The Small World of Words: English word association norms for over 12,000 cue words

Deyne, Simon De; Navarro, Daniel J.; Perfors, Amy; Brysbaert, Marc; Storms, Gerrit

doi:10.31234/osf.io/mb93p

Cited by 5 publications

(10 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Humans rely on rich semantic networks of conceptual knowledge in order to effectively process the world around them and to facilitate inference, decision making, and memory retrieval (Collins and Quillian, 1969;Collins and Loftus, 1975;Griffiths, Steyvers, and Tenenbaum, 2007;Griffiths, Steyvers, and Firl, 2007;De Deyne and Storms, 2008;Abbott et al, 2015;De Deyne et al, 2019;Lynn and Bassett, 2020). Understanding the structure of those semantic networks, and the computations that they support, has been central to the field of cognitive science for decades (Collins and Quillian, 1969;Collins and Loftus, 1975).…”

Section: Characterizing the Large-scale Structure Of Grounded Semanti...mentioning

confidence: 99%

Characterizing the Large-Scale Structure of Grounded Semantic Networks

Marjieh,

van Rijn,

Sucholutsky

et al. 2024

Preprint

View full text Add to dashboard Cite

The large-scale structure of semantic networks such as those derived from word association tasks and large lexical corpora has been shown to exhibit small-world and scale-free regularities. This is important as it challenges traditional accounts of semantic organization based on hierarchical and random structures, and it suggests how semantic growth can occur through a process of conceptual differentiation whereby new concepts inherit a subset of the associations of older concepts. Despite its significance, the evidence for the large-scale structure of semantic networks is indirect as the extent to which it reflects the naturally occurring semantic regularities of the environment has not been investigated systematically. To address this, we collected and analyzed semantic descriptors, human evaluations, and similarity judgments from four large datasets of naturalistic stimuli across three modalities (visual, auditory, and audio-visual) comprising 7,916 stimuli and 610,841 human responses. We show that the resulting grounded networks are not only predictive of direct sensory judgments on these domains, but also exhibit clear small-world and scale-free statistics. Our findings shed new light on the origins of the structure of semantic networks by grounding it in the semantic regularities of the environment.

show abstract

Section: Characterizing the Large-scale Structure Of Grounded Semanti...mentioning

confidence: 99%

Characterizing the Large-Scale Structure of Grounded Semantic Networks

Marjieh,

van Rijn,

Sucholutsky

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

Section: Network Building and Semantic Frame Reconstructionmentioning

confidence: 99%

“…In the literature about semantic networks, there exist several ways to connect cue words to their associative responses [19,20,36]. We chose to connect each cue word to all three of its responses since this method has been shown to provide more heterogeneity in associative responses [20] and has been used in previous works with forma mentis networks [1,63,65]. Moreover, this approach to network construction has been shown to improve the accuracy of many language-related prediction tasks (such as associative strength prediction) compared to other strategies, e.g., connecting the cue word to the first response only [20].…”

Section: Network Building and Semantic Frame Reconstructionmentioning

confidence: 99%

“…We chose to connect each cue word to all three of its responses since this method has been shown to provide more heterogeneity in associative responses [20] and has been used in previous works with forma mentis networks [1,63,65]. Moreover, this approach to network construction has been shown to improve the accuracy of many language-related prediction tasks (such as associative strength prediction) compared to other strategies, e.g., connecting the cue word to the first response only [20]. We also considered idiosyncratic associations, i.e., associations provided only one time, which are visually represented as narrower edges compared to non-idiosyncratic associations.…”

Section: Network Building and Semantic Frame Reconstructionmentioning

confidence: 99%

“…Firstly, in cognitive psychology, there must be an adequate match between a given implemented task measuring a target process and the cognitive theory or model used to explain that process [4,6]. For instance, past works have established a quantitative and theory-driven link between continued free association tasks -deriving free associations between concepts -and models based on such data whose network structure could explain aspects of conceptual recall from semantic memory [19,20] or even higher-level phenomena like problem solving [36]. For instance, according to the spreading activation model established by pioneering work of psychologists Collins and Loftus [16], providing an individual with a cue word activates a cognitive process acting on a network representation, such that concepts are nodes linked together via conceptual associations.…”

