1407 norm sets
Psycholinguistic research shows that numerous variables influence language comprehension and production.<b> </b>Despite the importance of norming words for neutralizing lexical confounding factors, facilitating cross-linguistic comparative research, and ensuring the replicability of psychological experiments, such lexical resources remain disproportionately scarce for minority and under-resourced languages. This dataset presents NORGAL, a comprehensive database comprising normative ratings for emotional valence, arousal, concreteness, imageability and subjective age of acquisition (AoA) of 1,585 Galician words. Highly frequent words – nouns and adjectives in particular – were extracted from Reference Corpus of Current Galician (CORGA) and normed on a 9-point Likert scale (except for AoA) by bilingual Galician-Spanish students from two University of Vigo campuses. This study also presents comparative analyses against established Spanish normative databases, ensuring strict methodological control and confirming the consistency of our stimulus ratings across studies. Finally, primary limitations associated with normative research in acquiring psycholinguistic indices in this specific context are explained, while paving the way for future avenues of application and expansion. In sum, NORGAL can serve as a valuable foundation for advancing experimental and psycholinguistic research in this language.
Résumé Cet article présente des normes d’imageabilité (ou valeurs d’imagerie) pour un ensemble de 1493 mots. Des analyses statistiques réalisées sur ces normes révèlent une fidélité élevée. Les scores d’imageabilité se révèlent par ailleurs assez modestement corrélés avec d’autres variables psycholinguistiques (par ex., fréquences lexicales, âge d’acquisition). Des analyses restreintes à un sous-échantillon de mots en français, ainsi que d’autres sur des mots normés pour l’anglais, révèlent que le nombre de traits sémantiques est modérément positivement corrélé aux scores d’imageabilité, contrairement à l’hypothèse selon laquelle la richesse sémantique est adéquatement indexée par l’imageabilité.
This normative dataset provides perceptual strength ratings for 5,500 Spanish words across five sensory modalities: touch, hearing, sight, smell, and taste. The word set was originally compiled from Spanish lexical databases and used in Díez-Álamo et al. (2019), ensuring broad coverage of psycholinguistic indices. A total of 671 participants rated approximately 200 words each, indicating the extent to which each sensory modality is involved when experiencing the concept (or property, in the case of adjectives), using a scale from 0 (not at all) to 5 (greatly). The procedure follows the methodology established by Lynott and Connell (2009, 2013). The dataset includes raw participant-level data, with fields for participant ID, task number, age, gender, word familiarity, and modality-specific ratings. This resource is valuable for research on conceptual representation, semantic processing, and multimodal lexical analysis in Spanish.
Over the last 40 years, object recognition studies have moved from using simple line drawings, to more detailed illustrations, to more ecologically valid photographic representations. Researchers now have access to various stimuli sets, however, existing sets lack the ability to independently manipulate item format, as the concepts depicted are unique to the set they derive from. To enable such comparisons, Rossion and Pourtois (2004) revisited Snodgrass and Vanderwart's (1980) line drawings and digitally re-drew the objects, adding texture and shading. In the current study, we took this further and created a set of stimuli that showcase the same objects in photographic form. We selected six photographs of each object (three color/three grayscale) and collected normative data and RTs. Naming accuracy and agreement was high for all photographs and appeared to steadily increase with format distinctiveness. In contrast to previous data patterns for drawings, naming agreement (H values) did not differ between grey and color photographs, nor did familiarity ratings. However, grey photographs received significantly lower mental imagery agreement and visual complexity scores than color photographs. This suggests that, in comparison to drawings, the ecological nature of photographs may facilitate deeper critical evaluation of whether they offer a good match to a mental representation. Color may therefore play a more vital role in photographs than in drawings, aiding participants in judging the match with their mental representation. This new photographic stimulus set and corresponding normative data provide valuable materials for a wide range of experimental studies of object recognition.
The field of false memories has been widely studied in cognitive psychology through the DRM paradigm (Deese, 1959;Roediger & McDermott, 1995), an experimental task used to induce false memories from materials that are conceptually and semantically related. This paradigm involves presenting lists of words that are semantically related to a non-presented critical word, with a subsequent memory test showing high levels of false recall and false recognition of that non-studied critical word. For example, after studying a list containing words, such as "butter", "food" and "sandwich", it is likely that in a subsequent free recall or recognition test, the word "bread" will be mistakenly identified as studied.False memory is generated, in part, by the relationship between the list words and the critical word (Gallo, 2010;Roediger & Gallo, 2016). This phenomenon has been explained by two theories: the fuzzy-trace theory (FTT) (Brainerd & Reyna, 1998), and the activationmonitoring framework (AMF) (Roediger et al., 2001). Both theories agree that, understanding the production of false memories requires considering two complementary processes: an error inflation process, identified as gist encoding in FTT and as activation in AMF; and an error editing process, identified as monitoring in AMF and as recollection rejection in FTT (Brainerd & Reyna, 2002).According to FTT (Brainerd & Reyna, 1998), "gist" refers to the general theme extracted from studied material. When a list of words related to a non-presented critical word is studied, both literal and semantic information are encoded. In a subsequent memory test, these literal and semantic memory traces operate simultaneously, providing information about the items. The retrieval of semantic information may lead to considering the critical word as having been studied due to its similarity to the presented words. However, the retrieval of literal information about the studied associates can counteract this effect by providing evidence that the critical word was not studied, a process known as recollection rejection (Brainerd & Reyna, 2002).According to the AMF (Roediger et al., 2001), two processes work together to produce false memory: activation and monitoring. Studying a list of words can trigger activation that spreads through the lexical-semantic system, creating implicit associations between interconnected words. This activation is moderated by a subsequent monitoring process that helps distinguishing between correct recall (studied words) and false memories (non-studied words).Even when considering different perspectives, both FTT and AMF agree that presenting a list of associates words activates a non-presented critical word, leading to error inflation. If the error inflation process is not accompanied by its corresponding error editing process, or if this process fails, false recall or false recognition may occur (Arndt & Gould, 2006). Therefore, studying the strategies used to avoid false memories is crucial for understanding the mechanisms underlying their formation.Theme identifiability of a list is one of the essential factors involved in this error editing process (Carneiro et al., 2009). In their normative study in Portuguese, they provided theme identifiability norms for 40 DRM associative lists selected from Albuquerque's (2005) study. Participants were presented with the lists and asked to generate a word that best described the general theme of each list. To study the effect of this factor on the production of false memories, they selected those lists with the highest and lowest levels of theme identifiability. The results showed that lists with high identifiability of the critical word as the theme produced lower levels of false recall and false recognition compared to lists where the critical word was not as easily identifiable. According to Carneiro et al. (2012), this outcome was attributed to an error editing strategy called "Identify to reject," which involves several stages: detecting that all the words in a list are related to a common theme, identifying the word that best describes the theme but is not present in the list, keeping it in mind to avoid recalling it in the future, and consequently, reducing false memories. This pattern has been observed in other studies using lists with an associative structure (Beato et al., 2023;Carneiro & Fernandez, 2013;Carneiro et al., 2012). Additionally, there are other normative studies that provide theme identifiability indices for associative lists in Spanish (Beato & Cadavid, 2016) and in English (Neuschatz et al., 2003).Most studies on false memories and error editing mechanisms using the DRM paradigm employ lists with an associative structure. However, ad hoc categorical relationships are less explored in the literature. Ad hoc categories are spontaneously constructed to achieve a specific goal in a given context, and their elements can come from different taxonomic categories (e.g., "Things that can fall on your head") (Barsalou, 1983). Both common and ad hoc categories can lead to similar memory distortions. While false memories are typically more pronounced for common categories, they are still robust for ad hoc categories (Soro et al., 2017).The mechanisms for avoiding memory distortions in such lists are not well understood. Ad hoc categories provide a valuable tool for studying situated concept representations, which are characterized by their flexibility and dynamism (Barsalou, 2005). Their use allows researchers to explore how individuals organize and retrieve information when categories are not predefined but emerge from context. This is particularly relevant for understanding how memory adapts to new information and situations, adding depth to theoretical debates on flexible concept representation.In the study with associative lists by Carneiro et al. (2009), the percentages of identification for critical words ranged from 1% to 77%. However, Soro et al. (2017) indicate that, unlike associative lists, in ad hoc categorical structured lists, theme identification typically refers to identifying the category label rather than the critical word itself. They used two criteria: exact identifiability, where participants identified the original theme of the lists (e.g., "Materials that cover the ground" for the category "Things that can be walked upon"), and comprehensive identifiability, where participants identified a theme that could include the critical word (e.g., a label that includes the critical word "grass" for "Things that can be walked upon"). Both criteria are important for describing our findings.To our knowledge, no previous studies have addressed false memories or theme identification with ad hoc categories in Spanish. Therefore, the aim of this research was to obtain theme identifiability indices for 70 lists that maintained ad hoc categorical relationships with a non-presented critical word using the DRM paradigm. These lists were created based on a normative study of ad hoc categories conducted in Spanish, which, to our knowledge, is the first of its kind in this language. Additionally, this research aims to lay the groundwork for studying the underlying mechanisms of error editing processes in lists with ad hoc categorical relationships in Spanish.In future research, these data may help in understanding the role of theme identifiability in the formation of false memories. Moreover, having these indices will enable more accurate predictions and better control over the experimental materials.A total of 188 students from the Psychology degree program at the University of La Laguna participated. All participants were native Spanish speakers (146 women, 42 men; mean age= 20.14, SD = 2.94).The material consisted of 70 ad hoc categorical lists, each containing 10 words. Both the critical words and their corresponding associates were