In cases of pervasive uncertainty, cognitive systems benefit from heuristics or committing to more general hypotheses. Here we have presented a hierarchical cognitive model of lexical processing that synthesizes advances in early rational cognitive models with modern-day neural architectures. Probabilities of higher-order categories derived from layers extracted from the middle layers of an encoder language model have predictive power in accounting for several reading measures for both predicted and unpredicted words and influence even early first fixation duration behavior. The results suggest that lexical processing can take place within a latent, but nevertheless discrete, space in cases of uncertainty.

A major challenge to equity among members of queer communities is the use of one’s chosen forms of reference, such as personal names or pronouns. Speakers often dimiss errors in pronominal use as unintentional, and claim that their errors reflect many decades of fossilized mainstream language use, including attitudes or expectations about the relationship between one’s appearance and acceptable forms of reference. Here, we propose a modeling framework that allows language use and speech communities to change over time, including the adoption of neopronouns and other forms for self-reference. We present a probabilistic graphical modeling approach to pronominal reference that is flexible in the face of change and experience while also moving beyond form-to-meaning mappings. The model critically also does not rely on lexical covariance structure to learn referring expressions. We show that such a model can account for individual differences in how quickly pronouns or names are integrated into symbolic knowledge and can empower computational systems to be both flexible and respectful of queer people with diverse gender expression.

ARBRES is an ongoing project of open science implemented as a platform (“wikigrammar”) documenting both the Breton language itself and the state of research and engineering work in linguistics and NLP. Along its nearly 15 years of operation, it has aggregated a wealth of linguistic data in the form of interlinear glosses with translations illustrating lexical items, grammatical features, dialectal variations... While these glosses were primarily meant for human consumption, their volume and the regular format imposed by the wiki engine used for the website also make them suitable for machine processing. ARBRES Kenstur is a new parallel corpus derived from the glosses in ARBRES, including about 5k phrases and sentences in Breton along with translations in standard French. The nature of the original data — sourced from field linguistic inquiries meant to document the structure of Breton — leads to a resource that is mechanically more concerned with the internal variations of the language and rare phenomena than typical parallel corpora. Preliminaries experiments in using this corpus show that it can help improve machine translation for Breton, demonstrating that sourcing data from field linguistic documentation can be a way to help provide NLP tools for minority and low-resource languages.

A majority of language technologies are tailored for a small number of high-resource languages, while relatively many low-resource languages are neglected. One such group, Creole languages, have long been marginalized in academic study, though their speakers could benefit from machine translation (MT). These languages are predominantly used in much of Latin America, Africa and the Caribbean. We present the largest cumulative dataset to date for Creole language MT, including 14.5M unique Creole sentences with parallel translations—11.6M of which we release publicly, and the largest bitexts gathered to date for 41 languages—the first ever for 21. In addition, we provide MT models supporting all 41 Creole languages in 172 translation directions. Given our diverse dataset, we produce a model for Creole language MT exposed to more genre diversity then ever before, which outperforms a genre-specific Creole MT model on its own benchmark for 23 of 34 translation directions.

Creoles represent an under-explored and marginalized group of languages, with few available resources for NLP research. While the genealogical ties between Creoles and a number of highly resourced languages imply a significant potential for transfer learning, this potential is hampered due to this lack of annotated data. In this work we present CreoleVal, a collection of benchmark datasets spanning 8 different NLP tasks, covering up to 28 Creole languages; it is an aggregate of novel development datasets for reading comprehension relation classification, and machine translation for Creoles, in addition to a practical gateway to a handful of preexisting benchmarks. For each benchmark, we conduct baseline experiments in a zero-shot setting in order to further ascertain the capabilities and limitations of transfer learning for Creoles. Ultimately, we see CreoleVal as an opportunity to empower research on Creoles in NLP and computational linguistics, and in general, a step towards more equitable language technology around the globe.

Zelda Rose is a command line interface for pretraining transformer-based models. Its purpose is to enable an easy start for users interested in training these ubiquitous models, but unable or unwilling to engage with more comprehensive — but more complex — frameworks and the complex interactions between libraries for managing models, datasets and computations. Training a model requires no code on the user’s part and produce models directly compatible with the HuggingFace ecosystem, allowing quick and easy distribution and reuse. A particular care is given to lowering the cost of maintainability and future-proofing, by making the code as modular as possible and taking advantage of third-party libraries to limit ad-hoc code to the strict minimum.

We propose a method for investigating the interpretability of metrics used for the coreference resolution task through comparisons with human judgments. We provide a corpus with annotations of different error types and human evaluations of their gravity. Our preliminary analysis shows that metrics considerably overlook several error types and overlook errors in general in comparison to humans. This study is conducted on French texts, but the methodology is language-independent.

Nous présentons OFCoRS, un système de résolution de coréférence, basé sur le français parlé et un ensemble de modèles Random Forest. L’objectif de ce papier est de comparer l’approche statistique d’OFCoRS avec l’approche neuronale du système DeCoFre. Nous soulignons particulièrement les similarités et différences entre les deux systèmes. Nous comparons ensuite leurs performances sur le corpus français ANCOR et observons que les performances d’OFCoRS s’approchent de celles de DeCoFre. Une analyse détaillée montre également que les deux systèmes affichent de faibles performances sur les coréférences indirectes, montrant ainsi qu’on ne peut pas considérer le traitement des anaphores complexes comme un problème résolu.

