Neural network models have recently made striking progress in natural language processing, but they are typically trained on orders of magnitude more language input than children receive. What can these neural networks, which are primarily distributional learners, learn from a naturalistic subset of a single child's experience? We examine this question using a recent longitudinal dataset collected from a single child, consisting of egocentric visual data paired with text transcripts. We train both language-only and vision-and-language neural networks and analyze the (...) linguistic knowledge they acquire. In parallel with findings from Jeffrey Elman's seminal work, the neural networks form emergent clusters of words corresponding to syntactic (nouns, transitive and intransitive verbs) and semantic categories (e.g., animals and clothing), based solely on one child's linguistic input. The networks also acquire sensitivity to acceptability contrasts from linguistic phenomena, such as determiner-noun agreement and argument structure. We find that incorporating visual information produces an incremental gain in predicting words in context, especially for syntactic categories that are comparatively more easily grounded, such as nouns and verbs, but the underlying linguistic representations are not fundamentally altered. Our findings demonstrate which kinds of linguistic knowledge are learnable from a snapshot of a single child's real developmental experience. (shrink)

The Affective grounding hypothesis suggests that affective experiences play a crucial role in abstract concepts’ processing (Kousta et al. 2011). Vigliocco and colleagues test the role of affective experiences as well as the role of language in learning words denoting abstract concepts, comparing children with typical and atypical development. They conclude that besides the affective experiences also language plays a critical role in the processing of words referring to abstract concepts.

Brain damage may doubly dissociate cognitive modules, but the practice of revealing dissociations is predicated on modularity being true (T. Shallice, 1988). This article questions the utility of assuming modularity, as it examines a paradigmatic double dissociation of reading modules. Reading modules illustrate two general problems. First, modularity fails to converge on a fixed set of exclusionary criteria that define pure cases. As a consequence, competing modular theories force perennial quests for purer cases, which simply perpetuates growth in the list (...) of exclusionary criteria. The first problem leads, in part, to the second problem. Modularity fails to converge on a fixed set of pure cases. The second failure perpetuates unending fractionation into more modules. (shrink)

Conceptual knowledge is acquired through recurrent experiences, by extracting statistical regularities at different levels of granularity. At a fine level, patterns of feature co-occurrence are categorized into objects. At a coarser level, patterns of concept co-occurrence are categorized into contexts. We present and test CONCAT, a connectionist model that simultaneously learns to categorize objects and contexts. The model contains two hierarchically organized CALM modules (Murre, Phaf, & Wolters, 1992). The first module, the Object Module, forms object representations based on co-occurrences (...) between features. These representations are used as input for the second module, the Context Module, which categorizes contexts based on object co-occurrences. Feedback connections from the Context Module to the Object Module send activation from the active context to those objects that frequently occur within this context. We demonstrate that context feedback contributes to the successful categorization of objects, especially when bottom-up feature information is degraded or ambiguous. (shrink)

A large literature suggests that the organization of words in semantic memory, reflecting meaningful relations among words and the concepts to which they refer, supports many cognitive processes, including memory encoding and retrieval, word learning, and inferential reasoning. The co‐activation of related items has been proposed as a mechanism by which semantic knowledge influences cognition, and contemporary accounts of semantic knowledge propose that this co‐activation is graded—that it depends on how strongly related the items are in semantic memory. Prior research (...) with adults yielded evidence supporting this prediction; however, there is currently no evidence of graded co‐activation early in development. This study provides the first evidence that in children the co‐activation of related items depends on their relational strength in semantic memory. Participants (N = 84, age range: 3–9 years) were asked to identify a target (e.g., bone) amid distractors. Children's responses were slowed down by the presence of a related distractor (e.g., puppy) relative to unrelated distractors (e.g., flower)—suggesting that children co‐activated related items upon hearing the name of the target. Importantly, the degree of this co‐activation was predicted by the strength of the target–distractor relation, such that distractors more strongly related to the targets slowed down children to a larger extent. These findings have important implications for understanding how organized semantic knowledge affects other cognitive processes across development. (shrink)