extracted from a normative study conducted to collect data on ad hoc categories in Spanish (Alonso et al., in preparation;Benítez, Alonso, Fernandez, & Díez, 2022). Ad hoc categories were selected from various normative studies in English and Portuguese and then translated into Spanish (Barsalou, 1982(Barsalou,, 1983(Barsalou,, 1985;;Hough & Pierce, 1989;Soro & Ferreira, 2017;Vallée-Tourangeau et al., 1998;van Overschelde et al., 2004). The general procedure used in the Spanish normative study was similar to that used by Battig & Montague (1969), with the exception that in the present study both the presentation of the material and the collection for responses were done by computer (see van Overschelde et al., 2004). The participants were instructed to generate as many exemplars as possible for each category within one minute. This study, currently in preparation, will provide indices of frequency, rank, and lexical availability for the exemplars of each category.The lists were constructed based on the frequencies obtained from the normative study. Critical words were selected as those with the highest frequency within each category, while their associates were the next most frequent words. Care was taken to ensure that the critical words did not appear in more than one list and that no associate was repeated across lists. The selected critical words were primarily nouns (with only 4 being verbs and 1 adjective), ranging from 1 to 5 syllables, with a mean frequency of occurrence in Spanish of 60.86 per million (Alonso et al., 2011).The 70 lists were divided into 5 blocks: 4 blocks containing 15 lists each and 1 block with 10 lists. For the theme identification test, a booklet was prepared with several pages. The first page collected participant information (name, age, gender, and degree). The second page included practice examples to familiarize participants with the task. The remaining pages were dedicated to the experimental lists. Each page displayed the list number and had three blank spaces for participants to write the word or words they believed identified the theme of the list (up to three), along with a Likert scale to indicate their confidence in whether the word given in the first position represented the list's theme. Procedure Data collection took place in November 2023 during group sessions, each consisting of approximately 35 participants and lasting around 30 minutes. Each group studied 15 lists, except for one group that studied only 10 lists. The order of list presentation within each group was randomly determined.Participants began by completing the demographic information on the first page of the booklet and then received instructions similar to those used in previous studies (Carneiro et al., 2009;Neuschatz et al., 2003). They were shown a series of lists in a PowerPoint presentation, with one word displayed every 2 seconds. Before the start of the experimental session, and to ensure participants fully understood the task, two practice trials were conducted. These trials involved the same task as the main session: following the presentation of each list, participants had 50 seconds to generate up to three words that they believed best described the theme of the list. They also provided a confidence judgment on how certain they were that the first word given represented the list's theme, using a Likert scale ranging from 1 ("not very confident") to 5 ("very confident").Before each list, a message appeared on the screen indicating the list number to be presented. The process of presenting the list and identifying the theme was repeated until the experimental session was complete.The spreadsheet file accompanying this report (Theme_Identifiability_Ad_hoc_Spanish.xls) consists of five sheets. The first sheet, "Theme identifiability", contains the raw data for each participant. It includes the 70 critical words and their respective lists of 10 ad hoc associates. The first column shows the participant number, the second column identifies the list, the third column specifies the type of relationship of the lists (ad hoc), and the fourth column indicates the type of word (studied vs. critical). The fifth column contains the words themselves, while the sixth column provides the English translations of the critical and studied words. Adjacent columns include all responses provided by participants in the first, second, and third positions, as well as the confidence judgements related to the first response. Additionally, intrusions are noted-i.e., words from a study list that a participant mistakenly identified as the theme of that list, and therefore are not considered valid responses.The second sheet, "First word", summarizes the total count of responses given in the first position for each list. It includes all the words generated by participants as the theme in the first position, associated with their respective critical word and list number. Additionally, it provides the English translation of the critical word, the total number of participants who responded to each list, the number and percentage of participants who identified a word as the theme, and the mean confidence judgement for each first-word response. Intrusions are also noted, including quantity and incorrectly identified words.The third and fourth sheets, "Second word" and "Third word", respectively, are dedicated to responses given in the second and third positions. The layout is identical to the previous sheets, but it does not include the column for mean confidence judgement. The fifth sheet, "Summary," presents the final summary, showing the most frequently identified theme for the first, second, and third positions for each list.Responses recorded in the database were maintained in their original format, with corrections made only for spelling errors. Singular/plural and masculine/feminine forms of words were counted separately.Table 1, available as supplementary material, shows the 70 critical words with their corresponding list identifier and English translation, the number of participants who responded to each list, the number of different themes given as the first response, the percentage of participants who indicated the critical word (comprehensive identification) as the theme of the list in the first position, and the mean confidence rating for the critical word as the first response. Additionally, it provides the percentages of participants who identified the ad hoc category label (exact identification) in the first position, as well as the mean confidence rating for these first-position responses.Theme identifiability is a crucial factor in the error editing process and thus influences the formation of false memories (Carneiro et al., 2009;Soro et al., 2017). However, the study of false memories in ad hoc categorical lists and the factors contributing to their formation remain unexplored in Spanish. Therefore, this study aimed to provide theme identifiability indices for 70 ad hoc lists within the framework of the DRM paradigm in Spanish.When examining identifiability levels using the same approach as in studies with associative lists (Carneiro et al., 2009), where the critical word is considered as the theme, the levels of identifiability are relatively low. Regarding comprehensive identifiability, the critical word was identified as a theme in the first position in only 10 out of the 70 lists, with identification percentages ranging from 2.1% to 17.2%. On average, the critical word was identified as the theme 0.98% (SD=3.11) of the time in the first position across all 70 lists. When considering only where the critical word was identified, this average increased to 6.88% (SD=5.39), with a mean confidence rating of 3.57 (SD=1.29). Given the ad hoc category exemplars can come from different categories, their membership is not immediately apparent without context, making the activation of the critical word challenging.In contrast, exact identifiability showed higher levels of identification. The theme of 59 lists (ranging from 2.1% to 87.5%) was identified in the first position, demonstrating a significant increase in identification compared to when only the critical word was considered. Participants often generated words that, while not precisely matching the category labels, were related to them, suggesting some thematic processing even if not explicitly expressed. On average, the theme was identified 22.22% (SD=23.75) of the time in the first position across all 70 lists. When considering only the lists where the theme was identified, this average rose to 26.36% (SD=23.67), with a mean confidence rating of 3.98 (SD=0.70) for the first response.According to Carneiro et al. (2009), in associative lists, the level of identification of the critical word as the theme is inversely related to the occurrence of false memories. Identifying the critical word as the theme triggers an error editing process that mitigates false memories in subsequent memory tests. However, this assumption may not hold true for categorical lists, particularly ad hoc categorical lists, where theme identifiability might not exert the same effect on the error editing process. Soro et al. (2017) suggested that for false memories to occur with ad hoc categories, a positive relationship with theme identifiability might be necessary due to the inherent variability among exemplars. Their study found no correlation between false recognition and theme identifiability. Instead, false recognition was influenced more by individual factors such as experience or creative thinking. Some participants generated more associations between exemplars, leading to an increased likelihood of errors in a subsequent recognition test.Our results suggest that context plays a fundamental role in theme identification within ad hoc categories, particularly concerning comprehensive identifiability. Therefore, considering the findings of Soro et al. (2017), the participant's ability to integrate exemplars into an appropriate context may significantly influence the activation of critical words. This aligns with the notion that memory is a dynamic process shaped by contextual factors (Barsalou, 2005).The present study provides valuable data on theme identifiability for a wide range of ad hoc categorical lists, underscoring the need for a different approach when studying error editing processes in this context. Our findings reveal a crucial aspect of ad hoc categories: while traditional associative lists tend to exhibit reduced false memories due to clearer theme identifiability, the inherent variability and contextual emergence of ad hoc categories introduce complexities in memory retrieval that merit further investigation. This area represents a promising opportunity for advancing our understanding of false memories.
Semantic representations arise from a distillation of multiple sources of information, including sensory, motor, affective, interoceptive, linguistic and cognitive experience. Experience of reward is a highly salient aspect of many human activities, and yet its contribution to semantic processing is not well understood. To address this, the present study took a psycholinguistic approach to measuring and evaluating associations with reward as a facet of word meaning. Behavioural and neurophysiological data suggest that reward processing involves multiple stages and mechanisms. For instance, systems associated with the experience and anticipation of pleasure in response to a reward appear distinct from motivational processes that underlie the pursuit of a stimulus. We sought to collect a novel set of word ratings that capture the full extent of reward-related experience. Initial explorations revealed that reward/pleasure ratings are highly correlated with existing norms of emotional valence. Ratings of association with motivation, however, were only moderately correlated with valence, suggesting they capture distinct semantic information. We therefore conducted a preregistered large-scale study to obtain motivation ratings for 8,601 words. Our analyses suggest these ratings capture aspects of word meaning which are distinct from other semantic dimensions, such as concreteness and valence. Moreover, they explain unique variance in participant performance on lexical, semantic, and recognition memory tasks. We combined motivation and emotional valence ratings to provide a composite measure that might approximate a more general 'reward' construct. However, this did not explain additional variance compared to the individual variables. We discuss the implications of these results for neurocognitive theories of semantics.
