Integrating language model embeddings into the ACT-R cognitive modeling framework

Iconic words are characterized by a sense of resemblance between their form and their meaning. The most common examples are onomatopoeias such as “cock-a-doodle-doo” or “woof-woof,” but recent research shows that speakers of various languages perceive a relationship between sound and meaning in various words of their language. This is the case for English words such as crunchy, zigzag or wiggle, whose sonority is perceived as mapping onto their meaning (Winter et al., 2023). Recently, psycholinguistic research has investigated whether iconicity influence language processing. Sidhu, Vigliocco and Pexman (2020) ran a lexical decision task with words varying in their iconicity level. Their results show that the more iconic a word is, the quicker participants react to it, thus suggesting that iconicity may have a facilitatory effect in word recognition. We plan to replicate their study with a sample of Brazilian Portuguese speakers and Brazilian Portuguese words to test their main claim that reaction times in a classic lexical decision task could be influenced by the stimuli’s level of iconicity in a neurotypical population of adults. Our analysis plan closely follows the original study. A successful replication should show that participants react faster to words with higher iconicity ratings.

Purpose: To evaluate brain activation areas during the processing of languages in multilingual volunteers by functional MRI and to examine the differences between the mother and foreign languages. Materials and Methods: Nine multilingual (Korean, French, and English speaking) Korean individuals were enrolled in this study. Functional images were acquired during a lexical decision task (LDT) and picture naming task (PNT) in each of the Korean, French and English languages. The areas activated were analyzed topographically in each language and task, and compared between languages. Results: Activation was noted in Broca’s area, supramarginal gyrus, fusiform gyrus during the LDT. During the PNT, activation was noted in Broca’s area, left prefrontal area, cerebellum, right extrastriated cortex. While Broca’s area activation was observed for all languages during LDT, there was more activation in Broca’s area and additional activation in the right prefrontal area with foreign languages. During the PNT, there was more activation in the left prefrontal area with foreign languages. Conclusion: Broca’s area, which is known as a major language region, was activated by all languages and tasks. The brain activation areas were largely overlapping with the mother and foreign languages. However, there were wider areas of activation and additional different activation areas with foreign languages. These results suggest more cerebral effort during foreign language processing.

This study introduces an artificial intelligence- based analytical model created to identify patterns, anomalies, and leads in long-standing open missing persons cases in India. By utilizing publicly available demographic, geographic, and temporal data from the Missing People Dataset (Kaggle), NCRB data, and district reports, the system employs several analytical layers: Pattern DNA clustering to reveal hidden demographic trends, Geo-Spatial Drift Analysis to predict potential movement areas using spatial attributes and logistic regression, and a Semantic Case Matching module that uses BERT embedding to connect cases with similar narratives for cold case support. The pipeline also features a risk prediction module based on time-series models like Prophet to forecast upcoming peaks in missing persons cases by state and time intervals. Extensive data preprocessing, exploratory data analysis, feature creation, and clustering were performed using Python. The resulting geo-risk heatmaps, similarity scores, and trend forecasts offer actionable insights to aid investigators in resource allocation, hotspot identification, and case prioritization. This framework demonstrates how AI can improve investigative efforts and highlights the potential for future integration with law enforcement databases and humanitarian programs, further emphasizing the value of technology-driven solutions for addressing key social issues. Keywords: Missing Person Analysis, Artificial Intelligence, Pattern Clustering, Geo-Spatial Analysis, Semantic Case Matching, Natural Language Processing, BERT embedding, Time Series Forecasting.

The contents and structure of semantic memory have been the focus of much recent research, with major advances in the development of distributional models, which use word co‐occurrence information as a window into the semantics of language. In parallel, connectionist modeling has extended our knowledge of the processes engaged in semantic activation. However, these two lines of investigation have rarely been brought together. Here, we describe a processing model based on distributional semantics in which activation spreads throughout a semantic network, as dictated by the patterns of semantic similarity between words. We show that the activation profile of the network, measured at various time points, can successfully account for response times in lexical and semantic decision tasks, as well as for subjective concreteness and imageability ratings. We also show that the dynamics of the network is predictive of performance in relational semantic tasks, such as similarity/relatedness rating. Our results indicate that bringing together distributional semantic networks and spreading of activation provides a good fit to both automatic lexical processing (as indexed by lexical and semantic decisions) as well as more deliberate processing (as indexed by ratings), above and beyond what has been reported for previous models that take into account only similarity resulting from network structure.

