United States

Selective avoidance of uncongenial facts to users’ preexisting false belief is a biased click tendency that detracts the utility of fact-checking websites. This study examined the strength of this avoidance tendency and whether it can be reduced with interventions. In a preregistered experiment with 1,203 participants, we compared two different types of interventions: An intervention with instruction that directly calls for reflection via text (instruction intervention); An intervention with a ranking-biased order that induces users to click on what they easily see and vice versa (ranking-biased intervention). The results showed no significant effect of the instruction intervention. However, the ranking-biased intervention showed preventive outcomes regarding the selective avoidance behaviors of users and promoted their clicking on links to uncongenial facts. The ranking-biased intervention was effective for users with high reflexiveness as well as for users with low reflexiveness. We discuss implications for designing fact-checking websites based on user understanding.

CogSci 2023

Fight Bias with Bias: Two Interventions for Mitigating the Selective Avoidance of Clicking Uncongenial Facts

**Welcome to EMNLP 2023!**

On behalf of the EMNLP 2023 Organizing Committee, I extend a warm and heartfelt welcome to all of you to the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP). It is with immense pleasure and excitement that we gather here in Singapore, a vibrant hub of innovation and technological advancement.

The conference program is packed with insightful presentations, thought-provoking workshops, and engaging networking opportunities. In addition to the technical sessions, we have also planned several social events that will provide you with the opportunity to connect with your colleagues.

The conference is held in person at the Resorts World Convention Centre in Singapore, and available on-line with the help of Underline.

For those in Singapore, I hope that you will find time to explore the exciting Sentosa Island, Gardens by the Bay, Singapore Botanic Gardens and the many other attractions unique to Singapore.

Of course, we are grateful to our sponsors and partners for their generous support of EMNLP 2023, whose contributions make it possible for us to host this world-class event.

Yuji Matsumoto (RIKEN AIP) 
EMNLP 2023 General Chair

To access the **EMNLP 2023** event page on Underline, you need to register for the Conference. 
Please follow **[this link](https://2023.emnlp.org/registration/)** for more details.

EMNLP 2023

EMNLP 2023 took place in Singapore from Dec 6th to Dec 10th, 2023.

poster

The 45th Annual Meeting of the Cognitive Science Society will take place in Sydney, Australia as a modified hybrid conference. Presentations from the main-session room will be live-streamed every day for all events. This will include all invited talks and award sessions. Recordings of all CogSci 2023 sessions will be made available for on-demand viewing after the conference. 
The in-person component will be held at the International Convention Centre Sydney at Darling Harbour. The Conference will take place Wednesday July 26th – Saturday July 29th, 2023 (Wednesday, July 26th is the Workshops Day).

Join us!

 [**The Program at a Glance can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2023/07/COgSci-PAAG-V2.pdf)
 [**The Detailed Program can be found here**](https://online.fliphtml5.com/ebtyf/mswz/)
 

#### **CogSci 2023 Program Book**
<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/tzgk/" seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

The 45th Annual Meeting of the Cognitive Science Society presents the latest research across cognitive science and highlights the theme of Cognition in Context.

Interaction plays an essential role in language acquisition, but the specific ways in which its underlying mechanisms contribute are unclear. To investigate this question, we conducted a study in which participants learnt an artificial language through interaction with a computer program, in a modified director-matcher paradigm. We manipulated the level of task interactivity and tracked learning progress. We found that participants in the interactive condition learnt the language faster and were more accurate by the end of their training, and that the difference persisted a day later. However, the manipulation did not affect grammatical or wordform accuracy. The results suggest that small changes in interactivity can have a sustained impact on language learning, particularly on word learning. Our study also has implications for the development of digital language learning materials, and it highlights the potential of this paradigm for exploring language learning in an interactive manner in an online environment.