The successes of contextual word embeddings learned by training large-scale language models, while remarkable, have mostly occurred for languages where significant amounts of raw texts are available and where annotated data in downstream tasks have a relatively regular spelling. Conversely, it is not yet completely clear if these models are also well suited for lesser-resourced and more irregular languages. We study the case of Old French, which is in the interesting position of having relatively limited amount of available raw text, but enough annotated resources to assess the relevance of contextual word embedding models for downstream NLP tasks. In particular, we use POS-tagging and dependency parsing to evaluate the quality of such models in a large array of configurations, including models trained from scratch from small amounts of raw text and models pre-trained on other languages but fine-tuned on Medieval French data.

Cet article présente un analyseur syntaxique en dépendances pour le français qui se compare favorablement à l’état de l’art sur la plupart des corpus de référence. L’analyseur s’appuie sur de riches représentations lexicales issues notamment de BERT et de FASTTEXT. On remarque que les représentations lexicales produites par FLAUBERT ont un caractère auto-suffisant pour réaliser la tâche d’analyse syntaxique de manière optimale.

Nous proposons la comparaison de deux méthodes de segmentation automatique du français parlé en périodes macro-syntaxiques, qui permettent d’analyser la syntaxe et la prosodie du discours. Nous comparons l’outil Analor (Avanzi et al., 2008) qui a été développé pour la segmentation des périodes prosodiques et les modèles de segmentations utilisant des CRF et des traits prosodiques et / ou morphosyntaxiques. Les résultats montrent qu’Analor divise le discours en plus petits segments prosodiques tandis que les modèles CRF détectent des segments plus larges que les périodes macro-syntaxiques. Cependant, les modèles CRF ont de meilleurs résultats qu’Analor en termes de F-mesure.

Cet article présente un travail qui consiste à étudier si les scores les plus utilisés pour l’évaluation de la résolution des coréférences constituent des métriques de similarité normalisées. En adoptant une démarche purement expérimentale, nous avons vérifié si les scores MUC, B3 , CEAF, BLANC, LEA et le meta-score CoNLL respectent les bonnes propriétés qui définissent une telle métrique. Notre étude montre que seul le score CEAFm est potentiellement une métrique de similarité normalisée.

Natural Language Processing in oral speech segmentation is still looking for a minimal unit to analyze. In this work, we present a comparison of two automatic segmentation methods of macro-syntactic periods which allows to take into account syntactic and prosodic components of speech. We compare the performances of an existing tool Analor (Avanzi, Lacheret-Dujour, Victorri, 2008) developed for automatic segmentation of prosodic periods and of CRF models relying on syntactic and / or prosodic features. We find that Analor tends to divide speech into smaller segments and that CRF models detect larger segments rather than macro-syntactic periods. However, in general CRF models perform better results than Analor in terms of F-measure.

Nous proposons une architecture neuronale avec les caractéristiques principales des modèles neuronaux de ces dernières années : les réseaux neuronaux récurrents bidirectionnels, les modèles encodeur-décodeur, et le modèle Transformer. Nous évaluons nos modèles sur trois tâches d’étiquetage de séquence, avec des résultats aux environs de l’état de l’art et souvent meilleurs, montrant ainsi l’intérêt de cette architecture hybride pour ce type de tâches.

We propose an end-to-end coreference resolution system obtained by adapting neural models that have recently improved the state-of-the-art on the OntoNotes benchmark to make them applicable to other paradigms for this task. We report the performances of our system on ANCOR, a corpus of transcribed oral French, for which it constitutes a new baseline with proper evaluation.

Cet article présente et analyse les premiers résultats obtenus par notre laboratoire pour la construction d’un modèle de résolution des coréférences en français à l’aide de techniques de classifications parmi lesquelles les arbres de décision et les séparateurs à vaste marge. Ce système a été entraîné sur le corpus ANCOR et s’inspire de travaux antérieurs réalisés au laboratoire LATTICE (système CROC). Nous présentons les expérimentations que nous avons menées pour améliorer le système en passant par des classifieurs spécifiques à chaque type de situation interactive, puis chaque type de relation de coréférence.

Cet article présente trois expériences de détection de mentions dans un corpus de français oral : ANCOR. Ces expériences utilisent des outils préexistants d’analyse syntaxique du français et des méthodes issues de travaux sur la coréférence, les anaphores et la détection d’entités nommées. Bien que ces outils ne soient pas optimisés pour le traitement de l’oral, la qualité de la détection des mentions que nous obtenons est comparable à l’état de l’art des systèmes conçus pour l’écrit dans d’autres langues. Nous concluons en proposant des perspectives pour l’amélioration des résultats que nous obtenons et la construction d’un système end-to-end pour lequel nos expériences peuvent servir de base de travail.