While high-performing language models are typically trained on hundreds of billions of words, human children become fluent language users with a much smaller amount of data. What are the features of the data they receive, and how do these features support language modeling objectives? To investigate this question, we train GPT-2 and RoBERTa models on 29M words of English child-directed speech and a new matched, synthetic dataset (TinyDialogues), comparing to OpenSubtitles, Wikipedia, and a heterogeneous blend of datasets from the BabyLM challenge. We evaluate the syntactic and semantic knowledge of these models using developmentally-inspired evaluations. Through pretraining experiments, we test whether the global developmental ordering or the local discourse ordering of children's training data supports high performance relative to other datasets. The local properties of the data affect model results, but surprisingly, global properties do not. Further, child language input is not uniquely valuable for training language models. These findings support the hypothesis that, rather than proceeding from better data, the child's learning algorithm is substantially more data-efficient than current language modeling techniques.

Is Child-Directed Speech Effective Training Data for Language Models?

Is “cow” more closely related to “grass” or to “chicken”? Speakers of different languages judge similarity in this context differently, but why? One possibility is that cultures co-varying with these languages induce differences in conceptualizations of similarity. Specifically, East Asian cultures may promote reasoning about thematic similarity, by which cow and grass are more related, whereas Western cultures may bias judgments toward taxonomic relations, like cow and chicken. We measure similarity judgments across the US, China, and Vietnam and replicate US-China differences, but do not find that responding in Vietnam patterns with China. Instead, similarity judgments in Vietnam are intermediate between the US and China. We also show that word embedding models (fastText models for each language) are related to judgments within each country, suggesting a possible alternative interpretation of cross cultural differences. Perhaps notions of similarity are similar across contexts, but the statistics of the linguistic environment vary.


Re-examining cross-cultural similarity judgments using language statistics

Young children acquire a wide range of linguistic and cognitive skills in the first three years of life. Decades of experimental work have established a solid empirical foundation for our understanding of cognitive development. But most experimental studies are limited in statistical power and focus on specific psychological constructs, thus making them unsuitable for describing developmental growth at scale. Here, we turned to meta-analyses of experimental research. We conducted a meta-meta-analysis to consolidate and integrate 23 meta-analyses compiled on MetaLab, a community-augmented meta-analysis platform. We found that most datasets can not meaningfully distinguish different functional forms for developmental change, but in those that could, there is great diversity in the best-fitting functional forms of the age model. We also evaluated the impact of a range of methodological factors. Overall, our work sheds light on the heterogeneous nature of developmental trajectories and the subtle interactions between research methods and experimental outcomes.

A synthesis of early cognitive and language development using (meta-)meta-analysis

By the age of two, children tend to assume that new word categories are based on objects’ shape, rather than their color or texture; this assumption is called the shape bias. They are thought to learn this bias by observing that their caregiver’s language is biased towards shape based categories. This presents a chicken and egg problem: if the shape bias must be present in the language in order for children to learn it, how did it arise in language in the first place? In this paper, we propose that communicative efficiency explains both how the shape bias emerged and why it persists across generations. We model this process with neural emergent language agents that learn to communicate about raw pixelated images. First, we show that the shape bias emerges as a result of efficient communication strategies employed by agents. Second, we show that pressure brought on by communicative need is also necessary for it to persist across generations; simply having a shape bias in an agent’s input language is insufficient. These results suggest that, over and above the operation of other learning strategies, the shape bias in human learners may emerge and be sustained by communicative pressures.

The Emergence of the Shape Bias Results from Communicative Efficiency

Relational reasoning is an important component of abstract thought that emerges early in development but shows substantial variation across contexts, with children in the US and China following distinct developmental trajectories in one paradigm between 18m and 4y. To understand the mechanisms through which variation in the learning environment influences the development of relational reasoning, we examine early relational reasoning in Punjabi speakers in India, who share some cultural and linguistic elements of their experience with children in both the US and China. In a causal relational match-to-sample task, we find that 3-year-olds in India exhibit performance that is intermediate to their high-performing peers in China and the relatively poor performance observed in the US at this age. These results suggest complexity and variability in the development of relational reasoning and lay a foundation for future research designed to tease apart the factors associated with early diversity in relational reasoning.

