Formal linguistic competence (getting the form of language right) and functional linguistic competence (using language to accomplish goals in the world) are distinct cognitive skills.The human brain ...contains a network of areas that selectively support language processing (formal linguistic competence), but not other domains like logical or social reasoning (functional linguistic competence).In the late 2010s, large language models trained on word prediction tasks began achieving unprecedented success in formal linguistic competence, showing impressive performance on linguistic tasks that likely require hierarchy and abstraction.Consistent performance on tasks requiring functional linguistic competence is harder to achieve for large language models and often involves augmentations beyond next word prediction.Evidence from cognitive science and neuroscience can illuminate the capabilities and limitations of large language models and pave the way toward better, human-like models of both language and thought.
Large language models (LLMs) have come closest among all models to date to mastering human language, yet opinions about their linguistic and cognitive capabilities remain split. Here, we evaluate LLMs using a distinction between formal linguistic competence (knowledge of linguistic rules and patterns) and functional linguistic competence (understanding and using language in the world). We ground this distinction in human neuroscience, which has shown that formal and functional competence rely on different neural mechanisms. Although LLMs are surprisingly good at formal competence, their performance on functional competence tasks remains spotty and often requires specialized fine-tuning and/or coupling with external modules. We posit that models that use language in human-like ways would need to master both of these competence types, which, in turn, could require the emergence of separate mechanisms specialized for formal versus functional linguistic competence.
Large language models (LLMs) have come closest among all models to date to mastering human language, yet opinions about their linguistic and cognitive capabilities remain split. Here, we evaluate LLMs using a distinction between formal linguistic competence (knowledge of linguistic rules and patterns) and functional linguistic competence (understanding and using language in the world). We ground this distinction in human neuroscience, which has shown that formal and functional competence rely on different neural mechanisms. Although LLMs are surprisingly good at formal competence, their performance on functional competence tasks remains spotty and often requires specialized fine-tuning and/or coupling with external modules. We posit that models that use language in human-like ways would need to master both of these competence types, which, in turn, could require the emergence of separate mechanisms specialized for formal versus functional linguistic competence.
Abstract
With a growing interest in heritage languages from researchers of bilingualism and linguistic theory, the field of heritage-language studies has begun to build on its empirical foundations, ...moving toward a deeper understanding of the nature of language competence under unbalanced bilingualism. In furtherance of this trend, the current work synthesizes pertinent empirical observations and theoretical claims about vulnerable and robust areas of heritage language competence into early steps toward a model of heritage-language grammar. We highlight two key triggers for deviation from the relevant baseline: the quantity and quality of the input from which the heritage grammar is acquired, and the economy of online resources when operating in a less dominant language. In response to these triggers, we identify three outcomes of deviation in the heritage grammar: an avoidance of ambiguity, a resistance to irregularity, and a shrinking of structure. While we are still a ways away from a level of understanding that allows us to predict those aspects of heritage grammar that will be robust and those that will deviate from the relevant baselines, our hope is that the current work will spur the continued development of a predictive model of heritage language competence.
Neural Network Acceptability Judgments Warstadt, Alex; Singh, Amanpreet; Bowman, Samuel R.
Transactions of the Association for Computational Linguistics,
11/2019, Letnik:
7
Journal Article
Recenzirano
Odprti dostop
This paper investigates the ability of artificial neural networks to judge the grammatical acceptability of a sentence, with the goal of testing their linguistic competence. We introduce the Corpus ...of Linguistic Acceptability (CoLA), a set of 10,657 English sentences labeled as grammatical or ungrammatical from published linguistics literature. As baselines, we train several recurrent neural network models on acceptability classification, and find that our models outperform unsupervised models by Lau et al. (2016) on CoLA. Error-analysis on specific grammatical phenomena reveals that both Lau et al.’s models and ours learn systematic generalizations like subject-verb-object order. However, all models we test perform far below human level on a wide range of grammatical constructions.
L’enseignement des langues étrangères vise à développer les compétences et capacités linguistiques et communicatives de l’apprenant. Écrire dans une langue étrangère est l’un des objectifs ...fondamentaux de ceux qui l’étudient. Ainsi, parmi les compétences sur lesquelles l’enseignant porte son attention, on retrouve notamment la capacité à écrire. Il est important d’adopter une bonne méthodologie pour atteindre cet objectif, c’est-à-dire s’assurer que l’élève sait écrire correctement dans la langue cible. Nous savons très bien que tout projet ou programme éducatif réussi s’accompagne d’un processus d’évaluation, depuis la phase d’esquisse du programme jusqu’à la phase de réalisation ou de mise en œuvre. Ainsi se dégage l’objectif de notre recherche à travers lequel nous entendons d’une part vérifier l’efficacité des techniques d’évaluation. D’autre part, connaître et découvrir les difficultés rencontrées dans l’évaluation de la production écrite dans l’enseignement de l’italien au lycée.
Word Meaning in Minds and Machines Lake, Brenden M.; Murphy, Gregory L.
Psychological review,
03/2023, Letnik:
130, Številka:
2
Journal Article
Recenzirano
Odprti dostop
Machines have achieved a broad and growing set of linguistic competencies, thanks to recent progress in Natural Language Processing (NLP). Psychologists have shown increasing interest in such models, ...comparing their output to psychological judgments such as similarity, association, priming, and comprehension, raising the question of whether the models could serve as psychological theories. In this article, we compare how humans and machines represent the meaning of words. We argue that contemporary NLP systems are fairly successful models of human word similarity, but they fall short in many other respects. Current models are too strongly linked to the text-based patterns in large corpora, and too weakly linked to the desires, goals, and beliefs that people express through words. Word meanings must also be grounded in perception and action and be capable of flexible combinations in ways that current systems are not. We discuss promising approaches to grounding NLP systems and argue that they will be more successful, with a more human-like, conceptual basis for word meaning.