It is generally accepted that a word's emotional valence (i.e., whether a word is perceived as positive, negative, or neutral) influences how it is accessed and remembered. There is also evidence that the affective content of some words is represented in nonarbitrary sound-meaning associations (i.e., emotional sound symbolism). We investigated whether more extensive statistical relationships exist between the surface form properties of English words and ratings of their emotional valence, that is, form typicality. We found significant form typicality for both valence and extremity of valence (the absolute distance from the midpoint of the rating scale, regardless of polarity). Next, using behavioral megastudy data sets, we show that measures of emotional form typicality are significant predictors of lexical access during written and auditory lexical decision and reading aloud tasks in addition to recognition memory performance. These findings show nonarbitrary form-valence mappings in English are accessed automatically during language and verbal memory processing. We discuss how these findings might be incorporated into theoretical accounts that implement Bayesian statistical inference. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Vocabulary knowledge is central to a speaker's command of their language. In previous research, greater vocabulary knowledge has been associated with advantages in language processing. In this study, we examined the relationship between individual differences in vocabulary and language processing performance more closely by (i) using a battery of vocabulary tests instead of just one test, and (ii) testing not only university students (Experiment 1) but young adults from a broader range of educational backgrounds (Experiment 2). Five vocabulary tests were developed, including multiple-choice and open antonym and synonym tests and a definition test, and administered together with two established measures of vocabulary. Language processing performance was measured using a lexical decision task. In Experiment 1, vocabulary and word frequency were found to predict word recognition speed while we did not observe an interaction between the effects. In Experiment 2, word recognition performance was predicted by word frequency and the interaction between word frequency and vocabulary, with high-vocabulary individuals showing smaller frequency effects. While overall the individual vocabulary tests were correlated and showed similar relationships with language processing as compared to a composite measure of all tests, they appeared to share less variance in Experiment 2 than in Experiment 1. Implications of our findings concerning the assessment of vocabulary size in individual differences studies and the investigation of individuals from more varied backgrounds are discussed.

A meta-analysis of 26 studies indicated that automatic semantic priming can occur without association. Priming did not vary substantially with differences in variables that affect automatic versus strategic processing, such as time spent processing the prime and target, relationship proportion, and task (except that average effects were smaller in the naming task). Although category coordinates were investigated in the majority of studies, synonyms, antonyms, and script relations also demonstrated priming; functional relations showed greater priming, and essential and perceptual relations showed less. The average effect size for semantic priming was smaller than that for associative priming, suggesting that there is an "associative boost" from adding an associative relationship to a semantic one. The implications of these findings for the modularity thesis and for models of semantic priming are discussed.

In a recent meta-analysis, Lucas (2000) concluded that there is strong evidence of an overall pure semantic priming effect but no evidence of priming based purely on association. In the present review, I critically examine the individual studies claiming evidence of featural and associative relations in semantic memory. The most important conclusion is that automatic priming appears to be due to both association strength and feature overlap. Mediated associates provide the strongest evidence of automatic associative priming, whereas functional associates, synonyms, and antonyms instead support priming based on feature overlap. In contrast, automatic priming does not occur for category coordinates or perceptually similar items, at least when presented in the visual modality. The status of other relations, such as collocates, episodic relatives, and script relations, is unclear and requires further experimentation. Implications for current models of semantic representation and priming are discussed.

Normative measures of verbal material are fundamental in psycholinguistic and cognitive research for the control of confounding in experimental procedures and for achieving a better comprehension of our conceptual system. Traditionally, normative studies have focused on classical psycholinguistic variables, such as concreteness and imageability. Recent works have shifted researchers' focus to perceptual strength, in which items are rated separately for each of the five senses. We present a resource that includes perceptual norms for 1,121 Italian words extracted from the Italian version of ANEW. Norms were collected from 57 native speakers. For each word, the participants provided perceptual-strength ratings for each of the five perceptual modalities. The perceptual norms performance in predicting human behavior was tested in two novel experiments, a lexical decision task and a naming task. Concreteness, imageability, and different composite variables representing perceptual-strength scores were considered as competing predictors in a series of linear regressions, evaluating the goodness of fit of each model. For both tasks, the model with imageability as the only predictor was found to be the best-fitting model according to the Akaike information criterion, whereas the model with the separately considered five modalities better described data according to the explained variance. These results differ from the ones previously reported for English, in which maximum perceptual strength emerged as the best predictor of behavior. We investigated this discrepancy by comparing Italian and English data for the same set of translated items, thus confirming a genuine cross-linguistic effect. We thus confirmed that perceptual experience influences linguistic processing, even though evaluations from different languages are needed to generalize this claim.

