Canada

Children do not learn language from passively analyzing correlations between language and observations, but from interaction with caregivers or peers. The non-nativist approach claims that the main driver of language learning should be to achieve communicative goals. Imitation, on the other hand, is another natural desire that many argue influences language learning. However, there are still gaps in the research on what roles communicative goals and imitating linguistic input play in language acquisition, due to the difficulty of performing comprehensive experiments with human learners. In this paper, we propose a computational framework using simulated experiments that allows us to compare the roles of the two drivers. Specifically, we simulate a two-way communication game between a speaker, corresponding to a language learner, and a listener, corresponding to a caregiver or teacher. The speaker&#39;s communicative goals are modeled as rewards for successful completion of a referential game, and imitation is performed by mimicking feedback from the listener. The listener adaptively chooses to give feedback and makes choices based on the speaker&#39;s utterances. 

With empirical results on naturalistic visual and language data, we find that communicative goals play an important role in driving language learning, whereas imitation accelerates the learning process. We also find that (1) models trained with communicative goals tend to use minimal vocabulary and utterances and overextend them to concepts outside the original word meanings; (2) the strategy with which the listener provides feedback also influences the learning results and speed. Code and data for replicating the experiments are available\footnote{\url{https://bit.ly/interactgym}} to spur future research on models for computational studies of language learning.

CogSci 2022

Language Learning from Communicative Goals and Linguistic Input

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
 The conference will be held in Toronto, Canada, July 9-14, 2023. 
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience. 
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**

ACL 2023

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

Children do not learn language from passively analyzing correlations between language and observations, but from interaction with caregivers or peers. The non-nativist approach claims that the main driver of language learning should be to achieve communicative goals. Imitation, on the other hand, is another natural desire that many argue influences language learning. However, there are still gaps in the research on what roles communicative goals and imitating linguistic input play in language acquisition, due to the difficulty of performing comprehensive experiments with human learners. In this paper, we propose a computational framework using simulated experiments that allows us to compare the roles of the two drivers. Specifically, we simulate a two-way communication game between a speaker, corresponding to a language learner, and a listener, corresponding to a caregiver or teacher. The speaker's communicative goals are modeled as rewards for successful completion of a referential game, and imitation is performed by mimicking feedback from the listener. The listener adaptively chooses to give feedback and makes choices based on the speaker's utterances. 

With empirical results on naturalistic visual and language data, we find that communicative goals play an important role in driving language learning, whereas imitation accelerates the learning process. We also find that (1) models trained with communicative goals tend to use minimal vocabulary and utterances and overextend them to concepts outside the original word meanings; (2) the strategy with which the listener provides feedback also influences the learning results and speed. Code and data for replicating the experiments are available\footnote{\url{https://bit.ly/interactgym}} to spur future research on models for computational studies of language learning.

technical paper

Join us for CogSci2022! Our 44th Annual Conference will be centered around the theme of diversity, with an accompanying range of keynotes and invited speakers who themselves highlight the diversity of our field. Our three symposia are Cognition across Diverse Populations, Cognition within Diverse Societies, and Global Challenges for Cognitive Science. The conference is being offered in a fully hybrid environment for onsite and online attendees.
 
 [**The Program at a Glance can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2022/07/Cogsci-2022-PAAG-Detailed-July-12final-1.pdf)
 [**The Detailed Program can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2022/08/DeArmond-CogSci2022-program-v113-final.pdf)

This conference requires a registration to attend. Please use the link below to secure your ticket. The website will be available to all registered attendees on July 25th. 

[Register to Attend](https://cognitivesciencesociety.org/registration/)

Registration Is Required

The 44th Annual Meeting of the Cognitive Science Society highlighted research on the theme Cognitive Diversity in addition to the full breadth and diversity of research topics offered by the society’s membership. 