Characterizing the development of relational reasoning in India

One problem language learners face is extracting word meanings from scenes with many possible referents. Despite the ambiguity of individual situations, a large body of empirical work shows that people are able to learn cross-situationally when a word occurs in different situations. Many computational models of cross-situational word learning have been proposed, yet there is little consensus on the main mechanisms supporting learning, in part due to the profusion of disparate studies and models, and lack of systematic model comparisons across a wide range of studies. This study compares the performance of several extant models on a dataset of 44 experimental conditions and a total of 1,696 participants. Using cross-validation, we fit multiple models representing theories of both
associative learning and hypothesis-testing theories of word learning, find two best-fitting models, and discuss issues of model and mechanism identifiability. Finally, we test the models' ability to generalize to additional experiments, including developmental data.

A large-scale comparison of cross-situational word learning models

Syntactic priming may be a key mechanism underlying children’s learning of novel words. Havron et al. (2019) exposed French-speaking children (ages 3 to 4) to a speaker biased by the use of either familiar verbs or nouns presented in the same syntactic context. This influenced participants’ interpretations of ambiguous novel words presented in the same syntactic frame. In Experiment 1, we successfully replicated Havron et al. with 77 French-speaking adults, using a web-based eye-tracking paradigm. Experiment 2 adapted this paradigm to English: Repeated exposure to a syntactic structure induced 102 English-speaking adults to update their expectations about the meanings of novel words. Our results indicate participants adapted to the specific linguistic structure used, not just the speaker’s tendency to mention actions or objects. These findings support the role of rapid adaptation during word learning. Experiments in progress investigate whether the English paradigm yields successful learning in 3- to 5-year-old children.

Syntactic adaptation and word learning in French and English

Human communication involves far more than words; speakers’ utterances are often accompanied by various kinds of emotional expressions. How do listeners represent and integrate these distinct sources of information to make communicative inferences? We first show that people, as listeners, integrate both verbal and emotional information when inferring true states of the world and others' communicative goals, and then present computational models that formalize these inferences by considering different ways in which these signals might be generated. Results suggest that while listeners understand that utterances and emotional expressions are generated by a balance of speakers’ informational and social goals, they additionally consider the possibility that emotional expressions are noncommunicative signals that directly reflect the speaker’s internal states. These results are consistent with the predictions of a probabilistic model that integrates goal inferences with linguistic and emotional signals, moving us towards a more complete formal theory of human communicative reasoning.

Integrating emotional expressions with utterances in pragmatic inference

Is “cow” more closely related to “grass” or “chicken”? Speakers of different languages judge similarity in this context differently, but why? One possibility is that cultures co-varying with these languages induce variation in conceptualizations of similarity. Specifically, East Asian cultures may promote reasoning about thematic similarity, by which cow and grass are more related whereas Western cultures may bias similarity judgements toward taxonomic relations, like cow-chicken. This difference in notions of similarity is the consensus interpretation for cross-cultural variation in this paradigm. We consider, and provide evidence for, an alternative possibility, by which notions of similarity are equivalent across contexts, but the statistics of the environment vary. On this account, similarity judgements are guided by co-occurrence in experience, and observing or hearing about cows and grass or cows and chickens more often could induce preferences for the relevant grouping, and account for apparent differences in notions of similarity across contexts.