Section: A Focus On the Evolution Of Math Perceptions In Llmsmentioning

confidence: 99%

See 2 more Smart Citations

Cognitive network science reveals bias in GPT-3, ChatGPT, and GPT-4 mirroring math anxiety in high-school students

Abramski,

Citraro,

Lombardi

et al. 2023

Preprint

View full text Add to dashboard Cite

Large language models are becoming increasingly integrated into our lives. Hence, it is important to understand the biases present in their outputs in order to avoid perpetuating harmful stereotypes, which originate in our own flawed ways of thinking. This challenge requires developing new benchmarks and methods for quantifying affective and semantic bias, keeping in mind that LLMs act as psycho-social mirrors that reflect the views and tendencies that are prevalent in society. One such tendency that has harmful negative effects is the global phenomenon of anxiety toward math and STEM subjects. Here, we investigate perceptions of math and STEM fields provided by cutting-edge language models, namely GPT-3, Chat-GPT, and GPT-4, by applying an approach from network science and cognitive psychology. Specifically, we use behavioral forma mentis networks (BFMNs) to understand how these LLMs frame math and STEM disciplines in relation to other concepts. We use data obtained by probing the three LLMs in a language generation task that has previously been applied to humans. Our findings indicate that LLMs have an overall negative perception of math and STEM fields, with math being perceived most negatively. We observe significant differences across the three LLMs. We observe that newer versions (i.e. GPT-4) produce richer, more complex perceptions as well as less negative perceptions compared to older versions and N=159 high-school students. These findings suggest that advances in the architecture of LLMs may lead to increasingly less biased models that could even perhaps someday aid in reducing harmful stereotypes in society rather than perpetuating them.

show abstract

Latent Relations at Steady‐state with Associative Nets

Shabahang,

Yim,

Dennis

2024

Cognitive Science

View full text Add to dashboard Cite

Models of word meaning that exploit patterns of word usage across large text corpora to capture semantic relations, like the topic model and word2vec, condense word‐by‐context co‐occurrence statistics to induce representations that organize words along semantically relevant dimensions (e.g., synonymy, antonymy, hyponymy, etc.). However, their reliance on latent representations leaves them vulnerable to interference, makes them slow learners, and commits to a dual‐systems account of episodic and semantic memory. We show how it is possible to construct the meaning of words online during retrieval to avoid these limitations. We implement a spreading activation account of word meaning in an associative net, a one‐layer highly recurrent network of associations, called a Dynamic‐Eigen‐Net, that we developed to address the limitations of earlier variants of associative nets when scaling up to deal with unstructured input domains like natural language text. We show that spreading activation using a one‐hot coded Dynamic‐Eigen‐Net outperforms the topic model and reaches similar levels of performance as word2vec when predicting human free associations and word similarity ratings. Latent Semantic Analysis vectors reached similar levels of performance when constructed by applying dimensionality reduction to the Shifted Positive Pointwise Mutual Information but showed poorer predictability for free associations when using an entropy‐based normalization. An analysis of the rate at which the Dynamic‐Eigen‐Net reaches asymptotic performance shows that it learns faster than word2vec. We argue in favor of the Dynamic‐Eigen‐Net as a fast learner, with a single‐store, that is not subject to catastrophic interference. We present it as an alternative to instance models when delegating the induction of latent relationships to process assumptions instead of assumptions about representation.

show abstract

The Small World of Words: English word association norms for over 12,000 cue words

Cited by 5 publications

References 43 publications

Characterizing the Large-Scale Structure of Grounded Semantic Networks

Characterizing the Large-Scale Structure of Grounded Semantic Networks

Cognitive network science reveals bias in GPT-3, ChatGPT, and GPT-4 mirroring math anxiety in high-school students

Latent Relations at Steady‐state with Associative Nets

Contact Info

Product

Resources

About