G) markers, which are sensitive to whether the meaning of an expression has been made salient in and thus recoverable from the preceding discourse; given expressions are de-accented/de-stressed. Focus (F) markers, which evoke alternatives to an expression, thereby establishing contrast with some part of the preceding discourse; focused expressions are emphasized/stressed. Together, G- and F- markers signal which parts of a given sentence matches (is anaphoric to) and contrasts with parts of the surrounding discourse. Here, we test the hypothesis that mutual exclusivity inferences can arise in a ME-paradigm as a result of the selective use conditions (emphasis/stress or lack thereof) imposed by G- and F- marking. While past studies did not systematically manipulate this variable to our knowledge, we suggest they may have tended to present their linguistic stimuli as in (3), with prosodic prominence--and thus, presumed F-marking--on the novel label. We think this is likely because that is the grammatical way, in standard English, to refer to an object that is being introduced into the discourse for the first time. Look at the car. Now look at [the DAX]F. The use condition on F-marking demands that the context makes salient a distinct alternative to dax (i.e. a contrasting meaning). There is only one salient alternative to dax, namely car. In order to meet the use conditions on F-marking, car must not entail dax, i.e. dax must not be another name for car, nor should dax be a subtype of car. Therefore, the dax must not pick out something that can be picked out by the term the car, i.e. the item already pointed to. If contrast between F-marking and G-marking, rather than contrast between the labels, drives ME-effects, we predict that removing the F-marker should result in a change in referent selection even as the labels stay the same. If, instead, dax was G-marked (indicated by prosodic de-accenting), the comprehender should look for something in the preceding discourse that has already made salient the relevant meaning. In this case, ME-effects should no longer arise -- G-marking should lead participants to infer that the previously labeled and known object -- the car -- can be referred to as a dax. In this study, we manipulate F- and G- marking on the prompted noun-phrase and ask whether rates of ME-driven selection shifts as a consequence. In the first version of our paradigm, we seek to establish whether the cues in information structure alone are sufficient to produce an ME effect when the nouns/labels do not necessarily contrast. To avoid a contrast between a known and a novel label, we will show participants two novel objects. First, we will unambiguously label one of two objects (e.g. blicket). Then in our test phrase we will use the NP “the toy” either F-marked, or G-marked with corresponding predictions of how children should interpret them. We use the noun-phrase “the toy” instead of “the dax” as the target phrase that probes the ME effect because “toy” could be -- but does not have to be -- a superordinate category that describes both objects. The use of non-contrasting labels might make ME-inferences (the blicket is not a toy) less likely while making co-referential readings (the blicket is a kind of toy) more likely as a whole. This creates the condition to test whether information structure alone can induce an ME effect, independently of whether contrast between labels is also sufficient to do so. When hearing F-marking on “the toy”, children should assume that “the toy” is an alternative to a salient discourse antecedent - in particular one that cannot entail toy. The only salient antecedent is “blicket”. What follows is that “blicket” cannot entail toy. If this is true then it follows that the referent of “the toy” can only be the other object. On the other hand, when “the toy” is G-marked children should assume that the meaning is recoverable from the available discourse. As “blicket” is the only salient discourse antecedent they should assume that “the toy” refers to that.

In this study, we are interested in how children connect communicative messages to pick out and track entities in the world. For example, a situation in which there are multiple objects present and someone says to a child, “I like that red apple”, how does the child go about identifying the intended referent, namely the apple that the speaker likes? Referential expressions like that red apple contain at least two sorts of informative content: its descriptive content, like apple, is helpful in identifying the kind of thing the referent is; in additional, grammatical elements/determiners like that indicate whether the referent was previously talked about or new to the discourse. We probe the relative contributions of these two pieces of information to referent identification in a task that puts the two in conflict. In our task, an object transforms from one kind to another (for instance from a toy duck to a toy car), and then participants have to select a suitable referent in response to different prompts. The prompts vary in the article (a/another/that) and the nouns (duck/car) used. Furthermore, across participants, we will manipulate context by varying whether this transformation was witnessed by the character who utters the critical referring expression.

When children hear a novel term in the context of two potential referents--one familiar or already-named, and one novel--they tend to assume that the novel word picks out the unfamiliar referent, a pattern typically described as the “Mutual Exclusivity” (ME) effect. While there are several competing explanations of ME, all invoke biases to avoid applying two labels to the same object. We instead propose that the exclusivity inference may be part of the linguistic message itself. In a previous experiment, we explored the hypothesis that the exclusivity inference is part of the linguistic message, and in turn the choice of referent, depends on whether or not the noun label in the carrier sentences is Focus-marked (expressed in English via prosodic accenting). Under this hypothesis F(ocus)-marking on the critical NP (indicated by accenting) prompts listeners to exclude contrastive alternatives in the context (e.g. the already labeled or familiar object), resulting in an ME inference even if the noun is not novel. But, if the NP is marked as given (indicated by de-accenting), listeners should look for a coreferential salient discourse antecedent, resulting in no ME effect. To test these predictions, we manipulated F- and G-markings on the noun-phrase prompt. Crucially, in this previous experiment, we used a target label in the carrier phrase that could in principle apply to both objects (“the toy”), instead of using a novel noun. We found that children derived ME when toy was F-marked, but not when it was G-marked; in the latter case, they opted for the previously labeled object. In the current study we ask whether this finding extends to mutual exclusivity phenomena more generally, by switching the target noun phrase from “the toy” to a target noun phrase that with a novel noun (e.g. “the dax). In this case we expect that the Focus condition is trivial given previous findings in the exclusivity literature (and our previous experiment) and focus exclusively on the Given condition. The results of the Given condition could help to adjudicate how general are the effects of information structure on referent disambiguation.

The distinctly human ability to both represent number exactly and develop symbolic number systems has raised the question of whether such number concepts are culturally constructed through symbolic systems. Although previous work with innumerate and semi-numerate groups has provided some evidence that understanding exact equality is related to numeracy, it is possible that previous failures were driven by pragmatic factors, rather than the absence of conceptual knowledge. Here, we test whether such factors affect performance on a test of exact equality in 3- to 5-year-old children by modifying previous methods to draw children’s attention to number. We find no effect of highlighting exact equality, either through framing the task as a “Number” game or as a “Sharing” game. Instead, we replicate previous findings showing a link between numeracy and an understanding of exact equality, strengthening the proposal that exact number concepts are facilitated by the acquisition of symbolic number systems.

How do humans develop the capacity to reason? In five studies, we examined infants’ emerging ability to make exclusion inferences using negation, as in the disjunctive syllogism (P or Q; not P; therefore Q). Inspired by studies of non-human animals and older children, Experiments 1–3 used an exclusion task adapted from Call’s (2004) 2-cup exclusion paradigm and Experiments 4–5 used an exclusion task adapted from the blicket detector paradigm (Sobel & Kirkham, 2006). In both tasks, we found failure to make exclusion inferences at 15 months, fragile success at 17 months, and robust success by 20 months of age. These data converge with some prior evidence that fails to find a capacity to represent negation in infants younger than 15 months of age and conflict with other evidence from different paradigms that suggests infants do have this capacity. We discuss three different resolutions of these conflicting data, and suggest lines of further work that might adjudicate among them.