United States

Word learning models are typically evaluated as the problem of observing words together with sets of atomic objects and learn-ing an alignment between them. We use ADAM, a Python software platform for modeling grounded language acquisition, to evaluate a particular word learning model, Pursuit (Stevens, Gleitman, Trueswell, &amp; Yang, 2017),under more realistic learning conditions (see e.g. Gleitman and Trueswell (2020) for review). In particular, we manipulate the degree of referential ambiguity and the salience of attentional cues available to the learner, and we present extensions to Pursuit which address the challenges of non-atomic meanings and exploiting attentional cues.

CogSci 2021

Grounding Word Learning Across Situations

### 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!<br><br>

**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/).


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.

Word learning models are typically evaluated as the problem of observing words together with sets of atomic objects and learn-ing an alignment between them. We use ADAM, a Python software platform for modeling grounded language acquisition, to evaluate a particular word learning model, Pursuit (Stevens, Gleitman, Trueswell, & Yang, 2017),under more realistic learning conditions (see e.g. Gleitman and Trueswell (2020) for review). In particular, we manipulate the degree of referential ambiguity and the salience of attentional cues available to the learner, and we present extensions to Pursuit which address the challenges of non-atomic meanings and exploiting attentional cues.

poster

The meaning of complex expressions (“two apples”) is computed by accessing and combining the concepts linked to their constituent words (“two”, “apples”). Across three eye-tracking experiments (N = 60), we demonstrate that preverbal infants can perform such computations and successfully derive the meaning of novel quantified noun phrases. Experiment 1 established that 12-month-olds can learn two distinct novel labels (pseudowords) denoting a singleton or a pair. Experiments 2-3 indicated that infants combine the meanings of the newly learnt quantity labels with those of familiar kind labels. When presented with four potential referents (e.g., 1duck, 2ducks, 1ball, 2balls) and asked to look at one ball, infants oriented to the target satisfying the meaning of both labels (1ball) over the distractors satisfying the meaning of the labels separately (2balls, 1duck). Conceptual combination skills that enable complex thought seem to be operational in infancy, and can be triggered by linguistic stimuli.

Unfants combine kind and quantity concepts

Iconicity refers to instances in which the form of language resembles its meaning (Perniss et al., 2010). The most prominent example in spoken language is onomatopoeia (e.g., woosh, which sounds like a gust of wind). Here we tested whether iconicity makes language more vivid by depicting the sensorimotor experiences being referred to. In Experiment 1, 44 participants read ten short passages (five iconic, five non-iconic), and then rated their vividness on several scales. These passages differed on two key words, which were either iconic (e.g., screech) or non-iconic (e.g., yell). We found no evidence that iconic language was more vivid. In Experiment 2, 199 participants each rated one longer passage that was either iconic or non-iconic (differing on eight key words). We found only marginal evidence of iconic language being more vivid, on subscales related to felt vividness. These results suggest that iconicity does not make written language more vivid.

Is Iconic Language More Vivid?

As children acquire the knowledge of their language's morphology, they invariably discover the productive processes that can generalize to new words. Morphological learning is made challenging by the fact that even fully productive rules have exceptions, as in the well-known case of English past tense verbs, which features the -ed rule against the irregular verbs. The Tolerance Principle is a recent proposal that provides a precise threshold of exceptions that a productive rule can withstand. Its empirical application so far, however, requires the researcher to fully specify rules defined over a set of words. We propose a greedy search model that automatically hypothesizes rules and evaluates their productivity over a vocabulary. When the search for broader productivity fails, the model recursively subdivides the vocabulary and continues the search for productivity over narrower rules. Trained on psychologically realistic data from child-directed input, our model displays developmental patterns observed in child morphology acquisition, including the notoriously complex case of German noun pluralization. It also produces responses to nonce words that, despite receiving only a fraction of the training data, are more similar to those of human subjects than current neural network models' responses are.

The Greedy and Recursive Search for Morphological Productivity

Analyses of political discourse typically focus on the semantic content of politicians’ statements. The approach treats the meaning of a speaker’s words as independent from the speaker’s identity itself; however, there are reasons to believe that one might influence the other. Features of a speaker’s identity influence others’ judgements of their character (e.g., Kinzler & DeJesus, 2013), and thus speaker identity could influence listeners’ assessment of the semantics and validity of the statements themselves. Here, we collect U.S. participants’ judgements of the political orientation of different statements, from liberal to conservative, heard with one of three accents, a generic U.S. accent, a Southern U.S. accent or an Australian accent. In comparison to identical statements conveyed in the generic U.S. accent, participants tended to perceive the U.S. Southern accented statements as more conservative and the Australian accented statements as more liberal.

Judgement of political statements are influenced by speaker identity

Production of referring expressions (the dog, it, Snoopy) is a complex process regulated by a combination of linguistic and cognitive constraints. In this paper, we explore the impact of world knowledge on the types of references speakers produce, focusing on predictability of event progressions. We argue that speakers are more likely to use a full noun phrase rather than drop the subject or use a pronoun when they describe an event progression they find surprising. In order to avoid the influence of distributional properties of event descriptions, we created an artificial world and trained subjects to recognize typical collision-event progressions within. Speakers then described novel scenes, which either conformed to their expectations or violated them, in a free production experiment. The results reveal that unpredictable event progressions lead to a more frequent production of full noun phrases, in contrast to reduced linguistic expressions (pronouns and null subjects). We conclude that speakers choose more informative descriptions to talk about surprising events.

