Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Kendra V. Dickinson has a total of 54 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Morphological processing of inclusive language in Spanish: A usage-based account

    by Kendra V. Dickinson, Ezequiel M. Durand-López

    We examine how form and usage shape the processing of the Spanish gender-inclusive morpheme -e. Within usage-based models, repeated exposure strengthens representations, allowing high-frequency forms to be retrieved holistically while infrequent forms remain compositionally demanding. To test these predictions, we compare the processing of the relatively frequent pronominal quantifier tod-e-s (‘everyone- inc - pl ’) with its masculine counterpart tod-o-s (‘everyone- m - pl ’) and plural inclusive adjectives such as content-e-s (‘happy- inc - pl ’) with masculine forms such as content-o-s (‘happy- m - pl ’). First-language Spanish speakers completed a word-by-word self-paced reading task in which these forms referred to mixed-gender groups, followed by a sociolinguistic questionnaire. The results showed no reliable processing cost for todes relative to todos at the precritical, critical and postcritical regions. By contrast, inclusive adjectives elicited longer reading times than masculine adjectives at the critical word and spillover region. This cost appeared across most participant groups but was absent among those reporting the highest use of inclusive language. A participant-level bootstrap comparison showed that the processing burden was reliably greater for inclusive adjectives than for todes. These findings indicate that inclusive morphology is not processed uniformly across grammatical contexts. Instead, grammatical categories and individual experience modulate entrenchment, suggesting adult speakers can incorporate emergent morphology into real-time processing routines through exposure.

  • Open Access

    Plataforma da Diversidade Linguística Brasileira: dados linguísticos para uma IA brasileira

    by Raquel Freitag, Marcia dos Santos Machado Vieira, Juliana Bertucci Barbosa, Miguel Oliveira Jr., Cleber Ataíde, Alana de Santana Correia, Amanda Post da Silveira, André Britto de Carvalho, Andréia Silva Araujo, Brayna Conceição dos Santos Cardoso, et al.

    Generative artificial intelligence is based on large-scale language models (LLMs), which are trained with data most often collected without consent or in breach of copyright. LLMs are trained with billions of words and millions of parameters, but we don't know exactly which texts are selected in the training or which parameters are controlled. While unsupervised learning requires a large volume of data, demanding more and more computational costs and generating energy impacts, supervised learning with structured and tagged data can optimize this process; more than that: supervised learning with structured and tagged data resulting from language documentation projects can contribute directly to the National Artificial Intelligence Plan: “Develop advanced language models in Portuguese, with national data that encompasses our cultural, social and linguistic diversity, to strengthen sovereignty in AI.” In Brazil, in addition to Portuguese and its varieties, there are more than 250 other languages (indigenous, immigration, sign language), which are neglected in digital inclusion due to a lack of structured data. The consortium of laboratories and research groups in this INCT aims to prepare linguistic data for the training of LLMs, considering Brazil's linguistic diversity, with the development of a joint protocol for collecting linguistic data in the field, to be replicated in the groups and laboratories longitudinally, as well as transcription procedures, as well as procedures for transcribing, aligning and labeling linguistic data to create a data set that represents Brazilian linguistic diversity, and conducting studies on linguistic processing of diversity to fine-tune LLMs, helping to reduce asymmetries and prejudice resulting from training LLMs with translations from English. A inteligência artificial generativa é baseada em modelos de língua em larga escala (LLMs), que são treinados com dados na maioria das vezes coletados sem consentimento ou infringindo direitos autorais. LLMs são treinados com bilhões de palavras e milhões de parâmetros, mas não sabemos exatamente quais textos são selecionados no treinamento ou quais parâmetros são controlados. Enquanto o aprendizado não supervisionado requer um grande volume de dados, demandando cada vez mais custos computacionais e gerando impactos energéticos, o aprendizado supervisionado, com dados estruturados e etiquetados pode otimizar esse processo; mais do que isso: o aprendizado supervisionado com dados estruturados e etiquetados resultantes de projetos de documentação linguística podem contribuir diretamente para o Plano Nacional de Inteligência Artificial: “Desenvolver modelos avançados de linguagem em português, com dados nacionais que abarcam nossa diversidade cultural, social e linguística, para fortalecer a soberania em IA.” No Brasil, além do português e suas variedades, há mais de 250 outras línguas (indígenas, de imigração, sinalizadas), negligenciadas na inclusão digital por falta de dados estruturados. O consórcio de laboratórios e grupos de pesquisa neste INCT visa a preparação de dados linguísticos para o treinamento de LLMs, considerando a diversidade linguística brasileira, com o desenvolvimento de um protocolo conjunto de coleta de dados linguísticos em campo, a ser replicado nos grupos e laboratórios longitudinalmente, assim como procedimentos de transcrição, alinhamento e etiquetagem de dados linguísticos para a constituição de um conjunto de dados que represente a diversidade linguística brasileira, e a realização de estudos sobre processamento linguístico da diversidade para o ajuste fino dos LLMs, contribuindo para a redução de assimetrias e preconceito resultantes do treino de LLMs com traduções do inglês.