A small increase in interactivity boosts language learning in a virtual environment

Language input is potentially ambiguous in a number of ways. In order to process language effectively, language users need to resolve these ambiguities quickly and efficiently. Many sources of information are recruited to complete this process including contextual constraints, prosody, and verb biases. The current work focuses on the development of verb biases given children’s overreliance on them. To explore this issue, I examined the effect of syntactic priming on the interpretation of PP-attachment ambiguities by 5-year-old children. Three priming experiments utilized three different but related prime types: globally ambiguous PP-attachments, unambiguous attachments disambiguated by syntax, and unambiguous attachments disambiguated by pragmatics. Results demonstrated priming in all three experiments, although it was strongest when the primes themselves were ambiguous. This finding provides further evidence for comprehension priming in children, and suggests that their verb biases can be overcome given the appropriate resources.

Priming Children's Interpretation of Globally Ambiguous Sentences

Verb learning is a difficult, important task for young learners (e.g., Childers & Tomasello, 2002). Children often see different events while hearing new verbs and have to deduce which event is relevant for learning that verb’s meaning. We simulated this experience by showing 3 ½- and 4 ½-year-olds (n= 28) 3 events on each trial and presenting a new verb. Across 3 trials, 4-year-olds were able to focus on the repeated action and extend the verb (p=.02); 3-year-olds performed at chance. In Study 2, the same-aged children (n=26 to date) were shown a repeated agent across trials (facilitating structural alignment), six learning trials (more instances for statistical learning), or both a repeated agent and six learning trials. Analyses of the first two conditions suggest a difference between conditions (p<.10); data collection in progress. These studies are important for understanding cognitive processes that may underlie a key task, verb learning.

The Key is Comparison: How Children Learn Verbs When Confronted with Referential Ambiguity

Modelling of early language acquisition aims to understand how infants bootstrap their language skills. The modelling encompasses properties of the input data used for training the models, the cognitive hypotheses and their algorithmic implementations being tested, and the evaluation methodologies to compare models to human data. Recent developments have enabled the use of more naturalistic training data for computational models. This also motivates development of more naturalistic tests of model behaviour. A crucial step towards such an aim is to develop representative speech datasets consisting of speech heard by infants in their natural environments. However, a major drawback of such recordings is that they are typically noisy, and it is currently unclear how the sound quality could affect analyses and modelling experiments conducted on such data. In this paper, we explore this aspect for the case of infant-directed speech (IDS) and adult-directed speech (ADS) analysis. First, we manually and automatically annotated audio quality of utterances extracted from two corpora of child-centred long-form recordings (in English and French). We then compared acoustic features of IDS and ADS in an in-lab dataset and across different audio quality subsets of naturalistic data. Finally, we assessed how the audio quality and recording environment may change the conclusions of a modelling analysis using a recent self-supervised learning model. Our results show that the use of modest and high audio quality naturalistic speech data result in largely similar conclusions on IDS and ADS in terms of acoustic analyses and modelling experiments. We also found that an automatic sound quality assessment tool can be used to screen out useful parts of long-form recordings for a closer analysis with comparable results to that of manual quality annotation.

Analysing the Impact of Audio Quality on the use of Naturalistic Long-Form Recordings for Infant-Directed Speech Research

We investigate the interpretation of disjunction in child and adult Romanian via a replication of Tieu et al. (2017). Specifically, we target the simple disjunction sau ‘or’ (with two intonation patterns: neutral and marked), and the complex disjunction markers sau…sau and fie…fie ‘either…or’. In a predictive Truth Value Judgment Task, participants evaluated a puppet’s disjunctive guesses (The hen pushed the bus or the plane) after seeing the outcome. Adults assigned predominantly exclusive interpretations to both simple and complex disjunctions (The hen pushed only one). Children, however, generally interpreted sau (with both intonational patterns) and sau…sau inclusively (The hen pushed one and possibly both), while they interpreted fie…fie conjunctively (The hen pushed both). It would appear that at an initial developmental stage, morphological/prosodic markedness does not affect children’s interpretation of disjunction. We discuss several possible accounts for the observed variation among complex disjunctions in child Romanian.