Affective word norms are essential for stimulus control in affective science, experimental psychology, and psycholinguistics, yet Urdu remains underrepresented in established lexical norm resources. The present study developed Urdu valence, arousal, and dominance (VAD) norms for a culturally adapted set of affective words derived from a Pakistan-grounded English VAD lexicon, which was then translated and culturally adapted and normed in Urdu. Using a cross-sectional, laboratory-based design, 240 Pakistani university students and young adults rated one of four stimulus lists. The final dataset comprised 320 Urdu words, with each word receiving 60 valid ratings. For each item, word-level means, standard deviations, valid ns, standard errors, and 95% confidence intervals were computed for valence, arousal, and dominance. The resulting lexicon showed broad coverage across affective space, high internal stability of word-level means across repeated random rater splits, and coherent dimensional associations among valence, arousal, and dominance. Category-based analyses further indicated that the item-construction strategy successfully distributed the final word set across distinct affective regions. The primary contribution of the study is an Urdu VAD norms resource that provides an empirically grounded basis for selecting, matching, and interpreting Urdu verbal stimuli in affective and psycholinguistic research. More broadly, the findings reinforce the importance of establishing affective properties in the target language rather than inferring them from translation alone.
This project hosts Mime Initiation Latency (MIL) norms for 189 object photographs, developed to support research on motor affordances and embodied cognition. The dataset provides behavioural latency measures indexing the time required for participants to initiate a pantomimed object-use action in response to colour photographs drawn from the Bank of Standardized Stimuli (BOSS; Brodeur et al., 2010, 2014). MIL constitutes a behavioural motor norm intended to complement existing subjective affordance measures (e.g., manipulability, graspability, mimability, body–object interaction). The norms were collected from Thai undergraduate participants using a controlled reaction-time paradigm implemented in E-Prime. After data screening and trimming procedures, normative statistics (means, standard deviations, distributional indices) are reported for 189 manipulable object photographs. Construct validity was evaluated through correlational and regression analyses with established motor and psycholinguistic norms. MIL was moderately to strongly associated with conceptually related motor variables and weakly or not associated with conceptually unrelated motor or lexical-semantic variables, supporting its interpretation as a behavioural index of motor affordance. The repository contains: Aggregated MIL norms (item-level statistics), and Raw data (response times, in milliseconds, for each participant). The shared materials are intended to facilitate stimulus selection, replication, and secondary analyses involving behavioural motor variables. This dataset accompanies the article: Ludington, J. D., & Clarke, A. J. B. (in press). Mime Initiation Latency for Object Photographs: A Behavioral Motor Norm. Journal of Psycholinguistic Research. Full bibliographic details (volume, issue, pages, DOI) will be added upon publication. The norms are released under a Creative Commons Attribution 4.0 International (CC BY 4.0) licence to facilitate reuse with appropriate attribution.
HelexKids 2.0 represents an annotated expansion of the original HelexKids, a word frequency database compiled from Greek and Cypriot elementary school textbooks. This article presents the extended annotation methodology employed to enrich the database with part-of-speech tags, phonological properties (syllable count, phonetic transcription, orthographic and phonological syllabification, and stress), and morphological features (case, number, gender, and inflection class for nouns). All original frequency and lexical measures are retained, and new ones (e.g., PLD20, bigram and biphone frequencies) are added. The annotation process combined automated tools with extensive manual annotation and expert verification. The database was redesigned to be dynamically assembled from the full set of textbooks, with results delivered on the fly. The annotated database comprises 67,802 types and over 1.3 million tokens. We report on stress pattern distributions across major parts of speech and grade levels, show how stress is distributed by word class, and provide hierarchies of word-level syllabic templates. HelexKids 2.0 is freely available through the GRADIENCE project webpage and serves as a valuable resource for psycholinguistic and morphophonological research, as well as for educational applications.
Norming studies on picture naming usually identify different correct names for each picture and provide more information on the dominant name (i.e., the most frequently produced name of a given picture) or weighted values based on all the names. The current study is among the first attempts to establish psycholinguistic norms for both the dominant and secondary names of pictures. The following norms in Mandarin Chinese were provided for 700 color photographs from the LinguaPix database: name agreement, naming latency, name length, image agreement, age of acquisition (AoA), and concept familiarity of the dominant and secondary names, as well as overall accuracy, number of names, H-statistic, familiarity, visual complexity, valence, and arousal of the pictures. This dataset increases the diversity of stimuli available for picture naming studies in Chinese and greatly facilitates stimuli selection by allowing researchers to manipulate not only common psycholinguistic properties of the dominant picture name but also those of the secondary name and the relations between them. The database is available in the Open Science Framework repository ( https://osf.io/5rphx/ ).
Lexical variables such as iconicity or age of acquisition are known to be important sources of variance in psycholinguistic experiments. To control for their influence, researchers working on German Sign Language (DGS) need to use stimuli rated for these variables (e. g., iconicity) by an independent group of participants before implementing their actual experiment. Up to now, several research groups have made such rating data publicly available, but a central resource is currently still lacking. Against this background, this short paper provides a roadmap for the collaborative creation of a so-called “DGS-LEX”, a lexical database for psycholinguistic research on DGS, similar to ASL-LEX. By integrating relevant data from different published and forthcoming studies, this joint effort aims to establish a new database for lexical variables in DGS primarily based upon subjective ratings. (Note: This paper is bilingual. The contents of the English and German versions are identical. The German version can be retrieved from https://doi.org/10.6084/m9.figshare.28282907. A summary in DGS is available from https://doi.org/10.6084/m9.figshare.28282898.)
The study of sign languages has received increasing interest in recent years; however there exist hardly any norms for psycholinguistic variables that have appeared to play an important role in sign language processing. Spoken and sign languages share the processes underlying the linguistic function, but also present specific features. One of these is the fundamental role of iconicity in sign languages, both in production and in their understanding. Research on the Italian Sign Language (LIS) is becoming more frequent and it is essential to have available linguistic databases. The objective of this study, therefore, is to offer a first corpus in LIS, which links iconicity with different psycholinguistic variables. Hearing Italian participants provided ratings of iconicity on a total of 234 sign of the www.spreadthesign.com site. Besides iconicity, the variables studied were: concreteness, imaginability, familiarity, age of acquisition and frequency. The analyzes suggest that imaginability is related to iconicity. This is the first psycholinguistic description based on a corpus of 234 signs and can be useful to other researchers interested in LIS.
Despite the excellent performance of black box approaches to modeling\nsentiment and emotion, lexica (sets of informative words and associated\nweights) that characterize different emotions are indispensable to the NLP\ncommunity because they allow for interpretable and robust predictions. Emotion\nanalysis of text is increasing in popularity in NLP; however, manually creating\nlexica for psychological constructs such as empathy has proven difficult. This\npaper automatically creates empathy word ratings from document-level ratings.\nThe underlying problem of learning word ratings from higher-level supervision\nhas to date only been addressed in an ad hoc fashion and has not used deep\nlearning methods. We systematically compare a number of approaches to learning\nword ratings from higher-level supervision against a Mixed-Level Feed Forward\nNetwork (MLFFN), which we find performs best, and use the MLFFN to create the\nfirst-ever empathy lexicon. We then use Signed Spectral Clustering to gain\ninsights into the resulting words. The empathy and distress lexica are publicly\navailable at: http://www.wwbp.org/lexica.html.\n
Over 70% of the more than 56,000 most frequently used words in simplified Chinese are two-character words (2C-words). The present study collected data on subtitle frequency, number of strokes, number of meanings, familiarity, concreteness, imageability, age of acquisition, subjective frequency, subjective number of meanings (sNOMs), compositionality, emotional experience rating, sensory experience rating (SER), and semantic transparency (ST) for 3,783 commonly used 2C-words. Correlative patterns were identified between the 13 variables, all of which, with the exception of imageability, SER, and sNOM, were significantly predictive of the changes in participants’ RTs in LDs on the target words. In conclusion, skilled readers’ awareness of 2C-words’ features such as concreteness, imageability, compositionality, sNOM, and ST is closely associated with their semantic perception of the constituent characters.
Theories of semantic representation propose that concepts are grounded in multiple forms of experience, yet normative resources capturing this multidimensionality remain scarce, especially in German. We present a database of experiential ratings for 1,985 German nouns, with particularly broad coverage of abstract concepts (a 3:1 ratio of abstract to concrete nouns). In total, 354 native German speakers rated the extent to which the meaning of each word is associated with 16 experiential dimensions spanning perceptual, motor, affective, cognitive, social and magnitude-related domains, complemented by concreteness and familiarity ratings. The database additionally provides the dominant experiential dimension of each word and an exclusivity score quantifying how strongly its profile is dominated by a single dimension relative to the others. Cross-database validation against existing German norms showed strong convergence with our concreteness, valence, and arousal ratings. Longitudinal test-retest assessment of all experiential dimensions, as well as concreteness and familiarity, in a subsample of 36 participants showed moderate to high temporal stability across ratings. To characterize the multidimensional structure captured by the database, we further applied exploratory factor analysis followed by hierarchical clustering in the factor-derived semantic space. These analyses identified three broad factors interpreted as socioemotional, perceptual, and magnitude-related, and four clusters of words with distinct experiential profiles. Summary-level and participant-level data are freely available through the project website, which also provides an interactive interface for exploring the norms. The database provides a resource for research on conceptual representation in psycholinguistics, cognitive neuroscience, and computational semantics.