Picture book reading is a common word-learning context from which parents repeatedly name objects to their child and it has been found to facilitate early word learning. To learn the correct word-object mappings in a book-reading context, infants need to be able to link what they see with what they hear. However, given multiple objects on every book page, it is not clear how infants direct their attention to objects named by parents. The aim of the current study is to examine how infants mechanistically discover the correct word-object mappings during book reading in real time. We used head-mounted eye-tracking during parent-infant picture book reading and measured the infant's moment-by-moment visual attention to the named referent. We also examined how gesture cues provided by both the child and the parent may influence infants' attention to the named target. We found that although parents provided many object labels during book reading, infants were not able to attend to the named objects easily. However, their abilities to follow and use gestures to direct the other social partner’s attention increase the chance of looking at the named target during parent naming.

Examining Real-time Attention Dynamics in Parent-infant Picture Book Reading

Children start to communicate and use language in social interactions from the very early stages in development. This allows them to experiment with their current linguistic knowledge and receive valuable feedback from their interlocutors. We conducted a large-scale corpus study to examine the quality of positive and negative Communicative Feedback signals that caregivers provide in terms of time-contingent responses and clarification requests. We found evidence for the effect of such feedback in supporting children’s production of intelligible speech.
The broad impact of this paper is to highlight how general social feedback mechanisms that govern human communication can also support child language acquisition.

Communicative Feedback as a Mechanism Supporting the Production of Intelligible Speech in Early Childhood

Do children learn language from the words that they produce themselves? Because children know that they have imperfect knowledge of language, they could simply ignore their own productions. However, children could also learn from their productions -- using what they say and how their caregivers respond to update their language models. Using irregular plurals as a case study, we conducted a large-scale corpus analysis and two experimental studies to understand the role of children's productions and caregivers' responses in language learning. We demonstrate that children do learn from their own production, with errorful utterances leading to more errors. However, at least in some contexts, children can use implicit corrections from parents to offset the negative effects of their errors. Children thus appear to learn not only from their caregivers' productions, but also from their own productions and from the relationship between the two.

Of mouses and mans: the role of production and feedback in language learning

According to cross-situational learning, infants aggregate statistical information across naming events to resolve ambiguous word-referent mappings. While lab experiments show that learners are sensitive to these statistics, in studies using naturalistic stimuli, adults often fail to infer the correct referent. Here, we examined how young learners' input “in the wild” differs from laboratory experiments. We analyzed the temporal and spatial regularities of parent naming events in a naturalistic dataset of parent-child play and tested their effect on infants' visual attention. Parents were less likely to name the same toy twice than to name two different toys in sequence, except at short lags (0>t>5s). Most of the visual scenes accompanying naming events were composed of several toys of approximately equal (small) size. Child-attention to the target toy appeared to be modulated primarily by size. These results underscore the importance of quantifying naturalistic statistical regularities for understanding the mechanisms of word learning.

Quantifying cross-situational statistics during parent-child toy play

Iconicity refers to a resemblance between word form and meaning. Previous work has shown that iconic words are learned earlier and processed faster. Here we examined whether iconicity would also affect a recognition memory task. We also manipulated the level at which items were encoded—with a focus on either their meaning or their form—in order to gain insight into the mechanism by which iconicity would affect memory. In comparison with non-iconic words, iconic words were associated with a higher false alarm rate, a lower d’ score, and a lower criterion C. We did not observe any interaction between iconicity and encoding condition. We also conducted an analysis of recognition memory megastudy data and found that iconicity was predictive of higher false alarms and a lower criterion C across 1,646 items. We interpret these results as suggesting that iconicity leads to a feeling of familiarity in recognition memory.

Effects of Iconicity in Recognition Memory

A hallmark of human memory is the ability to integrate discrete experiences into cognitive maps. A fundamental form of this integration is transitive inference (TI), in which overlapping premises (A < B, B < C) are integrated into a unified representation of a relational hierarchy (A < B < C). Few existing theories provide a mechanistic account of this construction of relational knowledge and how it is shaped by different training conditions. This study builds on recent findings that TI is facilitated by chaining of overlapping premises, with a new behavioral experiment confirming an advantage over non-overlapping sequences matched for premise frequency and spacing. A subsequent simulation study shows that the chaining effect is captured by a particle filter which performs approximate Bayesian inference about the latent hierarchy. These results provide a better understanding of how chaining shapes the construction of relational knowledge in the face of uncertainty and forgetting.