Re-examining cross-cultural similarity judgements using lexical co-occurrence

Curiosity is a fundamental driver of human behavior, and yet because of its open-ended nature and the wide variety of behaviors it inspires in different contexts, it is remarkably difficult to study in a laboratory context. A promising approach to developing and testing theories of curiosity is to instantiate them in artificial agents that are able to act and explore in a simulated environment, and then compare the behavior of these agents to humans exploring the same stimuli. Here we propose a new experimental paradigm for examining children's -- and AI agents' -- curiosity about objects' physical interactions. We let them choose which object to drop another object onto in order to create the most interesting effect. We compared adults' (N=155) and children's choices (N=66; 3-7 year-olds) and found that both children and adults show a strong preference for choosing target objects that could potentially contain the dropped object. Adults alone also make choices consistent with achieving support relations. We contextualize our results using heuristic computational models based on 3D physical simulations of the same scenarios judged by participants.

Predicting children's and adults' preferences in physical interactions via physics simulation

What do infants and young children tend to see in their everyday lives? Relatively little work has examined the categories and objects that tend to be in the infant view during everyday experience, despite the fact that this knowledge is central to theories of category learning. Here, we analyzed the prevalence of the categories (e.g., people, animals, food) in the infant view in a longitudinal dataset of egocentric infant visual experience. Overall, we found a surprising amount of consistency in the broad characteristics of children's visual environment across individuals and across developmental time, in contrast to prior work examining the changing nature of the social signals in the infant view. In addition, we analyzed the distribution and identity of the categories that children tended touch and interact with in this dataset, generalizing previous findings that these objects tended to be distributed in a Zipfian manner. Taken together, these findings take a first step towards characterizing infants' changing visual environment, and call for future work to examine the generalizability of these results and to link them to learning outcomes.

Characterizing the object categories two children see and interact with in a dense dataset of naturalistic visual experience

People perceive, think, and act in a multitude of different ways across cultures, and there is an extensive history of research documenting these differences. At the heart of much of this work is a contrast between Western and East Asian cultures that has inspired important efforts to document human psychology in populations outside of the WEIRD (Western, Educated, Industrial, Rich, Democratic) demographic, which is much overrepresented within psychological research (Henrich, Heine, & Norenzayan, 2010). Recent recommendations for measuring cultural distance (Muthukrishna et al., 2020) profile the US and China as focal points for cultural comparisons, but define cultural distance using explicit self-report measures. Here, we evaluate cross-cultural differences between the US and China using implicit and experimental measures. We attempt to reproduce and test extensions of prior work demonstrating cross-cultural differences in reasoning, vision, and social cognition with convenience snowball samples of university students. Few of these differences appeared in our sample.

Investigating cross-cultural differences in reasoning, vision, and social cognition through replication

Curiosity drives much of human behavior, but its open-ended nature makes it hard to study in the laboratory. Moreover, computational theories of curiosity -- models of how intrinsic motivation promotes complex behaviors -- have been challenging to test because of technical limits. To circumvent this problem, we develop a new way to assess intrinsic motivation for building: we assume people build what they find interesting, so we asked them to rate the "interestingness" of visual stimuli -- in this case, simple block towers.
Adults gave a range of ratings to towers built by children, with taller towers rated higher.
To probe interestingness further, we developed controlled tower stimuli in a simulated 3D environment. While tower height predicted much of the variation in ratings, people also favored more precarious towers, as inferred from geometric features and simulated dynamics. These ratings and features therefore give a clear target for computational accounts of curiosity to explain.

Measuring and predicting variation in the interestingness of physical structures

The ability to form novel conventions is a key signature of efficient linguistic communication. Reduction of referring expressions, one measure of convention formation, is found robustly in dyadic repeated reference games when the target images are initially difficult to name. In reference games, participants share the explicit goal of establishing joint reference. However, establishing reference is a key subgoal of many conversations where interlocuters have more complex goals. In the current work, we analyze a dataset where reference was embedded in strategic 3- player negotiation and coordination games. In these more complex games, we found that the patterns of reduction and convergence to conventions still held across two different incentive conditions, with some modest differences between the conditions.