This repository contains data and supplementary materials for the project “Semantic Transparency, Morphological Structure, and Lexical Characteristics of Mandarin Chinese Compound Words: A Large-Scale Normative Database.” The database provides semantic transparency ratings for 4,405 Mandarin Chinese two-character compound words, including first-constituent transparency, second-constituent transparency, and whole-word transparency. It also includes morphological structure, headedness, and lexical variables such as frequency, character frequency, family size, stroke count, and homophone/homograph-related measures.The shared files support reproducibility and reuse of the database for research on Chinese word recognition, compound processing, morphology, semantic transparency, and psycholinguistic norming. Participant-level data have been anonymized.
Understanding the lexical characteristics of Chinese characters is crucial given their extensive usage and unique logographic structure. In this study, we normed affective ratings (valence and arousal) for 3971 Chinese characters. We investigated the relationships between intensity (mean rating) and ambiguity (rating variability) of these affective variables, alongside additional lexico-semantic variables from Su et al., Behavior Research Methods, 55(6), 2989-3008, (2022). Drawing on lexical data from 25,281 two-character words available in the Chinese Lexicon Project (Tse et al., Behavior Research Methods, 49(4), 1503-1519, 2017, Behavior Research Methods, 55(8), 4382-4402, 2023; Chan & Tse, Behavior Research Methods, 56(7), 7574-7601, 2024), we further explored cross-level relationships between character-level and word-level variables. Multiple regression analyses controlling for various lexical variables revealed several noteworthy patterns. First, we identified a quadratic valence-arousal relationship, such that characters with extreme valence ratings (either highly positive or highly negative) elicited higher arousal compared to neutral characters. This relationship was moderated by arousal ambiguity, partially consistent with previous findings (Brainerd et al. Journal of Experimental Psychology: General, 150(8), 1476-1499, 2021a), Second, we observed consistent quadratic intensity-ambiguity relationships across all variables, supporting the quadratic law proposed by Brainerd et al. Journal of Memory and Language, 121, 104286, (2021b). Finally, significant positive associations occurred between character-level variables and their corresponding word-level variables for both the first and second characters. The strength of these cross-level relationships varied across affective and lexico-semantic variables and may further be influenced by semantic transparency. Overall, our findings advance the understanding of affective and semantic features of Chinese characters and offer insights into the cross-level integration of characters' and words' lexical characteristics. The data reported in this paper are available at: https://osf.io/kh4yx.
Abstraction enables us to categorize experience, learn new information, and form judgments. Language arguably plays a crucial role in abstraction, providing us with words that vary in specificity (e.g., highly generic: tool vs. highly specific: muffler). Yet, human-generated ratings of word specificity are virtually absent. We hereby present a dataset of specificity ratings collected from Italian native speakers on a set of around 1K Italian words, using the Best-Worst Scaling method. Through a series of correlation studies, we show that human-generated specificity ratings have low correlation coefficients with specificity metrics extracted automatically from WordNet, suggesting that WordNet does not reflect the hierarchical relations of category inclusion present in the speakers' minds. Moreover, our ratings show low correlations with concreteness ratings, suggesting that the variables Specificity and Concreteness capture two separate aspects involved in abstraction and that specificity may need to be controlled for when investigating conceptual concreteness. Finally, through a series of regression studies we show that specificity explains a unique amount of variance in decision latencies (lexical decision task), suggesting that this variable has theoretical value. The results are discussed in relation to the concept and investigation of abstraction.
This study examined how well large language models (LLMs) approximate human psychological ratings for early-acquired English words. We used four state-of-the-art LLMs, including GPT-4o and Meta-Llama-3.1, to evaluate 21 static psychological features for 695 words and compared these estimates with human norms. The results showed that LLMs aligned well with human ratings for some features (e.g., Concreteness, Bodily Interactiveness) in terms of rank correlations (rs >.82) and distributional similarities but diverged notably for others (e.g., Iconicity, Arousal; rs <.48). Compared with content words, function words showed more pronounced discrepancies between human and LLM ratings. We also assessed how similarly human- and LLM-derived psychological features predicted words' age of acquisition (AoA), revealing both strong correspondences and systematic biases, depending on the model (differences in correlations ranged from -.27 to.28). Based on these analyses, we identified which features may be reliably estimated using LLMs, which require further refinement, and what methodological considerations are necessary for applying LLM-based measures in cognitive science. We discuss the implications of using LLMs as methodological tools in psychology and cognitive science, highlighting both their practical advantages (e.g., data coverage and data collection efficiency) and theoretical relevance. The present study provides a novel framework for evaluating the cognitive plausibility of LLMs by using lexical psychological features, complementing existing benchmarks.
This paper presents research on word familiarity rate estimation using the 'Word List by Semantic Principles'. We collected rating information on 96,557 words in the 'Word List by Semantic Principles' via Yahoo! crowdsourcing. We asked 3,392 subject participants to use their introspection to rate the familiarity of words based on the five perspectives of 'KNOW', 'WRITE', 'READ', 'SPEAK', and 'LISTEN', and each word was rated by at least 16 subject participants. We used Bayesian linear mixed models to estimate the word familiarity rates. We also explored the ratings with the semantic labels used in the 'Word List by Semantic Principles'.
English: This dataset provides cross-cultural name agreement norms for a set of 237 standardized color photographs. The stimuli were selected from the Bank of Standardized Stimuli (BOSS; Brodeur et al., 2010, 2014) or obtained under a Creative Commons license. We gratefully acknowledge Mathieu Brodeur, Ph.D., for authorizing the use and sharing of the BOSS photographs included in the present dataset, for which a CC-BY licence was obtained. Naming data were collected through online picture-naming questionnaires completed by adult French speakers from two linguistic and cultural contexts: Quebec French speakers in Quebec, Canada, and France French speakers in France. The dataset includes item-level norms for each French variety, including the modal name, modal name agreement, H-value, number of distinct names, two alternative names with their corresponding agreement values, and the percentage and count of blank responses. Available lexical frequency measures are also provided: objective frequency norms for France French using FreqFilms2 from Lexique 2 (New et al., 2004), and subjective frequency ratings for Quebec French using FREQ_MEAN from Desrochers and Thompson (2009). The dataset further includes the 237 photographic stimuli, a supplementary file documenting the post-collection data processing procedure, and R scripts used to compute the norms and perform statistical analyses. Participants did not consent to the deposition of the raw and processed individual-level data, or the data-processing log, in an institutional repository. The dataset is intended to support experimental, clinical, and cross-cultural research involving picture naming and culturally adapted assessment tools for French-speaking populations. -------- Français: Ce jeu de données fournit des normes transculturelles d’accord sur le nom pour un ensemble de 237 photographies couleur standardisées. Les stimuli ont été sélectionnés à partir de la Bank of Standardized Stimuli (BOSS; Brodeur et al., 2010, 2014) ou obtenus sous licence Creative Commons. Nous remercions sincèrement Mathieu Brodeur, Ph. D., d’avoir autorisé l’utilisation et le partage des photographies de la BOSS incluses dans le présent jeu de données, pour lesquelles une licence CC BY a été obtenue. Les données de dénomination ont été recueillies au moyen de questionnaires de dénomination d’images en ligne remplis par des locuteurs adultes du français issus de deux contextes linguistiques et culturels: des locuteurs du français québécois au Québec, Canada, et des locuteurs du français de France en France. Le jeu de données comprend des normes par item pour chaque variété de français, incluant le nom modal, le pourcentage d’accord sur le nom modal, la valeur H, le nombre de noms distincts, deux noms alternatifs accompagnés de leurs valeurs d’accord correspondantes, ainsi que le pourcentage et le nombre de réponses vides. Les mesures de fréquence lexicale disponibles sont également fournies: des normes de fréquence objective pour le français de France, à partir de FreqFilms2 dans Lexique 2 (New et al., 2004), ainsi que des jugements de fréquence subjective pour le français québécois, à partir de la variable FREQ_MEAN de Desrochers et Thompson (2009). Le jeu de données comprend également les 237 stimuli photographiques, un fichier supplémentaire documentant la procédure de traitement des données après la collecte, ainsi que les scripts R utilisés pour calculer les normes et réaliser les analyses statistiques. Les participants n’ont pas consenti à ce que les données brutes et nettoyées, ni le journal de nettoyage des données, soient déposés dans un dépôt institutionnel. Ce jeu de données vise à soutenir la recherche expérimentale, clinique et transculturelle portant sur la dénomination d’images et sur les outils d’évaluation adaptés culturellement aux populations francophones.