Modeling the effect of chained study in transitive inference

When concepts are retrieved from memory, this process occurs within a rich search space where multiple sources of information interact with each other. Although the mapping from wordform to meaning is generally considered to be arbitrary, there is recent evidence to suggest that form and meaning may be correlated in natural language, and semantic and phonological cues may interact during retrieval. However, whether phonology interacts with meaning-related information in deeper semantic retrieval tasks, and whether this interaction has broader implications for how we conceptualize semantic retrieval remains relatively understudied. We examined these questions within the framework of the semantic fluency task, where individuals were asked to retrieve as many exemplars as they could from a given category (e.g., animals) within a fixed period of time. Responses were more phonologically similar during later stages of retrieval, and greater phonological similarity across responses was associated with greater number of items produced. We formulated a nested set of optimal foraging models to evaluate the combined influence of semantic and phonological information on retrieval likelihood. Model comparisons revealed that a model that combined frequency, semantic, and phonological information locally to make within-category transitions but relied on only frequency as a global cue to make between-category transitions produced the best explanation of the behavioral data.

Mouse-mole-vole: The inconspicuous benefit of phonology during retrieval from semantic memory

Previous studies have shown that representationally complex referents are encoded slower into working memory (WM) but are retrieved faster (Hofmeister, 2011; Karimi & Ferreira, 2016). However, the cost of maintaining complex representations is still not well understood. Through two self-paced reading experiments, we investigated the cost of encoding, maintaining and retrieving complex representations in WM. While we replicated the facilitatory effect during retrieval, the slowdown during encoding was not consistent across our experiments. More critically, for the first time, our experiments demonstrated that maintaining complex representations in WM is less costly than maintaining their simple counterparts. Furthermore, we found that WM maintenance cost is reduced because complex target noun phrases are more distinct from other competing referents in WM than simple ones. Overall, our results showed that the semantic elaboration of complex representations can reduce maintenance cost and provided new perspectives into this understudied WM process.

Noun phrase representational complexity reduces maintenance cost in working memory by increasing distinctiveness between referents

Knowledge of cause and effect allows people to navigate and understand the complex systems of the world. Despite the importance of causal knowledge to everyday reasoning, little is known about how people transfer causal knowledge learned in one situation to novel contexts. In two experiments, we examine when people choose to generalize two types of causal knowledge, causal mechanisms (Experiment 1) and causal strength (Experiment 2), across various domains. We find that people willingly transfer causal knowledge to novel contexts when the entities in those contexts share high categorical relatedness with the source of the causal knowledge. The extent to which people are willing to transfer causal knowledge decreases as category similarity decreases. We discuss future research that could delineate the boundaries of causal transfer.

Transferring Novel Causal Knowledge

Self-directed learners are often described as ‘intuitive scientists’, yet they also tend to struggle in assessments of their scientific reasoning. We investigate a novel explanation for this apparent gap between formal and informal scientific inquiry. Specifically, we consider whether learners’ documented failure to correctly apply the control of variables strategy might stem from a mismatch between their causal intuitions and task presentation. Children (7- and 9-year-olds) and adults were tested on a version of a traditional multivariate reasoning task (Tschirgi, 1980) that we modified to clarify ambiguous elements of the causal logic. A significant majority of participants in all age groups selected informative experiments on this modified task, avoiding confounded actions with positive tangible outcomes. This finding contrasts with the longstanding claim that learners do not correctly employ control of variables without extensive training and suggests that self-directed scientific inquiry may be intuitively suited to support causal learning goals.

Premium content

Downloads

Next from CogSci 2022

Examining Real-time Attention Dynamics in Parent-infant Picture Book Reading

Similar lecture

Speaking the Language of Your Listener: Audience-Aware Adaptation via Plug-and-Play Theory of Mind