Conceptualizations of competence that permeate applied linguistics systematically fail to account for the role of racialization in language learning and assessments thereof. To interrogate the ...racialization of linguistic competence, we first examine its discursive emergence in conjunction with the ideological construction of linguistic homogeneity as central to the naturalization of race within the context of European colonialism. We then track how ideas about linguistic competence took shape jointly with a genre of the human that is overrepresented as white, as well as how this particular genre of the human informed foundational conceptualizations of communicative competence. After analyzing relevant examples of how communicative competence has been taken up in ways that reify this racializing ideology, we end with an alternative conceptualization of the goals of language learning that focuses on the worldviews and lifeways of racialized communities to move beyond universalizing conceptions of competence as the desired outcome.
A one‐page Accessible Summary of this article in non‐technical language is freely available in the Supporting Information online and at https://oasis‐database.org
This article examines the racialized relationship between ideologies of language standardization and what I term “languagelessness.” Whereas ideologies of language standardization stigmatize ...particular linguistic practices understood to deviate from prescriptive norms, ideologies of languagelessness call into question linguistic competence–and, by extension, legitimate personhood–altogether. Throughout the article I show how these ideologies interact with one another, and how assessments of particular individuals' language use often invoke broader ideas about the (in)competence and (il)legitimacy of entire racialized groups. I focus specifically on dimensions of the racialized relationship between ideologies of language standardization and languagelessness in contemporary framings of U.S. Latinas/os and their linguistic practices. I draw on a range of evidence, including ethnographic data collected within a predominantly Latina/o U.S. high school, institutional policies, and scholarly conceptions of language. When analyzed collectively, these sources highlight the racialized ways that ideologies of language standardization and languagelessness become linked in theory, policy, and everyday interactions. In my examination of these data through the lens of racialization, I seek to theorize how ideologies of language standardization and languagelessness contribute to the enactment of forms of societal inclusion and exclusion in relation to different sociopolitical contexts, ethnoracial categories, and linguistic practices.
Este artículo examina la relación racializada entre las ideologías de estandarización lingüística y lo que llamo “languagelessness.” Mientras que las ideologías de estandarización lingüística estigmatizan prácticas lingüísticas específicas consideradas como ajenas a normas preceptivas, las ideologías de languagelessness ponen en duda la competencia lingüística – y por extensión, la persona legítima – por completo. A lo largo del artículo, se muestra cómo estas ideologías interactúan entre sí, y cómo las evaluaciones de la práctica lingüística de ciertas personas proyectan a menudo ideas más generales sobre la (in)competencia y la (i)legitimad de grupos racializados. El artículo se focaliza en ciertas dimensiones de la relación racializada entre las ideologías de estandarización lingüística y de languagelessness en los marcos contemporáneos de l@s Latin@s y sus prácticas lingüísticas. Para esto, este trabajo se basa en un corpus que incluye datos etnográficos recogidos en un instituto estadounidense predominantemente Latin@, programas institucionales, y teorías académicas del lenguaje. Al analizarlos conjuntamente, estos datos resaltan la manera racializada en que las ideologías de estandarización lingüística y de languagelessness se relacionan con los ámbitos de la teoría, la política, y las interacciones diarias. Al examinar estos ejemplos mediante el concepto de la racialización, se pretende teorizar cómo las ideologías de estandarización lingüística y de languagelessness contribuyen a la reproducción de formas de inclusión y exclusión social en relación con distintos contextos sociopolíticos, categorías etnoraciales, y prácticas lingüísticas.
The aim of this article is to analyze the situation of foreign language teaching and learning at the Institute of Foreign Languages of Vytautas Magnus University (VMU), within the framework of the ...promotion of multilingualism in the European Union and the language policy of VMU. Knowledge of the mother tongue and at least two foreign languages at a sufficient level for effective communication is an aspiration set out in the European Union documents. In this study, we examine how the university’s language policy contributes to the goal of EU multilingualism and the challenges posed by its implementation. For this purpose, data on language teaching at VMU Institute of Foreign Languages over the last ten years are analyzed. In addition, a survey conducted in spring 2022 is presented which aimed at finding out which factors influence the students’ learning of other languages. The results of these two investigations reveal that, while students acquire the proficient user level in English, there is a steady decline in the interest to study other languages apart English, and the number of students who choose other languages has been decreasing. Only a small percentage of them reach the proficient user level (B1–B2 on the CEFR scale), whereas the majority of them become familiar with a new language and end their studies at initial levels (A1–A2 on the CEFR scale). The students’ responses suggest that there is room for improvement in the university system to promote multilingualism; the lack of awareness of the importance of languages, the increasing predominance of English and the lack of motivation, as well as the negative experiences associated with learning, have a significant influence on the decline of interest in other foreign languages, thus constituting a major threat to the implementation of multilingualism policies.
Двете книги на проф. Владко Мурдаров предлагат много езикови знания, поднесени достъпно и интригуващо, за да достигнат до голям кръг читатели. Целта в тях, както и в предходната „Разни съвети за ...глаголите", е предлаганата информация да бъде осмислена така, че гласът на езиковеда да бъде чут от всички, за които тези съвети са предназначени, и да спомогне съществено за повишаване на езиковата им култура.
This article confirms the growing gap between the competency and the practice of the Basque language in the Basque territories. There are two reasons for this: on the one hand, the level of relative ...competency of the Basque language is decreasing, because the proportion of those who feel more comfortable in Castilian or French than in Basque is increasing, which is reflected in a lesser feeling of linguistic competence; and, on the other hand, the density of Basque speakers in the relational network of speakers tends to decrease, which in turn reduces the opportunities to practice Basque.