Recent metaphor research has revealed that metaphor comprehension involves both categorization and comparison processes. This finding has triggered the following central question: Which property determines the choice between these two processes for metaphor comprehension? Three competing views have been proposed to answer this question: the conventionality view (Bowdle & Gentner, 2005), aptness view (Glucksberg & Haught, 2006b), and interpretive diversity view (Utsumi, 2007); these views, respectively, argue that vehicle conventionality, metaphor aptness, and interpretive diversity determine the choice between the categorization (...) and comparison processes. This article attempts to answer the question regarding which views are plausible by using cognitive modeling and computer simulation based on a semantic space model. In the simulation experiment, categorization and comparison processes are modeled in a semantic space constructed by latent semantic analysis. These two models receive word vectors for the constituent words of a metaphor and compute a vector for the metaphorical meaning. The resulting vectors can be evaluated according to the degree to which they mimic the human interpretation of the same metaphor; the maximum likelihood estimation determines which of the two models better explains the human interpretation. The result of the model selection is then predicted by three metaphor properties (i.e., vehicle conventionality, aptness, and interpretive diversity) to test the three views. The simulation experiment for Japanese metaphors demonstrates that both interpretive diversity and vehicle conventionality affect the choice between the two processes. On the other hand, it is found that metaphor aptness does not affect this choice. This result can be treated as computational evidence supporting the interpretive diversity and conventionality views. (shrink)

The pervasive use of distributional semantic models or word embeddings for both cognitive modeling and practical application is because of their remarkable ability to represent the meanings of words. However, relatively little effort has been made to explore what types of information are encoded in distributional word vectors. Knowing the internal knowledge embedded in word vectors is important for cognitive modeling using distributional semantic models. Therefore, in this paper, we attempt to identify the knowledge encoded in word vectors by conducting (...) a computational experiment using Binder et al.'s (2016) featural conceptual representations based on neurobiologically motivated attributes. In an experiment, these conceptual vectors are predicted from text‐based word vectors using a neural network and linear transformation, and prediction performance is compared among various types of information. The analysis demonstrates that abstract information is generally predicted more accurately by word vectors than perceptual and spatiotemporal information, and specifically, the prediction accuracy of cognitive and social information is higher. Emotional information is also found to be successfully predicted for abstract words. These results indicate that language can be a major source of knowledge about abstract attributes, and they support the recent view that emphasizes the importance of language for abstract concepts. Furthermore, we show that word vectors can capture some types of perceptual and spatiotemporal information about concrete concepts and some relevant word categories. This suggests that language statistics can encode more perceptual knowledge than often expected. (shrink)

The aim of this study is to assess the role of readers’ proficiency and of the base-word distributional properties on eye-movement behavior. Sixty-two typically developing children, attending 3rd, 4th, and 5th grade, were asked to read derived words in a sentence context. Target words were nouns derived from noun bases (e.g., umorista, ‘humorist’), which in Italian are shared by few derived words, and nouns derived from verb bases (e.g., punizione, ‘punishment’), which are shared by about 50 different inflected forms and (...) several derived words. Data showed that base and word frequency affected first-fixation duration for nouns derived from noun bases, but in an opposite way: Base frequency had a facilitative effect on first fixation, whereas word frequency exerted an inhibitory effect. These results were interpreted as a competition between early accessed base words (e.g., camino, chimney) and target words (e.g., caminetto, fireplace). For nouns derived from verb bases, an inhibitory base frequency effect but no word frequency effect were observed. These results suggest that syntactic context, calling for a noun in the target position, lead to an inhibitory effect when a verb base was detected, and made it difficult for readers to access the corresponding base+suffix combination (whole word) in the very early processing phases. Gaze duration was mainly affected by word frequency and length: for nouns derived from noun bases, this interaction was modulated by proficiency, as length effect was stronger for less proficient readers, while they were processing low-frequency words. For nouns derived from verb bases, though, all children, irrespective of their reading ability, showed sensitivity to the interaction within frequency of base+suffix combination (word frequency) and target length. Results of this study are consistent with those of other Italian studies that contrasted noun and verb processing, and confirm that distributional properties of morphemic constituents have a significant impact on the strategies used for processing morphologically complex words. (shrink)