In psycholinguistic research, careful selection and control of stimuli are essential for gaining insights into cognitive processes. Within this field, pictures often serve as stimuli, which requires the use of image databases to investigate linguistic, mnemonic, and visual perceptual phenomena in different populations (children without disabilities, adults, elderly people, illiterate, and brain-damage patients; see Soares et al., 2018 for more detail).Although several image databases provide norms for variables such as naming agreement (the most common name assigned to a picture by individuals; Snodgrass & Vanderwart, 1980), conceptual familiarity (the frequency with which individuals encounter or think about the depicted object; Snodgrass & Vanderwart, 1980) and visual complexity (judgments regarding the number of lines, intricacies, and details in an image; Snodgrass & Vanderwart, 1980; see also Székely & Bates, 2000 for an objective measure of picture visual complexity), these norms are available in multiple languages but are often restricted to a limited set of black-and-white line drawings (less than 300). Notably, such black-and-white images have been found to elicit weaker recognition compared to colored pictures (Sanfeliu & Fernandez, 1996;Rossion & Pourtois, 2004). In recent decades, there has been an increased effort to develop colored image datasets in various languages. However, many of them consider small datasets (usually less than 500 pictures, but see Brodeur et al., 2014;and Krautz & Keuleers, 2022, for more extensive datasets) and/or use different normalization protocols that complicate the process of comparing data and planning and executing cross-linguistic experiments (Soares et al., 2018;Duñabeitia et al., 2022;and Zhong et al, 2024 for overviews).The Multilingual Picture (MultiPic) database (Duñabeitia et al., 2022) was designed to address the limitations of previous databases by providing researchers with norms for naming agreement and concept familiarity for a set of colored images (500), selected from an initial pool of 750 images (Duñabeitia et al., 2018). To date, this database spans thirty-three languages, including American English, Australian English, Basque, Belgium Dutch, British English, Cantonese, Catalan, Cypriot Greek, Czech, Finnish, French, German, Greek, Hebrew, Hungarian, Italian, Korean, Lebanese Arabic, Malay, Malaysian English, Mandarin Chinese, Netherlands Dutch, Norwegian, Polish, European Portuguese, Quebec French, Rioplatense Spanish, Russian, Serbian, Slovak, Spanish, Turkish, and Welsh. The images depict specific concepts and general knowledge items, and the same data collection and preprocessing protocols were consistently applied across all languages.Expanding MultiPic to include additional languages, dialects, and language varieties worldwide would enable researchers to investigate lesser-studied languages beyond the predominant focus on English, facilitate direct cross-linguistic comparisons, and deepen our understanding of cognitive processes that are universal versus those that are language-specific. The primary objective of the present study was to norm MultiPic in Galician, a relatively under-researched language, enabling researchers to conduct studies with it. Galician is a Western Ibero-Romance language predominantly spoken in Galicia, an autonomous community in northwestern Spain, where it holds co-official status with Spanish.Psycholinguistic studies on Galician are less prevalent than those on Spanish, Portuguese, Catalan, and Basque (see Comesaña & Sá-Leite, 2024). This paucity of research investigating the specific cognitive mechanisms involved in both the comprehension and production of Galician is likely attributable to several factors, including the language's recent standardization in the 1980s (Pintos, 2025) and the scarcity of databases that would allow for the careful selection of linguistic materials for various experiments. The development of these tools would reinvigorate research on Galician by ensuring that experimental outcomes accurately mirror core cognitive processes. Consequently, they would provide essential scientific evidence to inform public policies related to Galician. This language, which coexists with Spanish, presents a distinctive opportunity to examine psycholinguistic theories of language processing within bilingual contexts.Although we generally adhere to the same data collection and preprocessing procedures described in the MultiPic database (Duñabeitia et al., 2022), several adaptations were necessary to accurately reflect Galician's linguistic reality. These modifications accounted for the linguistic diversity across the region, shaped by the contact between Galician and Spanish. For instance, because Galician remains subordinate to Spanish in many social contexts, speakers often incorporate Spanish words or adapted terms, with variations across different regions (cf. Rei-Doval, 2025). Thus, we considered the diverse linguistic varieties and regional differences within Galician, ensuring that the dataset represents the full spectrum of language use across different areas. This approach not only respects the sociolinguistic context of Galician but also allows for a more comprehensive understanding of the cognitive processes involved in bilingual language processing. That is, it will enable researchers to examine which cognitive processes are general and which are specific to different sociolinguistic contexts. Nevertheless, the experimental method, the preprocessing protocol, and the data structure are comprehensively detailed to provide researchers with the necessary framework for adapting the MultiPic to other languages with comparable sociolinguistic contexts.In conclusion, MultiPic and the Galician MultiPic, in particular, serve as valuable tools that enable researchers to design studies in Galician and other languages, where the properties of the materials have been rigorously tested in parallel.The complete dataset, including the data file, is publicly available in the following repositories: https://figshare.com/articles/dataset/Untitled_Item/19328939 and https://osf.io/ank4g/?view_only=43367d3dd27543b0aa66dfb8e71ce1fc.2 MethodParticipants were recruited over two months through social media and local newspaper advertisements. Their participation was voluntary. A total of 88 Galician speakers were initially recruited, surpassing the median sample size in the original MultiPic project (i.e., 80; Duñabeitia et al., 2022)). Still, three were excluded for not following task instructions (e.g., responding in a language other than Galician or basing answers on familiarity with the picture instead of its name).From the remaining 85 participants (47 women, 34 men, four preferred not to disclose their sex; mean age = 42 [age range of 18 to 82], SD = 19.05), around 28% of participants were from O Grove, 8% from Santiago de Compostela, 8% from A Coruña, 6% from Vigo, and the rest from 24 different places. Even though all were speakers of Galician, in their daily lives, around 28% spoke only Galician, 32% spoke more Galician than Castilian Spanish, 18% spoke both languages equally, 16% spoke more Castilian Spanish than Galician, and 6% spoke only Castilian Spanish. More than half had a university degree.We used the 500 colored pictures from the MultiPic database representing common concrete concepts. These pictures were in PNG format with a 300 × 300 pixels resolution and were initially created by a local artist commissioned by the authors of the original study (Duñabeitia et al., 2018). The set of 500 elements depicted was the same as those used in Duñabeitia et al. (2022), consisting of a pictorial set of digital line drawings derived from a list of imageable and concrete Spanish words taken from ESPAL (Duchon et al., 2013).The Galician MultiPic norming followed the standardized protocol of the original MultiPic project. Instructions were provided in Galician. Sociolinguistic data, including age, gender, number of languages spoken fluently, and possession of a university degree, were collected. However, unlike other languages, the sociolinguistic data for Galician was gathered in greater detail to ensure an accurate understanding of the sociolinguistic reality of the Galician language. Thus, questions were added regarding place of birth and place of residence, age of acquisition of Galician and Spanish, language balance, educational level, and socioeconomic status.Participants received a link and completed the tasks on their computers, tablets, or phones. However, fifteen elderly adults who were not computer literate gave their responses orally, which a team member transcribed. All participants completed the two tasks in the same order using the Gorilla Experiment Builder. First, participants were provided with a link and completed the tasks by typing their responses using a computer, tablet, or smartphone. However, fifteen elderly adults who were not computer literate provided their responses orally, which a team member transcribed. Considering this, all participants named each of the 500 randomly presented images, using no more than one word per concept. Then, they rated their familiarity with each concept on a 100-point scale, ranging from 0 (not familiar at all) to 100 (very familiar). If they did not know the name of an image, they could select the "?" button, which was recorded as an "I don't know" response. Before starting, participants completed two practice trials to familiarize themselves with the procedure. The experiment lasted approximately one hour, with breaks every 50 trials. Responses were coded to account for linguistic variations, including standard (i.e., the form accepted by the Real Academia Galega [Royal Galician Academy]) versus colloquial forms, dialectal differences, and influences from Spanish. To this end, and following preceding studies (Duñabeitia et al., 2018;2022), a native speaker of Galician reviewed and corrected spelling errors while also standardizing responses by merging basic variants of the same names (e.g., hyphenated or pluralized forms).A version of the Galician MultiPic is also available at https://figshare.com/articles/dataset/Untitled_Item/19328939. However, only the data considering the Galician nouns are provided here, even when the Galician noun was not the modal name (which occurred with 52 nouns [highlighted in yellow in the dataset provided in OSF]). Thus, the Galician MultiPic at Figshare includes nine columns (from A to I) as occurs with the other 33 languages of MultiPic, which corresponds to the Language provided, the Code (number of the picture), the Number of Responses, the H statistic, the Modal Response, the Modal Response Percentage, the "I don't know" Response Percentage, the Idiosyncratic Response Percentage, and the Familiarity.Sheet H-STATISTIC contains the calculation of the H-STATISTIC for each picture.Sheet CODEBOOK contains a detailed description of the information collected in both the raw and cleaned data frames used for analyses.Regarding the naming task, two measures were considered as in earlier Multipic studies: the mean H statistic and the mean modal response percentagefoot_0. These were analyzed, and the familiarity measures were recorded as well.The most notable finding is that the data exhibits an averaged H statistic of 0.71 and a mean modal response percentage of 73.56%. As mentioned above, only 52 pictures out of 500 had one unique response. The H statistic for Galician is higher than the average for MultiPic across 33 other languages (0.55). Indeed, only 5 out of 33 languages (Malay, Lebanese, Korean, Mandarin, and Cantonese) have higher H statistic values than Galician, and only 2 (Mandarin and Cantonese) lower mean modal response percentages (73.28% and 59.17%, respectively). Interestingly, of the official languages in the Iberian Peninsula, including Basque, Catalan, Galician, Portuguese, and Castilian Spanish, Galician exhibits the highest H statistic and the lowest mean modal response percentage. In comparison, the H statistic and the mean modal response percentage for Basque are 0.66 and 82.94%, for Catalan 0.45 and 88.98%, for Portuguese 0.37 and 90.38%, and for Castilian Spanish 0.30 and 93%, respectively. Table 1 summarizes the norms for the 500 images of the MultiPic in each of the languages of the Iberian Peninsula. The relatively high mean H statistic and low mean modal response percentages obtained in the current dataset suggest a higher lexical variability when compared to most languages included in the Multipic database and to