  • Open Access

    Predicative possession choice in Argentinian Spanish

    by Martin Fuchs, Kendra Dickinson, Scott Schwenter

    This study investigates the expression of predicative possession in Argentinian Spanish, focusing on the alternation between two constructions: tener (‘have’) + NP and estar con (‘be with’) + NP. Building on previous research, we explore the factors that determine speakers’ choices between these constructions, particularly the influence of temporal context and the presence of adverbs. Using a forced-choice experimental design, participants were presented with vignettes varying in temporal duration (durative vs. non-durative) and adverbial modification (extending, limiting, or none). Results show a strong overall preference for the tener + NP construction, particularly in durative contexts. However, the estar con + NP construction is more likely to be selected in non-durative contexts, especially when a limiting adverb is present. These findings suggest that the distinction between the two constructions is not purely semantic but also pragmatically modulated by temporal and contextual factors. We argue that this pattern reflects a subset-superset relationship between the two constructions, where tener + NP can cover a broader temporal scope than estar con + NP. This overlap mirrors cross-linguistic findings on possessive constructions and aspectual distinctions, with implications for understanding grammaticalization processes in Romance languages.

  • Open Access

    Plataforma da Diversidade Linguística Brasileira: Dados linguísticos para uma IA brasileira

    by Raquel Meister Ko Freitag, Marcia dos Santos Machado Vieira, Juliana Bertucci Barbosa, Miguel Oliveira Jr., Cleber Ataíde, Alana de Santana Correia, Amanda Post da Silveira, André Britto de Carvalho, Andréia Silva Araujo, Brayna Conceição dos Santos Cardoso, et al.

    Generative artificial intelligence is based on large-scale language models (LLMs), which are trained with data most often collected without consent or in breach of copyright. LLMs are trained with billions of words and millions of parameters, but we don't know exactly which texts are selected in the training or which parameters are controlled. While unsupervised learning requires a large volume of data, demanding more and more computational costs and generating energy impacts, supervised learning with structured and tagged data can optimize this process; more than that: supervised learning with structured and tagged data resulting from language documentation projects can contribute directly to the National Artificial Intelligence Plan: “Develop advanced language models in Portuguese, with national data that encompasses our cultural, social and linguistic diversity, to strengthen sovereignty in AI.” In Brazil, in addition to Portuguese and its varieties, there are more than 250 other languages (indigenous, immigration, sign language), which are neglected in digital inclusion due to a lack of structured data. The consortium of laboratories and research groups in this INCT aims to prepare linguistic data for the training of LLMs, considering Brazil's linguistic diversity, with the development of a joint protocol for collecting linguistic data in the field, to be replicated in the groups and laboratories longitudinally, as well as transcription procedures, as well as procedures for transcribing, aligning and labeling linguistic data to create a data set that represents Brazilian linguistic diversity, and conducting studies on linguistic processing of diversity to fine-tune LLMs, helping to reduce asymmetries and prejudice resulting from training LLMs with translations from English.

  • Open Access

    Evaluating the form of third-person anaphoric direct objects in Portuguese: A cross-dialectal study

    by Kendra V Dickinson, Luana Lamberti, Scott Schwenter

    Both Brazilian Portuguese (BP) and European Portuguese (EP) permit definite third-person null direct objects (DOs) in anaphoric contexts, but differ with regard to the overt DO variant, with the former variety favoring tonic pronouns, the latter clitics. Previous research on language production has shown that the choice between variants in both varieties is constrained by semantic-pragmatic features such as animacy and specificity. We analyze speaker evaluation of these forms in EP and BP, using an experimental acceptability judgement task in which stimuli varied according to DO variant, animacy, and specificity. Data are drawn from the evaluative responses (n= 1752) of 215 Portuguese-speaking participants. Our results demonstrate that the null variant is evaluated most positively overall in both varieties. For EP and BP respectively, the clitic and tonic variants were evaluated most positively with animate specific referents. Our findings show that the patterns of variation previously found in production are reflected in gradient evaluations of anaphoric DOs in EP and BP. This provides support for the hypothesis that the potential shifts in Brazilian Portuguese regarding the transition from clitic to tonic direct object (DO) pronouns, suggest that the overt variant may have adopted characteristics similar to the clitic DOs in EP.