Productivity is a central concept in the study of language and language acquisition. As a test case for exploring the notion of productivity, we focus on the noun slots of verb frames, such as __want__, __see__, and __get__. We develop a novel combination of measures designed to assess both the flexibility and creativity of use in these slots. We do so using a rigorously controlled sample of child speech and child directed speech from three English-speaking children between the ages of (...) 2–3 years and their caregivers. We find different levels of creativity and flexibility between the adult and child samples for some measures, for some slots, and for some developmental periods. We discuss these differences in the context of verb frame semantics, conventionality versus creativity and child errors, and draw some tentative conclusions regarding developmental changes in children's early grammatical representations. (shrink)

The extant literature on discourse comprehension distinguishes between two types of texts: narrative and expository. Narrative discourse tells readers a story by giving them an account of events; the narration informs and/or persuades the readership by using textual elements such as theme, plot, and characters. Expository discourse explains or informs the readership by using concepts and techniques such as definition, sequence, categorization, and cause-effect relations. The present study is based on two experiments. In Experiment 1, we compared the two discourse (...) types to examine if college students would be better at extracting the meanings of novel words from one of the two types of discourse structure than from the other. The findings indicated that participants were significantly better at inferring the meaning of novel words from narrative compared to expository discourse. In Experiment 2, we examined the number of situation models that a reader is required to mentally construct, as a possible characteristic that influences the difficulty of learning new word meaning within narrative discourse. Contrary to intuition, fewer novel words were learned in a single-situation, as opposed to a multi-situation model condition, suggesting that the additional inferencing needed to construct multiple models also promotes word learning. Results are discussed with respect to how the structure of written discourse can facilitate word learning in a reader’s native language. Implications for education and assessment are also discussed. (shrink)

The focus of this paper is two-fold. First, similarities generated from six semantic models were compared to human ratings of paragraph similarity on two datasets—23 World Entertainment News Network paragraphs and 50 ABC newswire paragraphs. Contrary to findings on smaller textual units such as word associations (Griffiths, Tenenbaum, & Steyvers, 2007), our results suggest that when single paragraphs are compared, simple nonreductive models (word overlap and vector space) can provide better similarity estimates than more complex models (LSA, Topic Model, SpNMF, (...) and CSM). Second, various methods of corpus creation were explored to facilitate the semantic models’ similarity estimates. Removing numeric and single characters, and also truncating document length improved performance. Automated construction of smaller Wikipedia-based corpora proved to be very effective, even improving upon the performance of corpora that had been chosen for the domain. Model performance was further improved by augmenting corpora with dataset paragraphs. (shrink)

We present statistical analyses of the large‐scale structure of 3 types of semantic networks: word associations, WordNet, and Roget's Thesaurus. We show that they have a small‐world structure, characterized by sparse connectivity, short average path lengths between words, and strong local clustering. In addition, the distributions of the number of connections follow power laws that indicate a scale‐free pattern of connectivity, with most nodes having relatively few connections joined together through a small number of hubs with many connections. These regularities (...) have also been found in certain other complex natural networks, such as the World Wide Web, but they are not consistent with many conventional models of semantic organization, based on inheritance hierarchies, arbitrarily structured networks, or high‐dimensional vector spaces. We propose that these structures reflect the mechanisms by which semantic networks grow. We describe a simple model for semantic growth, in which each new word or concept is connected to an existing network by differentiating the connectivity pattern of an existing node. This model generates appropriate small‐world statistics and power‐law connectivity distributions, and it also suggests one possible mechanistic basis for the effects of learning history variables (age of acquisition, usage frequency) on behavioral performance in semantic processing tasks. (shrink)