Not all complex disjunctions are alike: On inclusive and conjunctive interpretations in child Romanian

The link between language and cognition is evident in the first months of life. As early as three months of age, a parent labeling an object that the infant is looking at, influences the infants’ thoughts about those objects (Ferry et al., 2010). Prior research investigating this question has been conducted using face-to-face methodologies (Fulkerson & Waxman, 2007). Due to COVID, we sought to validate an online counterpart of the categorization task using preferential-looking paradigms. In Experiment 1, we found that language labels facilitated object categorization for 10- to 12-month-old infants. In contrast, a control condition that presented the same labeling phrases played in reverse did not facilitate categorization. In Experiment 2, we found that labels did not facilitate object categorization among 6-month-old infants. In sum, the advantage of online infant testing may be restricted to older ages.

Mixed results testing categorization in infants using a preferential-looking paradigm on Zoom

Core vocabulary is a topic of huge interest in linguistics and has been studied from a wide variety of perspectives, such as language learning, dictionary studies, and cross-linguistically. In many of these conceptions, word frequency is widely considered the conventional measure of a word's coreness; however, this approach overlooks important aspects of mental representation like centrality in an associative semantic network. In this experiment, we compare different approaches to defining core words in a task that involves predicting missing words in sentences. Results showed that core words (regardless of definition) were easier to guess than non-core words, but that frequency-defined ones did not perform as well as expected given their higher predictability and the nature of the task. Analysis of incorrect responses also showed that people preferred to guess core words, simple synonyms, and words that are taxonomically related to the target. The findings suggest that how core vocabulary is defined depends in part on the nature of the task and that aspects of both mental representation and the linguistic environment play an important role.

Word Prediction in Context: An Empirical Investigation of Core Vocabulary

Objects can either be represented as independent individuals (“object-files”) or as members of a single collection (“ensemble”). A great deal of work over the past 40 years has explored these representational systems, largely in the visual domain. Far less is known about auditory objects. Here, we show that a property characteristic of visual object representation – that it can be modulated by simple linguistic framing – also applies to auditory objects. In particular, we show that using the expression “each sound” versus “every sound” can bias auditory object construal in the same way that using “each circle” versus “every circle” can bias visual object construal. These findings support the idea that object-files and ensembles are not limited to the visual domain, but are representational formats found more generally throughout cognition.

Language Shifts the Representation of Sounds in Time: From Auditory Individuals to Auditory Ensembles

The effect of a first language (L1) on the acquisition of a second language (L2) is widely acknowledged across psycho-linguistics field. However, prior studies have focused on limited L1-L2 pairs and raise questions to how do linguistic structures (syntactic, phonological, or genetics) either promote or hinder the learning of a new language. In current study, we collected standardized English assessment scores from English Language Learners at international schools from 36 countries and addressed the language pairs by comparing linguistic similarities between more than 30 L1s. We also examined if different linguistic structures of L1 influence the acquisition of L2 in terms of syntactic, phonological, and genetic similarities. As a result, we found older ELLs learn English at a faster rate compared to younger ELLs. Among older ELLs, students who speak L1 that shares linguistic similarities with English learn faster than ELLs whose L1 has a relatively distant relationship with English.

Will children learn English faster if their native language is similar to English?

By 24 months, toddlers can process language incrementally using phonology to disambiguate lexical candidates. Adult research shows incremental processing of word meaning; however, this aspect is unclear in research with toddlers. Thus, we investigated the effect of phonological and semantic competition on word recognition in 18- and 24-month-old toddlers. In a preferential-looking task, toddlers looked at semantic-related (taxonomic or thematic) or unrelated pairs of pictures with different or similar phonological onset. The results showed that the 18-month group looked at the target only when the onset of both pictures was different but not when they were similar -regardless of the type of relationship. By contrast, the 24-month group had a similar look pattern as the 18-month group, only in semantically related but not in unrelated pairs. This research suggests that at 24 months of age, toddlers can interpret language incrementally based on phonological and semantic representations.

Premium content

Downloads

Next from CogSci 2023

A small increase in interactivity boosts language learning in a virtual environment

Similar lecture

Analyzing Semantic Faithfulness of Language Models via Input Intervention on Question Answering | VIDEO