Communicative reduction in referring expressions within a multi-player negotiation game

In many cultures, adults use simple, slow, and dynamic speech when talking to infants ("parentese," or infant-directed speech) and make expansive, repetitive movements when demonstrating object properties to infants ("motionese," or infant-directed actions). These modifications enhance infants’ attention to and learning about language and goal-directed actions. Adults’ interactions with infants are also full of emotions –– do adults also modify their emotional expressions when interacting with infants? Here we showed parents of infants (aged 6 to 14 months; N = 25) emotion-evoking pictures including colorful bubbles, adorable stuffed animals, yummy snacks, broken toys, dangerous fire, and rotten fruits. We asked parents to describe their feelings about these pictures either to their infant or to an adult partner (i.e., an experimenter). While the parents’ use of emotion words did not differ between conditions, their emotional expressions did: Their infant-directed emotional expressions were more positive when they discussed positive pictures and more negative when they discussed negative pictures compared to their adult-directed emotional expressions. These findings suggest that besides "parentese" and "motionese," there is also a unique form of emotional communication in parent-child interaction – "emotionese."

Adults tailor their emotional expressions to infants through "emotionese"

This poster session includes Main Conference posters and Findings from the following areas:

NLP Applications • Information Retrieval and Text Mining • Generation • Linguistic Theories, Cognitive Modeling and Psycholinguistics 

In-Person Poster Session A (Riverfront Hall)

poster

## Welcome to EMNLP 2024! 
We are excited to welcome you to one of the most prominent conferences in the field of Natural Language Processing. This year, EMNLP 2024 is being held in a hybrid format,
offering both virtual and in-person participation in beautiful Miami. Due to a record-breaking number of submissions, we've expanded the total number of accepted papers to accommodate more cutting-edge research from around the globe.
### [Conference Handbook](https://drive.google.com/file/d/1WPROgxjLAC96AJL7Ugy0tEnYm7dkrbHt/view?usp=sharing)

You are required to register for this event. **Please register [here](https://2024.emnlp.org/registration/).** The EMNLP 2024 event page on Underline will be open to public one week prior to the event.

Please register!

EMNLP 2024

EMNLP 2024 will take place in Miami, Florida from Nov 12th to Nov 16th, 2024, at the Hyatt Regency Miami Hote and on Underline for remote participants.

T.19: Culture and Language 1

technical paper

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>

CogSci 2023

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

T.03: Language Acquisition 1



[![](https://s3.amazonaws.com/pf-upload-01/u-59356/0/2021-09-18/me23j72/button_workshop-website.png)](https://www.conll.org/2021) 

W22: CoNLL 2021 Day 2

workshop paper

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible. The experience for on-site participants would closely approximate a normal pre-COVID *ACL conference, with 5-6 thematically organized parallel sessions and live Q/A and interactive discussion immediately after the talks. Presentations by virtual participants will be equitably interleaved with those of on-site participants, projected on the auditorium screens as if on-site, and also followed immediately by live Q/A and interactive discussion at a time during reasonable waking hours for the virtual presenter. For all participants, on-site and virtual, who are unable to attend a session due to either time-zone issues or because they are participating in another session live, talk recordings and slides will be available online at a minimum after the live presentation (and in many cases before as well), and questions may be submitted in advance on session-specific discussion boards and answered live in session with the usual visual aids if desired.

<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/ceby/" seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

Please Note: The EMNLP registration system is not currently connected to the underline site as we are still in the process of building out EMNLP 2021. You will receive access instructions from underline the week of November 1st. 

Access is given only to EMNLP upon registration, if you have not registered please do so [here](https://2021.emnlp.org/registration).

Registered attendees will receive access the week of November 1st.

EMNLP 2021

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible.


**For a full list of posters in this session, click the See All link on the right side of the page.**

Poster Session 1 -  D: Education, Development and Perspectives in Cognitive Science

### Looking for Cogsci 2023? Please [click here](https://underline.io/events/405/reception) to switch to 2023 edition.


**Dear Colleagues,**

Welcome to the 43rd Annual Conference of the Cognitive Science Society! 