Context is an important construct in many domains of cognition, including learning, memory, and emotion. We used dynamical systems methods to demonstrate the episodic nature of experience by showing a natural separation between the scales over which within-context and between-context relationships operate. To do this, we represented an individual's emails extending over about 5 years in a high-dimensional semantic space and computed the dimensionalities of the subspaces occupied by these emails. Personal discourse has a two-scaled geometry with smaller within-context dimensionalities (...) than between-context dimensionalities. Prior studies have shown that reading experience and visual experience have a similar two-scaled structure. Furthermore, the recurrence plot of the emails revealed that experience is predictable and hierarchical, supporting the constructs of some influential theories of memory. The results demonstrate that experience is not scale-free and provide an important target for accounts of how experience shapes cognition. (shrink)

The paper focuses on the embodied view of cognition applied to language. First we discuss what we intend when we say that concepts are “embodied”. Then we briefly explain the notion of simulation, addressing also its neuro-physiological basis. In the main part of the paper we will focus on concepts mediated by language, presenting behavioral and neuro-physiological evidence of the action/perception systems activation during words and sentences comprehension.

Recent work has shown that many problematic aspects of metaphor use and comprehension can be resolved through an account that includes both relevance and perceptual simulation. It has also been shown that metaphors often imply stories, and that stories are often metaphorical. Previous research on narratives has focused primarily on stories that appear either in formal literature or in structured interviews; this essay focuses on stories that occur as an integral part of conversation. It extends recent work on metaphor comprehension (...) to show how use and comprehension of stories in natural, informal conversations can be usefully analyzed in terms of perceptual simulations. Conversely, it extends previous work on storytelling to reveal the metaphorical element in many stories, and to emphasize the social (bonding through shared enjoyment) as well as informative and persuasive functions of storytelling. (shrink)

Abstract Since their inception, distributional models of semantics have been criticized as inadequate cognitive theories of human semantic learning and representation. A principal challenge is that the representations derived by distributional models are purely symbolic and are not grounded in perception and action; this challenge has led many to favor feature-based models of semantic representation. We argue that the amount of perceptual and other semantic information that can be learned from purely distributional statistics has been underappreciated. We compare the representations (...) of three feature-based and nine distributional models using a semantic clustering task. Several distributional models demonstrated semantic clustering comparable with clustering-based on feature-based representations. Furthermore, when trained on child-directed speech, the same distributional models perform as well as sensorimotor-based feature representations of children’s lexical semantic knowledge. These results suggest that, to a large extent, information relevant for extracting semantic categories is redundantly coded in perceptual and linguistic experience. Detailed analyses of the semantic clusters of the feature-based and distributional models also reveal that the models make use of complementary cues to semantic organization from the two data streams. Rather than conceptualizing feature-based and distributional models as competing theories, we argue that future focus should be on understanding the cognitive mechanisms humans use to integrate the two sources. (shrink)

Computational techniques comparing co-occurrences of city names in texts allow the relative longitudes and latitudes of cities to be estimated algorithmically. However, these techniques have not been applied to estimate the provenance of artifacts with unknown origins. Here, we estimate the geographic origin of artifacts from the Indus Valley Civilization, applying methods commonly used in cognitive science to the Indus script. We show that these methods can accurately predict the relative locations of archeological sites on the basis of artifacts of (...) known provenance, and we further apply these techniques to determine the most probable excavation sites of four sealings of unknown provenance. These findings suggest that inscription statistics reflect historical interactions among locations in the Indus Valley region, and they illustrate how computational methods can help localize inscribed archeological artifacts of unknown origin. The success of this method offers opportunities for the cognitive sciences in general and for computational anthropology specifically. (shrink)

Empirical studies indicate that analogy consists of two main processes: retrieval and mapping. While current theories and models of analogy have revealed much about the mainly structural constraints that govern the mapping process, the similarities that underpin the correspondences between individual representational elements and drive retrieval are understood in less detail. In existing models symbol similarities are externally defined but neither empirically grounded nor theoretically justified. This paper introduces a new model (EMMA: the environmental model of analogy) which relies on (...) co‐occurrence information provided by LSA (Latent Semantic Analysis; Landauer & Dumais, 1997) to ground the relations between the symbolic elements aligned in analogy. LSA calculates a contextual distribution for each word encountered in a corpus by counting the frequency with which it co‐occurs with other words. This information is used to define a model that locates each word encountered in a high‐dimensional space, with relations between elements in this space representing contextual similarities between words. A series of simulation experiments demonstrate that the environmental approach to semantics embodied in LSA can produce appropriate patterns of analogical retrieval, but that this semantic measure is not sufficient to model analogical mapping. The implications of these findings, both for theories of representation in analogy research and more general theories of semantics in cognition, are explored. (shrink)