the languages that coexist in the Iberian Peninsula.Correlation analyses on the H statistic and Familiarity values across languages of the Iberian Peninsula were conducted to validate individual dataset quality. We focus on comparative analyses in these languages because speakers share not only historical and linguistic connections but also cultural and educational influences that shape familiarity judgments. This is particularly relevant for Romance languages like Castilian Spanish, European Portuguese, Catalan, and Galician, which have significant lexical and structural similarities, as well as for Basque, which, despite being non-Romance coexists in the same sociolinguistic environment. While cross-linguistic correlations can occur even between typologically distant languages, as shown in previous studies, our focus here is on a more controlled linguistic and cultural space, allowing for a more precise interpretation of familiarity effects.A correlation analysis performed on the H statistic showed that all the Pearson pairwise correlation coefficients were significant at the p < 0.001 level, with r values ranging between 0.27 (Galician vs. European Portuguese) and 0.59 (Spanish vs. Catalan). The reason why r values between Galician and European Portuguese are the lowest despite their status as closely related languages with a shared medieval history as part of Galician-Portuguese, may be attributed to the distinct sociolinguistic contexts in which they have developed. These differing contexts have played a significant role in shaping lexical variation between the two languages. Note that Galician coexists with Castilian Spanish, a language of high prestige, which has led to the incorporation of numerous lexical borrowings from this language into Galician (see Dubert, 2025). Furthermore, the establishment of an official written standard for Galician did not occur until 1980, highlighting the relatively recent process of linguistic standardization. In contrast, Portuguese is the main language in Portugal and does not coexist with another widely spoken language, except for Mirandese, which is used in the specific region of Miranda do Douro. Additionally, Portuguese has a long-established linguistic tradition, with its first grammar and dictionary dating back to the 16th century. Since these early efforts, a strong normative tradition has been maintained (Santos, 2018).Likewise, the correlation analyses performed on different familiarity scores obtained for each item in each language showed that all the Pearson pairwise correlation coefficients were significant at the p < 0.001 level, with r values ranging between 0.79 (Catalan vs. Basque) and 0.83 (Galician vs. European Portuguese).Besides, a correlation analysis was conducted between the H statistic and Familiarity values in each official or co-official language from the Iberian Peninsula already tested in the MultiPic database. We found low to moderate negative and significant correlations in all of them. That is, the higher the values in familiarity, the lower the values in the H statistic, which makes sense as the higher the H statistic, the lower the name agreement. To be more precise, all Pearson pairwise correlation coefficients were significant at the p < 0.001 level, except for the Galician language ( p =.02), with r values ranging between -.10 (for the Galician) and -.44 (for the Catalan). The smallest correlation was found for Galician. At first, we thought that this was probably because it has greater lexical variability than the other languages. Indeed, if we look at the second language from the Iberian Peninsula that has a high lexical variability (Basque), we can see that it also showed a small correlation value between H statistic and Familiarity (-.25). However, when compared with other languages like Chinese or Malay that also have a great lexical variability we found high significant correlations (-.49 and -.90, respectively). Therefore, a more plausible explanation may lay on the fact that familiarity modulates agreement (and not the other way around). That is, if someone is not familiar (or that much familiar) with an object, they would be hesitant when naming it, and as a consequence, this would lead to lower agreement scores across participants. This would be true for all the languages. However, for Galician more variables than familiarity may be explaining this result such as the already mentioned coexistence with the Castilian language, the recent official written standard for Galician, which means that it is not perfectly implemented, and the desire of some people to reflect their dialectal variant. We recognize, however, that this is a tentative explanation that deserves further examination.Each variable's inter-rater reliability was determined by calculating intraclass correlations (ICCs) via a two-way random consistency model. ICCs revealed acceptable reliability for H statistic (ICC = 0.78 [0.75, 0.81]), and an excellent reliability for familiarity (ICC = 0.92 [0.91, 0.93]).Although all correlations were significant, findings underscore how social and regional factors, such as dialectal variation, hyper-Galician forms, and Spanish influence, shape lexical variation in Galician. A closer look at the responses given to each picture shows that some pictures had multiple interpretations that seem to reflect the variety of realities of the population, i.e., participants used nouns with different meanings (e.g., magnet vs. horseshoe), for example, by using different co-hyponyms or elements of the same semantic field ( figo vs. cebola [fig vs. onion]), or, in some cases, by focusing on different elements or areas of the image (e.g., for the picture of a shoulder, participants used names like costas, pel, or marrón, i.e., back, skin, or brown in English). On the other hand, some pictures were named with synonyms, such as xornal and periódico, two different Galician words to name a newspaper. Importantly, in many cases, the noun participants used depended on the dialectal variety of their region (e.g., vespa, avespa, and avéspora for wasp). Also, in some cases, participants used "hyper-Galician" forms, i.e., linguistic that are created when speakers to use they as or Galician. often these to the of Castilian Spanish, a shaped by the unique context of language contact and the relatively standardization of Galician in Galicia, as For example, the Galician form for is, but the hyper-Galician form is many of lexical from Spanish can be such as the Spanish word for or for In in more than of the images the modal form is the Spanish word (e.g., or or the adapted Spanish word (e.g., and in and Spanish, or and in and Spanish, may however, that the lexical variability in Galician is by the pictures than by the of the language as an In other the same concepts consistently the lowest agreement across languages. This does not seem to be the when calculating the mean H statistic of the Multipic database including all the languages tested this was = and the mean modal response percentage was = These values are to those provided in earlier studies with different of stimuli (e.g., et al., et al., et al., et al., et al., et al., and as Duñabeitia et al. have already out when comparing the data of languages, relatively low mean H statistic and the high mean modal response percentages of the current dataset suggest high name agreement across items, languages, and the materials for their use in different of experiments and these The of data from specific regions and the written of the experiment questions about and to the relatively recent standardization of the Galician language, many speakers are with the spelling or This questions about the of this task in a written Nevertheless, this the of comprehensive planning and cultural when the Galician MultiPic valuable insights for cross-linguistic studies and research, recognition of linguistic diversity while providing a framework for adaptations in other Galician of MultiPic a in psycholinguistic research by providing standardized norms for an language. This the lexical variation in Galician, shaped by regional and language contact with like regional and Multipic is a written than spoken the Galician MultiPic is an essential for cross-linguistic studies and the of bilingual cognitive valuable insights for research on language
The publication of the Vai dictionary based on August Klingenheben’s collection and edited by Raimund Kastenholz is a very fortunate event for African studies. It finally makes the rich lexical database that was compiled by August Klingenheben (1886-1967) accessible. The thorough and consistent editorial work carried out by Kastenholz has made available the results of Klingenheben’s research and documentation project on Vai, which, as we learn from the introduction, spanned a period of over f...
This paper introduces and describes the second edition of the Mixtec Sound Change Database (Auderset & Campbell, 2024), which now includes a module on tone change. Tone change is an under-researched topic in historical linguistics and is virtually absent from cross-linguistic databases. The Mixtec languages from southern Mexico provide an ideal starting point for studies in this area, as they are all tonal and tone can be reconstructed to the proto-language. This updated version of the database thus serves as a model for including tonal data in synchronic and diachronic studies of languages and language families where this feature is present. This will support efforts towards a better understanding of suprasegmental change in Mixtec and beyond.
Valence, Arousal and Dominance ratings for over 13k Italian content words (including nouns, adjectives and verbs)
Abstract In the present study, we developed affective (valence and arousal) and sensory–motor (concreteness and imageability) norms for 210 English idioms rated by native English speakers (L1) and English second-language speakers (L2). Based on internal consistency analyses, the ratings were found to be highly reliable. Furthermore, we explored various relations within the collected measures (valence, arousal, concreteness, and imageability) and between these measures and some available psycholinguistic norms (familiarity, literal plausibility, and decomposability) for the same set of idioms. The primary findings were that (i) valence and arousal showed the typical U-shape relation, for both L1 and L2 data; (ii) idioms with more negative valence were rated as more arousing; (iii) the majority of idioms were rated as either positive or negative with only 4 being rated as neutral; (iv) familiarity correlated positively with valence and arousal; (v) concreteness and imageability showed a strong positive correlation; and (vi) the ratings of L1 and L2 speakers significantly differed for arousal and concreteness, but not for valence and imageability. We discuss our interpretation of these observations with reference to the literature on figurative language processing (both single words and idioms).
Emotion lexicons are useful in research across various disciplines, but the availability of such resources remains limited for most languages. While existing emotion lexicons typically comprise words, it is a particular meaning of a word (rather than the word itself) that conveys emotion. To mitigate this issue, we present the Emotion Meanings dataset, a novel dataset of 6000 Polish word meanings. The word meanings are derived from the Polish wordnet (plWordNet), a large semantic network interlinking words by means of lexical and conceptual relations. The word meanings were manually rated for valence and arousal, along with a variety of basic emotion categories (anger, disgust, fear, sadness, anticipation, happiness, surprise, and trust). The annotations were found to be highly reliable, as demonstrated by the similarity between data collected in two independent samples: unsupervised (n = 21,317) and supervised (n = 561). Although we found the annotations to be relatively stable for female, male, younger, and older participants, we share both summary data and individual data to enable emotion research on different demographically specific subgroups. The word meanings are further accompanied by the relevant metadata, derived from open-source linguistic resources. Direct mapping to Princeton WordNet makes the dataset suitable for research on multiple languages. Altogether, this dataset provides a versatile resource that can be employed for emotion research in psychology, cognitive science, psycholinguistics, computational linguistics, and natural language processing.