  • Open Access

    Regularization and Innovation: A Usage-Based Approach to Past Participle Variation in Brazilian Portuguese

    by Kendra V. Dickinson

    This project explores the synchronic variation of participle forms in Brazilian Portuguese (BP). Despite general systematicity, the language maintains many historically irregular participles, which often compete with regularized variants. The language has also developed innovative participles, which tend to exist in variation with regular forms. Adopting a usage-based framework, the study examines how analogical processes affect persistent irregular participles and short-form forms in BP, emphasizing the role of grammatical context and frequency. Data are drawn from the Portuguese Web 2011 corpus, including 12 verbs with long-form Latinate irregulars (n = 4800) and 8 verbs with short-form forms (n = 3200). The results show that long-form Latinate irregulars are more common as adjectives and with the verb estar, while regularized forms are prevalent with ser and in perfect constructions. Conversely, short-form participles occur least frequently in perfect constructions, showing a tendency towards the maintenance of regularity in this context. Additionally, verbs that occur more often in perfect constructions are most resistant to innovation. These findings indicate that perfect constructions play a dual role in promoting and preserving regularity in BP and shed light on how grammar–internal relationships and contexts of occurrence play a role in language variation and change.


  • Open Access

    What Does It Meme? English–Spanish Codeswitching and Enregisterment in Virtual Social Space

    by Kendra V. Dickinson

    This project investigates English–Spanish codeswitching in internet memes posted to the Facebook page, We are mitú (mitú), and analyzes how lexical insertions and quotatives contribute to the enregisterment of linguistic patterns and the construction of collective identity among U.S. Latinx millennials in virtual social spaces. Data include instances of lexical insertion (n = 280) and quotative mixed codes (n = 114) drawn from a collected corpus of 765 image–text memes. The most frequent lexical insertions included food items (e.g., elote and pozole), kinship terms (e.g., abuelita and tía), and culturally specific artifacts or practices (e.g., quinceañera and lotería), which reflect biculturalism and rely on a shared set of references for the construction of a group identity. Additionally, the quotatives in the data construct Spanish-speaking characterological figures that enregister a particular brand of U.S. Latinx millennial identity that includes being bilingual, having Spanish-speaking parents, and having strong ties to Latinx culture. Overall, this work highlights not only internet memes as a vehicle for enregisterment, but also, and more importantly, how the language resources employed within them work to enregister linguistic and cultural norms of U.S. Latinx millennials, and thereby, play a role in identity construction in virtual social spaces.


  • Overt Subject Pronoun Use in Switch-Reference Contexts in Child Spanish Developmental Language Disorder: A Discriminant Function Analysis

    by Kendra V. Dickinson, Pedro Antonio Ortiz-Ramírez, Ana Arrieta-Zamudio, John Grinstead, Blanca Flores-Ávalos

    Purpose: Our study analyzes probabilistic constraints on subject expression previously found in adult Spanish in the speech of typically developing (TD) Spanish-speaking children and children with developmental language disorder (DLD). Previous work shows that children with DLD produce fewer overt subjects than typically developing children, and that the latter acquire constraints on subject expression as they age into adolescence. Our study complements these findings and provides further substance to the grammatical profile of children whose morphosyntactic development diverges from that of typically developing children. Method: Data are drawn from unstructured spontaneous production data from a sample of 19 monolingual Mexican, Spanish-speaking children, collected in 2006–2007. This sample includes 19 children diagnosed with DLD and 19 age-matched, typically developing children. We collected all instances of finite verbs that either did or could have occurred with a subject personal pronoun uttered by the child participants and coded them for several factors including tense–mood–aspect, switch reference, and person and number. Results: We find that children with DLD produce fewer overt subject pronouns in switch reference contexts than typically developing controls, with a significant interaction of group and switch reference. Furthermore, a discriminant function analysis shows that overt pronoun use in switch reference contexts can form part of a useful diagnostic discriminant function, with high levels of sensitivity and specificity. Conclusions: Overall, we find important differences between TD Spanish-speaking children and those diagnosed with DLD regarding rates of overt subjects and sensitivity to the probabilistic constraint of switch reference. This finding contributes to our understanding of the morphosyntactic profiles of children with DLD, as well as the utility of factors such as switch reference in the identification of language disorders.