Two experiments investigated whether infants represent goal‐directed actions of others in a way that allows them to draw inferences to unobserved states of affairs (such as unseen goal states or occluded obstacles). We measured looking times to assess violation of infants' expectations upon perceiving either a change in the actions of computer‐animated figures or in the context of such actions. The first experiment tested whether infants would attribute a goal to an action that they had not seen completed. The second (...) experiment tested whether infants would infer from an observed action the presence of an occluded object that functions as an obstacle. The looking time patterns of 12‐month‐olds indicated that they were able to make both types of inferences, while 9‐month‐olds failed in both tasks. These results demonstrate that, by the end of the first year of life, infants use the principle of rational action not only for the interpretation and prediction of goal‐directed actions, but also for making productive inferences about unseen aspects of their context. We discuss the underlying mechanisms that may be involved in the developmental change from 9 to 12 months of age in the ability to infer hypothetical (unseen) states of affairs in teleological action representations. (shrink)

On the surface, bi- and multilingualism would seem to be an ideal context for exploring questions of typological proximity. The obvious intuition is that the more closely related two languages are, the easier it should be to implement the two languages in one mind. This is the starting point adopted here, but we immediately run into the difficulty that the overwhelming majority of cognitive, computational, and linguistic research on bi- and multilingualism exhibits a monolingual bias (i.e., where monolingual grammars are (...) used as the standard of comparison for outputs from bilingual grammars). The primary questions so far have focused on how bilinguals balance and switch between their two languages, but our perspective on typology leads us to consider the nature of bi- and multi-lingual systems as a whole. Following an initial proposal from Hsin ( 2014 ), we conjecture that bilingual grammars are neither isolated, nor (completely) conjoined with one another in the bilingual mind, but rather exist as integrated source grammars that are further mitigated by a common, combined grammar (Cook, 2016 ; Goldrick et al., 2016a, b ; Putnam and Klosinski, 2017 ). Here we conceive such a combined grammar in a parallel, distributed, and gradient architecture implemented in a shared vector-space model that employs compression through routinization and dimensionality reduction. We discuss the emergence of such representations and their function in the minds of bilinguals. This architecture aims to be consistent with empirical results on bilingual cognition and memory representations in computational cognitive architectures. (shrink)

The Abstract Conceptual Feature (ACF) framework predicts that word meaning is represented within a high‐dimensional semantic space bounded by weighted contributions of perceptual, affective, and encyclopedic information. The ACF, like latent semantic analysis, is amenable to distance metrics between any two words. We applied predictions of the ACF framework to abstract words using eyetracking via an adaptation of the classical “visual word paradigm” (VWP). Healthy adults (n = 20) selected the lexical item most related to a probe word in a (...) 4‐item written word array comprising the target and three distractors. The relation between the probe and each of the four words was determined using the semantic distance metrics derived from ACF ratings. Eye movement data indicated that the word that was most semantically related to the probe received more and longer fixations relative to distractors. Importantly, in sets where participants did not provide an overt behavioral response, the fixation rates were nonetheless significantly higher for targets than distractors, closely resembling trials where an expected response was given. Furthermore, ACF ratings which are based on individual words predicted eye fixation metrics of probe‐target similarity at least as well as latent semantic analysis ratings which are based on word co‐occurrence. The results provide further validation of Euclidean distance metrics derived from ACF ratings as a measure of one facet of the semantic relatedness of abstract words and suggest that they represent a reasonable approximation of the organization of abstract conceptual space. The data are also compatible with the broad notion that multiple sources of information (not restricted to sensorimotor and emotion information) shape the organization of abstract concepts. While the adapted “VWP” is potentially a more metacognitive task than the classical visual world paradigm, we argue that it offers potential utility for studying abstract word comprehension. (shrink)