SUBTLEX-SR is a subtitle-based frequency norm for Serbian, the Serbian member of the SUBTLEX family of psycholinguistic frequency resources (following Brysbaert & New, 2009). It provides word-form and lemma frequencies, contextual diversity, and dispersion measures derived from the Serbian portion of OpenSubtitles v2018. **Contents.** The resource consists of two lexical tables: - **Wordform table** (2,198,809 entries): one row per surface form, with frequency from a 50-million-token lemmatized subsample, frequency from the full 64,842-film cleaned corpus, contextual diversity, three dispersion measures (Gries DP, DPnorm, Juilland D over decade buckets), and POS distribution.- **Lemma table** (330,535 entries): one row per lemma, with subsample-derived frequency, contextual diversity, and POS distribution. Both tables are provided in two scripts (Latin and Cyrillic) and in two formats (CSV and Apache Parquet). Methodology JSON files documenting all construction decisions, and a deterministic per-film manifest of the lemmatized subsample, are included for reproducibility. Four figures from the accompanying paper (baseline correlations, register divergence, Zipf distribution, CD vs. frequency) are also included. **Headline numbers.** 64,842 distinct films (the contextual-diversity base); 287.6 million alphabetic tokens in the cleaned corpus; 50.5 million classla-tokens in the lemmatized subsample; year coverage 1902–2020 with frequency data restricted to films from 1950 onwards. **Construction summary.** Source: OpenSubtitles v2018 raw Serbian (178,596 XML files). Cleaning included repair of a CP-1250-as-CP-1252 mojibake encoding error affecting 90.6% of source documents, two-phase deduplication consolidating 178,494 cleaned uploads into 64,842 distinct films, language-script normalization, and contamination filtering. Lemmatization performed with classla 2.2.1 (standard model variant) on a stratified subsample of 8,933 films. **Validation.** SUBTLEX-SR correlates strongly with OPUS's pre-computed Serbian frequency table (Pearson *r* = 0.97 on 498,489 forms; internal-consistency check) and moderately with the web-derived srLex baseline (*r* = 0.68 on 216,260 forms; *r* = 0.69 on 35,968 lemmas). The reduced srLex correlation reflects a register difference between subtitle dialogue and web prose; the divergent vocabulary sorts coherently into dialogue-characteristic classes (negated future-tense auxiliaries, vocatives, interjections) on one side and news/government-characteristic classes (country and region names, politicians' surnames, formal connectives) on the other. **Limitations.** No behavioral validation against lexical decision RT data has been performed for this release; this is being pursued through collaboration with the Laboratory for Experimental Psychology, University of Novi Sad. classla's Serbian model lemmatizes a small set of Serbian forms to their Croatian variants (e.g., *šta → što*, *koga → tko*); this affects all classla-sr users and is documented in the README. The corpus contains translation residue from non-Serbian source films (predominantly English-language Hollywood and BBC content) and some Bosnian/Croatian-orthography forms typical of the BCMS continuum. See the README and methodology JSON files for full documentation. **Companion paper.** Popović, M. (in preparation). *SUBTLEX-SR: A subtitle-based frequency norm for Serbian, with attention to register coverage in existing Serbian frequency resources.* Submitted to Language Resources and Evaluation. **License.** CC BY-SA 4.0. Source subtitle text is not redistributed with this deposit; see the README for source-corpus access via OPUS. **References.** - Brysbaert, M., & New, B. (2009). Moving beyond Kučera and Francis: A critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for American English. *Behavior Research Methods*, 41(4), 977–990.- Lison, P., & Tiedemann, J. (2016). OpenSubtitles2016: Extracting large parallel corpora from movie and TV subtitles. *Proceedings of LREC 2016*, 923–929.- Ljubešić, N., & Dobrovoljc, K. (2019). What does neural bring? Analysing improvements in morphosyntactic annotation and lemmatisation of Slovenian, Croatian and Serbian. *Proceedings of BSNLP 2019*, 29–34.
In this paper, we present our recent experience in constructing a first-of-its-kind functional corpus based on the theoretical framework of Systemic Functional Linguistics. Annotated on selected texts from the Penn Treebank, the corpus was built by a collaborative team on a web-based annotation platform with several advanced features. After a discussion on the background and motivation of the project, we present our solutions to some of the challenges encountered in the collaborative annotation process. With fine-grained annotations of an initial corpus now available, the corpus can serve as a valuable linguistic resource that complements existing semantically annotated corpora and aids in the development of a larger-scale resource crucial for automated systems for analysis of linguistic function.
. In Study 2, we used estimates and human ratings to predict valence effects in a lexical decision task. Again, we found that norms derived from vector space models and label lists for adults outperformed other combinations and approximated the functional form of children's and adults' valence effects best. We discuss the practical implications of these findings. Additionally, a new set of valence ratings from children (mean age = 12.5 years) for 535 German words is made available.
A long-standing goal shared by researchers has been to design optimal experimental procedures, including the selection of appropriate stimuli. Pictures are commonly used in different research fields. However, until recently, researchers have relied mostly on line-drawings, which can have poor ecological validity. We developed a set of high quality standardized photographs of objects from six different categories, recorded under two camera viewpoints, and five presentation conditions (on its own, held by clean hands, and by hands covered with different substances: sauce, chocolate and mud). These various staging conditions can be used to induce different emotional states while maintaining the object of interest constant. We first report normative data on the objects' name agreement and familiarity collected from North American and Portuguese participants. Results showed high name agreement and familiarity in both samples. Next, arousal, disgust and valence ratings were collected for the stimuli under either an emotional-activating or a neutral context. Subjective ratings varied according to the staging condition and the context, confirming that the same items can effectively be used in different emotional conditions. This database allows researchers to select more ecologically-valid stimuli according to their research purposes while considering several variables of interest and avoiding item-selection problems commonly present when comparing responses to neutral and emotional items.
Normative data for naming photographs are essential in psycholinguistic research. However, image naming norms are typically derived from young adults, limiting their relevance for older populations, who are at greater risk for language impairments due to neurological conditions such as stroke, traumatic brain injury or dementia. Further, lexical retrieval declines also in healthy aging, making it essential to establish norms for older adults to distinguish normal from impaired word retrieval. This study provides normative data for 600 photographs of the Bank of Standardized Stimuli (BOSS) focusing on three age cohorts (40-50, 51-65, and 66+). We examined naming accuracy, name agreement, H values, and response times (RT) to explore age-related differences in image naming. Participants completed a web-based oral picture naming task via video conferencing. Results revealed overall high naming accuracy (mean = 80.5%) and name agreement (mean = 87.4%) across the full sample, with modest variability across the range of adults self-reportedly free of neurological deficits. The 51-65 cohort showed the highest accuracy and fastest RTs. Significant correlations between RT and name agreement and H value support the inclusion of RT as key indices of naming difficulty. We discuss the implications of these findings considering psycholinguistic norms, demographic influences, and methodological differences from previous image norming studies. Novel contributions of this study include normative data for a large sample of middle to older age adults including RT and alternative names, expanding the utility of the BOSS image set for examining aging-related changes in lexical access. The study underscores the importance of including RT measures alongside traditional naming norms for improved characterization of visual stimuli. Open access to the updated dataset aims to facilitate future research into age-related language processing and supports personalized applications in cognitive and clinical settings.
Project description This project hosts TUNorms, a set of psycholinguistic word norms for Thai developed to support research on lexical–semantic processing and embodied cognition. The dataset provides normative ratings for imageability, body–object interaction (BOI), and subjective frequency for 627 mono- and multi-syllabic Thai words. The norms were collected from Thai university students using standardised rating procedures. Reliability was assessed through internal consistency and cross-linguistic comparisons with existing norms for overlapping items, and the measures were further validated in a semantic categorisation task demonstrating independent effects of imageability and BOI on response latencies after controlling for established lexical variables. The repository contains the aggregated normative data and the full rating instructions (Thai and English). Raw participant-level data are not included, as this is a completed normative study and the shared materials are intended to support controlled stimulus selection, replication, and secondary analyses. This dataset accompanies a journal manuscript currently under submission. Upon acceptance, the final citation will be added to this record. The norms are released under a Creative Commons Attribution 4.0 International (CC BY 4.0) licence to facilitate reuse.
We introduce a database (IDEST) of 250 short stories rated for valence, arousal, and comprehensibility in two languages. The texts, with a narrative structure telling a story in the first person and controlled for length, were originally written in six different languages (Finnish, French, German, Portuguese, Spanish, and Turkish), and rated for arousal, valence, and comprehensibility in the original language. The stories were translated into English, and the same ratings for the English translations were collected via an internet survey tool (N = 573). In addition to the rating data, we also report readability indexes for the original and English texts. The texts have been categorized into different story types based on their emotional arc. The texts score high on comprehensibility and represent a wide range of emotional valence and arousal levels. The comparative analysis of the ratings of the original texts and English translations showed that valence ratings were very similar across languages, whereas correlations between the two pairs of language versions for arousal and comprehensibility were modest. Comprehensibility ratings correlated with only some of the readability indexes. The database is published in osf.io/9tga3, and it is freely available for academic research.