Speakers Use More Informative Referring Expressions to Describe Surprising Events

In this paper we use pupillometry, a non-invasive, naturalistic method of measuring attention and cognitive load, to measure the effect of stress clash (Chinése shíp) and its metrical repair (Chínese shíp) during auditory sentence processing. We addressed two main research questions. The first question explores whether phonologically-disfavored metrical structures induce processing costs indexed by changes in pupil size. The second investigates whether the application of an optional process of stress retraction called the Rhythm Rule (Liberman & Prince, 1977) ameliorates or compounds any general penalty for stress clash. We find that unrepaired stress clash leads to greater pupil diameter relative to non-clashing sequences, indicating increased attention and cognitive load. We also find that repaired sequences lead to a decrease in overall pupil diameter, indicating facilitation.

The online advantage of repairing metrical structure: Stress shift in pupillometry

When do people use knowledge about the world in order to comprehend language? We asked whether pronoun resolution decisions are influenced by knowledge about physical plausibility. Results showed that referents which are more physically plausible in described events were more likely to be selected as antecedents of ambiguous pronouns, implying that resolution decisions were driven by physical inference. An alternative explanation is that these decisions were driven instead by distributional word knowledge. We tested this by including predictions of a statistical language model (BERT) and found that physical plausibility explained variance on top of the statistical language model predictions. This indicates that at least part of people's pronoun resolution judgments comes from knowledge about the world and not the word. This result constrains psycholinguistic models of comprehension—world knowledge must influence propositional interpretation—and raises the broader question of how non-linguistic physical inference processes are incorporated during comprehension.

The Role of Physical Inference in Pronoun Resolution

It is often assumed that language is strongly compositional, i.e., that the meaning of complex expressions is uniquely determined by the meanings of their constituents and their mode of composition (Fodor, 1987). Compositionality naturally connects to a broadly modular architecture of the language faculty, according to which our capacity for assigning meaning relies exclusively on lexical and syntactic knowledge (Baggio et al., 2015). Here, we discuss several arguments against strong compositionality. One such argument focuses on novel experimental data on the interpretation of privative adjectives (e.g., ‘fake’) (Partee, 2007). These data show that the interpretation of these adjectives is inexorably connected to the conceptual structure of the modified noun. We argue that lexical and syntactic information serve as important cues for, but do not uniquely determine, the process of meaning assignment (Martin, 2016). We discuss consequences for semantic theorising and the cognitive architecture of the language faculty.

Compositionality, modularity, and the architecture of the language faculty

During lexical access, listeners turn unfolding phonetic input into words. We tested how participants interpret words that aren't in their lexicon, either due to their coarticulation cues or because they label a novel object. In a 2-picture Visual World study, 57 adults saw a familiar object and an unfamiliar object, while hearing sentences directing their gaze to the target in 3 conditions: with a familiar word (“crib”), a novel word (“crig”), or a familiar word with coarticulation cueing a novel word (“cri(g)b”). When coarticulation cues matched the novel word (“cri(g)b”), participants looked more at the unfamiliar object than when the cues matched the familiar word, suggesting lexical competition can include a novel word under appropriate circumstances. When hearing a novel word (e.g. “crig”), participants showed two patterns: Roughly half looked more at the unfamiliar object, as expected, while the rest surprisingly looked more at the familiar object. We discuss the interaction of mutual exclusivity, phonetic similarity, and coarticulation cues in driving lexical access.

Listeners can use coarticulation cues to predict an upcoming novel word

Metaphor is an important part of everyday thought and language. Although we are often not aware of metaphor in everyday speech, on occasion, a particularly creative or novel use of metaphor will make us pay attention. It has been hypothesized that one of the driving cognitive factors behind the use of novel metaphor is a need to describe a new reality (as opposed to a preexisting reality) that would otherwise be difficult to convey using conventionalized metaphor. To this extent, novel metaphor use in everyday language may be more associated with episodic memory demands in contrast to conventional metaphor that is associated with semantic memory. To test this idea we analyzed novel metaphor use in the Hippocorpus --- a corpus of more than 5000 recalled and imagined stories about memorable life events in the first person perspective. In this dataset, recalled events have been shown to rely on episodic memory to a greater extent than descriptions of imagined events (i.e., narrating an event as if it happened to you but not describing an event that actually happened to you), which largely draw on semantic memory. We hypothesized that novel metaphor use during event narration should be modulated by the extent to which language users are able to draw on primary experience to describe events. We found that novel metaphor counts in recalled events were significantly higher than imagined events. Importantly, we found that factors that influence the extent to which language users are able to draw on primary experience during event narration (i.e., openness to experience, similarity to one's own experience, and how memorable or important an event was) modulated novel metaphor use in different ways in imagined compared to recalled events. The work paves the way for using large scale corpora to analyze underlying cognitive processes that modulate metaphorical language use.

Downloads

Next from CogSci 2021

Unfants combine kind and quantity concepts

Similar lecture

Modelling Human Communication as a Rejection Game