Categories such as 'animal' or 'furniture' play a pivotal role in processing, organizing, and communicating world knowledge. Many theories and computational models of categorization exist, but evaluation has disproportionately focused on artificially simplified learning problems (e.g., by assuming a given set of relevant features or small data sets); and on English native speakers. This paper presents a large-scale computational study of category and feature learning. We approximate the learning environment with natural language text, and scale previous work in three ways: We (1) model the full complexity of the learning process, acquiring learning categories and structured features jointly; (2) study the generalizability of categorization models to five diverse languages; and (3) learn categorizations comprising hundreds of concepts and thousands of features. Our experiments show that meaningful representations emerge across languages. We further demonstrate a joint model of category and feature acquisition produces more relevant and coherent features than simpler models, suggesting it as an exploratory tool to support cross-cultural categorization studies.

Word associations are among the most common paradigms for studying the human mental lexicon. While their structure and underlying relation types have been widely studied, surprisingly little attention has been given to the question of why participants produce the observed associations. Answering this question would not only advance understanding of human cognition, but could also aid machines in learning and representing basic commonsense knowledge. Here, we introduce WAX: a large, crowd-sourced data set of 19K English word associations paired with human-generated free-text explanations and relation type labels. We present an efficient framework for eliciting associations together with explanations, and a comprehensive analysis of the emerging types of explanations and relations. We test language models in their ability to predict associations and generate explanations, demonstrating that models struggle to capture the diversity of human-produced associations, and suggesting WAX as a fertile resource for future research.

Categorization is ubiquitous in human cognition and society, and impacts how we perceive and understand the world. In reflecting the needs and perspectives of their creators, no categorization system is entirely objective, and inbuilt biases can have harmful social consequences. Here, we propose methods for quantifying three kinds of category biases in hierarchical category systems. We present a study on two widely used library classification systems (the DDC and LCC) as large-scale examples of human categorization, and use our methods to quantify bias towards content associated with western (vs non-western) concepts in topic areas including history and religion. We find consistent evidence for western bias and show that the DDC tends to exhibit more western bias than the LCC. Our methods are general, and can be used to survey biases across topic areas, bias attributes, and hierarchical category systems.

Natural language expresses new concepts by reusing existing words or coining new ones. Previous studies have examined these word formation strategies separately through a functional lens, but it is unclear why one strategy might be preferred over another. In this study, we hypothesize that communicative and cognitive efficiency might predict the choice between lexical reuse and compounding for expressing an emerging concept. We test our hypothesis by developing a computational analysis of English word meanings that emerged over the past century. Our results suggest that strategy choice may be explained partly by a pressure for least effort. Our work contributes a novel connection between strategy choice in word formation and functional theories of language.

Compounding is a common type of word formation extensively studied in linguistics and cognitive psychology. A growing line of research suggests that the lexicon supports efficient communication by balancing informativeness and simplicity. We propose that the formation of novel compounds reflects a similar tradeoff between informativeness and word length. We formalize this hypothesis in information-theoretic terms and develop a computational procedure to evaluate our hypothesis on English noun compounds that emerged over the past century. We find that attested compounds achieve more efficient tradeoffs between informativeness and word length than do alternative word forms. Our work demonstrates how word formation and compositionality can be connected with information-theoretic approaches to the design of the lexicon.