A verb as the fundamental part of a sentence is important and its retrieval consists of different cognitive stages. Additionally, verb retrieval difficulty is reported in some types of aphasia and other neurological diseases, and some psycholinguistic variables can influence the verb retrieval process. This study aimed to provide a normative database in the Persian language for 92 black and white action pictures and related verbs in two groups of young ages (20 to 40 years old) and middle ages (41 to 64 years old). A total of 150 volunteers participated in this study, and the groups had similar characteristics due to education. The pictures were normed for variables such as name agreement, familiarity, visual complexity, age of acquisition, and image agreement. Correlation coefficients were calculated values among these measures, and comparisons were made between the two age groups. The results of the comparisons between the two groups showed that name agreement and familiarities were age-dependent. The results revealed that all measures varied with age. Also, the present study provided a set of verbs and their pictures in the Persian language and normative data were obtained on the psycholinguistic variables such that it can be used for clinical practice and research in the areas of verb processing and their naming.
The concreteness-abstractness continuum is considered a primary dimension in the representation of semantic networks. Its theoretical importance and clinical significance are widely acknowledged. To assist and enhance future research, this study collected and evaluated concreteness/abstractness ratings for 9,877 two-character Chinese words retrieved from the MEga study of Lexical Decision in Simplified CHinese (MELD-SCH, Tsang et al, 2018). The ratings were validated through comparisons with previous rating studies on concreteness and imageability of smaller word samples. Relations of word concreteness with word frequency, age-of-acquisition, and efficiency of lexical processing were also examined. These ratings provide an additional dimension of information to two-character words in the database MELD-SCH, permitting not only more comprehensive research on the Chinese language, but also cross-language investigation of the concreteness effect between Chinese and other languages such as English and Dutch where a large database of concreteness ratings is also available.
Word characteristics such as frequency, imageability, concreteness and length are considered good predictors of performance in lexical tasks like picture naming, word comprehension or lexical decision-making. There is also evidence that the age of acquisition (AoA) of words can partly explain aspects of word processing behaviour in later childhood and adulthood (Morrison et al., 1992; Brysbaert & Cortese, 2010).In the present study, we collected AoA norms for 158 nouns and 142 verbs in 22 languages: Afrikaans, British English, Catalan, Danish, Finnish, German, Hebrew, Irish, IsiXhosa, Italian, Lithuanian, Luxembourgish, Maltese, Norwegian, Polish, Russian, Serbian, Slovak, South African English, Spanish, Swedish and Turkish. In a preparatory picture naming procedure, adult native speakers of 34 languages were asked to name 508 object and 504 action pictures. Words shared among the target languages were retained for the final corpus. Our study followed the typical procedure for establishing AoA (see Morrison et al. 1997) and was performed on-line (see www.words-psych.org). 804 adult participants (at least 20 for each language) were asked to specify the age at which they learned the words in their native language. The vast majority of words were rated as acquired by the age of 7 years, demonstrating overlap in early vocabulary across diverse languages. Significant correlations between all language pairs point to a similar developmental sequence for the words under investigation. No previous study has compared AoA judgements on a shared set of words in a wide range of languages. 'The AoA data collected in the 22 languages provides word characteristics that should assist the design of cross-linguistic psycholinguistic experiments and the preparation of materials for use in the assessment and treatment of language disorders in preschool children. The AoA data are currently being used to control for AoA in the construction of cross-linguistic lexical tasks assessing word knowledge in monolingual and bilingual children.
This project aims to create the first open normative database of Basque emotion words by collecting emotional valence, arousal, and concreteness ratings for 4,200 words from native and highly proficient Basque speakers, using ANEW-based methods and making the results openly available for research and language technology applications. The poster was presented at the fourth CLARIAH-EUS workshop, on November 28, 2025 in Vitoria-Gasteiz.
Concreteness is a fundamental dimension of word semantic representation that has attracted more and more interest to become one of the most studied variables in the psycholinguistic and cognitive neuroscience literature in the last decade. Concreteness effects have been found at both the brain and the behavioral levels, but they may vary depending on the constraints of the context and task demands. In this study, we collected concreteness norms for English and Italian words presented in different context sentences to allow better control and manipulation of concreteness in future psycholinguistic research. First, we observed high split-half correlations and Cronbach's alpha coefficients, suggesting that our ratings were highly reliable and can be used in Italian- and English-speaking populations. Second, our data indicate that the concreteness ratings are related to the lexical density and accessibility of the sentence in both English and Italian. We also found that the concreteness of words in isolation was highly correlated with that of words in context. Finally, we analyzed differences between nouns and verbs in concreteness ratings without significant effects. Our new concreteness norms of words in context are a valuable source of information for future research in both the English and Italian language. The complete database is available on the Open Science Framework (doi: 10.17605/OSF.IO/U3PC4).
A pluricentric language is a language that is used in at least two countries where it has the official status of a state, commonwealth or regional language with at least partially its own (codified) norms that usually contribute to the personal identity of speakers. Pluricentric languages have one dominant variant and (one or) several non-dominant varieties. As a result of the political fragmentation of the Hungarian language area that developed after the First World War, and then, confirmed by the peace treaties after the Second World War, the Hungarian language is one of the pluricentric languages in Europe. The article examines the results of close linguistic contacts in non-dominant varieties of the modern Hungarian language used outside Hungary. The consequences of language contacts are highlighted on the basis of lexical borrowings, which are fixed in a specific online dictionary. The dictionary consists of borrowed words of foreign origin used by autochthonous Hungarian minorities living in the Carpathian Basin outside Hungary. In addition to words and phrases that are used exclusively in the speech and writing of Hungarians in countries neighboring Hungary, words that are also used in Hungary, but with a different meaning, were also collected in the database. As of the end of September 2022, the dictionary database contained 5,034 dictionary entries (words). Since this online loanword list contains direct borrowings from many languages of the Carpathian Basin that are in contact with Hungarian (mostly from the official or state languages of Hungary's neighboring countries, including Slovak, Ukrainian, Romanian, Serbian, Croatian, Slovenian, and German), the database is a rich source for the study of contacts between Hungarian and Indo-European languages. Based on the material of the online dictionary, it was found that among the lexical borrowings of the Hungarian language –as a result of centuries-old contacts between Hungarian and various Slavic languages –borrowings of Slavic origin constitute the largest layer of vocabulary of foreign origin in the Hungarian language. The result of the project is a dictionary database that provides an opportunity for a comparative analysis of the vocabulary of non-dominant variants of the pluricentric Hungarian language.
<ns4:p>Over the past decade, there have been several attempts to standardize cross-linguistic datasets. Since language comparison is a notoriously difficult endeavor, it requires tools that facilitate standardization and are convenient to use. The Concepticon is based on a toolkit provided for cross-linguistic comparison and offers a reference catalog for comparable concepts that appear in concept lists. While curating the Concepticon, we found that a variety of studies in distinct research fields collected information on word properties. However, until recently, no resource existed that contained these data to enable the comparison of the different word properties across languages. This gap was filled by the Database of Norms, Ratings, and Relations (NoRaRe), which is an extension of the Concepticon. Here, we present the major release of both resources - Concepticon Version 3.0 and NoRaRe Version 1.0 - which represents an important step in our data development. We show that extending and adapting the data curation workflow in Concepticon to NoRaRe is useful for the standardization of cross-linguistic datasets. In addition, combining datasets from different research fields enables studies grounded in language comparison. Concepticon and NoRaRe include lexical data for various languages, tools for test-driven data curation, and the possibility for data reuse. The first major release of NoRaRe is also accompanied by a new web application that allows convenient access to the data.</ns4:p>
Anxiety is the unease about a possible future negative outcome. In recent years, there has been growing interest in understanding how anxiety relates to our health, well-being, body, mind, and behaviour. This includes work on lexical resources for word-anxiety association. However, there is very little anxiety-related work on larger units of text such as multiword expressions (MWE). Here, we introduce the first large-scale lexicon capturing descriptive norms of anxiety associations for more than 20k English MWEs. We show that the anxiety associations are highly reliable. We use the lexicon to study prevalence of different types of anxiety- and calmness-associated MWEs; and how that varies across two-, three-, and four-word sequences. We also study the extent to which the anxiety association of MWEs is compositional (due to its constituent words). The lexicon enables a wide variety of anxiety-related research in psychology, NLP, public health, and social sciences. The lexicon is freely available: https://saifmohammad.com/worrylex.html
Naturalistic paradigms provide ecologically valid insights into affective and cognitive processes but often require costly and time-consuming human annotations. Large language models (LLMs) provide a scalable tool for generating human-like affective ratings that could complement traditional behavioral approaches. In this study, we compared affective ratings of narrative segments obtained from young adults, five OpenAI's GPT models, Meta's Llama 3.1, and lexical-level norms from the SCOPE metabase. LLM-derived ratings of hedonic valence showed strong correlations with human ratings and outperformed lexical norms. When applied to fMRI data, LLM-derived ratings identified affective brain networks that substantially overlapped with those revealed by human ratings. These findings demonstrate that LLMs can approximate group-level affective ratings from young adults in naturalistic contexts and serve as a useful complement to traditional behavioral data collection, while underscoring the need for careful evaluation of their generalizability and potential biases.