This year’s conference brings together the latest research from all Cognitive Science and also highlights this year’s theme **Comparative Cognition: Animal Minds.** 

The program features four plenary speakers: Friederike Range (University of Veterinary Medicine, Vienna), Onur Güntürkün (Ruhr University Bochum), Julia Fischer (University of Göttingen), and Erich Jarvis (The Rockefeller University). The keynotes and invited symposia showcase the broad spectrum of ideas that encompass the conference theme of comparative cognition. 

Further, the program includes the Rumelhart Prize keynote presentation by Susan Goldin-Meadow (University of Chicago), and symposia to celebrate the winner of the Elman Prize: Wei Ji Ma (New York University), the Heineken Award winner Robert Zatorre (McGill University), and the Robert J. Glushko Dissertation Prizes session.

The program committee for CogSci 2021 received 1,025 submissions, including 710 full papers, 275 member abstracts, 8 publication-based talks, as well as 15 proposals for symposia, and 17 for workshops and tutorials. After a rigorous review process, the committee selected 152 papers for oral presentation and inclusion in the conference proceedings (14.8%), 332 papers for poster presentation and inclusion in the proceedings (32.4%), and 175 papers for poster presentation with inclusion of abstracts in the proceedings (17.1%) We also selected 275 submitted member abstracts. In addition, we accepted 6 publication-based talks, 12 symposia, and 9 workshops and tutorials to make for an exciting and inclusive program.

We hope that you enjoy the conference this year! 

**Your Program Chairs,**
![](https://assets.underline.io/uploads/markdown_image/1/image/99984908caba5e6453e5fc5ba34f3507.jpg)

**Learn more about the Vienna Cognitive Science Hub at the University of Vienna, hosts of CogSci 2021**

<iframe width="560" height="315" src="https://www.youtube.com/embed/464h1m3iU68" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>

To review the proceedings for the conference or review the searchable program, please visit the [website](https://cognitivesciencesociety.org/program/).

CogSci 2021

The Cognitive Science Society brings together researchers from around the world who hold a common goal: understanding the nature of the human mind. The mission of the Society is to promote Cognitive Science as a discipline, and to foster scientific interchange among researchers in various areas of study, including Artificial Intelligence, Linguistics, Anthropology, Psychology, Neuroscience, Philosophy, and Education. The Society is a non-profit professional organization and its activities include sponsoring an annual conference and publishing the journals Cognitive Science and TopiCS.

**Date: **Wednesday, July 28, 2021 
**Time: **17:20 CEST - 19:00 CEST 
**Session Description: ** Review the posters you are interested in, watch the videos, and even leave messages for the authors in the Q&A section before joining the poster presenters and exhibitors in Gather.town for live discussion. 
**For a full list of posters in this session, click the See All link on the right side of the page.**

Poster Session 2 -  D: Education, Development and Perspectives in Cognitive Science

Poster Session 2 -  E: Linguistics

Poster Session 3 -  E: Linguistics

**Session Chair: **Daphna Buchsbaum 
**Technical Session Chair: **Alon Hafri 
**Session Description: ** The session is comprised of 15 minute talks followed by up to 5 minutes of live Q&A time for each presenter.

IP8: Development models

**Session Chair: **Kenny Smith 
**Technical Session Chair: **Merle Kortendieck 
**Session Description: ** The session is comprised of 15 minute talks followed by up to 5 minutes of live Q&A time for each presenter.

IP26: Social - Groups, norms, culture

**Session Chair: **Caren Walker 
**Technical Session Chair: **Jonathan Kominsky 
**Session Description: ** The session is comprised of 15 minute talks followed by up to 5 minutes of live Q&A time for each presenter.

IP31: Risk & decision making

This session has no scheduled virtual poster presentation time, but attendees will be able to comment and ask questions about your work/research asynchronously throughout the conference. 

**If you are a virtual poster presenter**, you can leave a message in your poster area chat space to advise times (and time zone) that you will be available either in the chat box, or by setting up your own online meeting room (Zoom, etc.). 
**If you are an attendee**, please check each poster's chat box and arrange time with presenter.

Michael Frank

15

9

Presentations