Language is one of the most significant and immense aspects of human evolution. Not surprisingly, it has been a research topic of scientists working in many disciplines. Numerous questions have been raised from this important research: Why is language so quickly acquired by children living in very different cultures with very different languages? How does the human brain produce and comprehend language? How is second-language learning influenced by the native language? Is the neural network that mediates language processing universal across languages? For years, scientists have attempted to answer these questions. With the advent of new technology, such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), the past decade has seen an extremely rapid expansion of research on the aforementioned issues. Such technical advances have given researchers great power to visualize the activity of the intact human brain that is associated with language processing and language learning, and have consequently made significant contributions to the vast amount of discoveries in brain research. Whilst imaging technologies are improving every year, because language processing demands higher-order cognition, its neuro-cognitive mechanisms are very complex and much research remains to be done. Cross-linguistic studies provide a powerful contrast to elucidate the universality and specificity of the neural basis of language. Although the direct subtraction of the processing of the native language in one population (Language A) from the processing in another population (Language B) might seem like an interesting comparison, cognitive and psycholinguistic research has established that even at the lexical level, cognitive processing involves dozens of variables, such as visual and auditory complexity, spelling–sound regularity, word frequency, word familiarity, and age of acquisition. Thus, in brain mapping research, direct subtraction of two different populations may include a confounding of linguistic variables in two different languages, and these variables are often un-confoundable. For example, it might be possible to find seemingly identical words in two different languages (“banana” is banana in English and banana in Italian). However, the processing might be quite different due to many factors such as syllable structure (banana has a much more typical structure in Italian than in English), how common the spelling–sound relationships are (because far fewer spelling–sound relationships exist in Italian than in English, they are much more frequent), how consistent the spelling–sound relationships are (graphemes in English often map onto many different sounds but in Italian they typically do not), and so on. A better understanding of the generality and specificity of neural mechanisms associated with language should build on knowledge and conclusions from research on individual languages, rather than from a direct subtraction approach. A number of studies in this special issue show how cross-language research can lead to important insights. This special issue emerged from a research symposium in April 2002 at the University of Hong Kong, in which we as co-organizers invited researchers who were very active in the fields of cognitive neuroscience, cognitive psychology, and linguistics to present their functional brain imaging findings. These pages feature contrastive studies of Chinese and English, two languages that differ markedly in almost every aspect: phonology, orthography, semantics, and syntax. We hope that our readers will learn from these new findings, and will be inspired to further such research with their own particular knowledge and skills.

Individual Differences in Language Acquisition and Processing

The effect of relatedness proportion (the percentage of related words in a list) on associative priming was investigated in two masked priming experiments. In Experiment 1, subjects were randomly assigned to either a high (80%) or low (20%) relatedness proportion condition and performed a lexical decision task. Semantically related and unrelated primes were briefly flashed (50 ms) before the targets and were preceded by a mask that prevented their overt identification. In Experiment 2, subjects were tested in both relatedness proportion conditions in a categorization task; the N400, an electrophysiological index of lexical and semantic priming, was measured in trials that did not require a button press. Behavioral and electrophysiological priming effects were observed in both high and low relatedness proportion conditions. The observed priming effects were not modulated by relatedness proportion. The results are discussed in terms of current theories of associative priming.

Size is an important visuo-spatial characteristic of the physical world. In language processing, previous research has demonstrated a processing advantage for words denoting semantically “big” (e.g., jungle) versus “small” (e.g., needle) concrete objects. We investigated whether semantic size plays a role in the recognition of words expressing abstract concepts (e.g., truth). Semantically “big” and “small” concrete and abstract words were presented in a lexical decision task. Responses to “big” words, regardless of their concreteness, were faster than those to “small” words. Critically, we explored the relationship between semantic size and affective characteristics of words as well as their influence on lexical access. Although a word’s semantic size was correlated with its emotional arousal, the temporal locus of arousal effects may depend on the level of concreteness. That is, arousal seemed to have an earlier (lexical) effect on abstract words, but a later (post-lexical) effect on concrete words. Our findings provide novel insights into the semantic representations of size in abstract concepts and highlight that affective attributes of words may not always index lexical access.

In their review, Murgiano, Motamedi, and Vigliocco (2020) lay out a new perspective in which they argue that language should be understood as a situated phenomenon. This perspective has implications for the study of written language, which is fundamentally an un-situated phenomenon. We consider the implications of this perspective for iconicity as it appears in written language. We argue that typical methods for studying word processing (e.g., the lexical decision task) may be bound to underestimate the relevance of iconicity for language. In addition, the typical approach of collecting ratings of individual words on a lexical-semantic dimension may not be well suited to quantifying iconicity. Nevertheless, we believe the field should continue to explore effects of iconicity in language processing, and we discuss some potential ways to adjust traditional word processing tasks.

The term schizotypy refers to a group of stable personality traits with attributes similar to symptoms of schizophrenia, usually classified in terms of positive, negative or cognitive disorganization symptoms. The observation of increased spreading of semantic activation in individuals with schizotypal traits has led to the hypothesis that thought disorder, one of the characteristics of cognitive disorganization, stems from semantic disturbances. Nevertheless, it is still not clear under which specific circumstances (i.e., automatic or controlled processing, direct or indirect semantic relation) schizotypy affects semantic priming or whether it does affect it at all. We conducted two semantic priming studies with volunteers varying in schizotypy, one with directly related prime-target pairs and another with indirectly related pairs. Our participants completed a lexical decision task with related and unrelated pairs presented at short (250 ms) and long (750 ms) stimulus onset asynchronies (SOAs). Then, they responded to the brief versions of the Schizotypal Personality Questionnaire and the Oxford-Liverpool Inventory of Feelings and Experiences, both of which include measures of cognitive disorganization. Bayesian mixed-effects models indicated expected effects of SOA and semantic relatedness, as well as an interaction between relatedness and directness (greater priming effects for directly related pairs). Even though our analyses demonstrated good sensitivity, we observed no influence of cognitive disorganization over semantic priming. Our study provides no compelling evidence that schizotypal symptoms, specifically those associated with the cognitive disorganization dimension, are rooted in an increased spreading of semantic activation in priming tasks.