Text normalization methods have been commonly applied to historical language or user-generated content, but less often to dialectal transcriptions. In this paper, we introduce dialect-to-standard normalization – i.e., mapping phonetic transcriptions from different dialects to the orthographic norm of the standard variety – as a distinct sentence-level character transduction task and provide a large-scale analysis of dialect-to-standard normalization methods. To this end, we compile a multilingual dataset covering four languages: Finnish, Norwegian, Swiss German and Slovene. For the two biggest corpora, we provide three different data splits corresponding to different use cases for automatic normalization. We evaluate the most successful sequence-to-sequence model architectures proposed for text normalization tasks using different tokenization approaches and context sizes. We find that a character-level Transformer trained on sliding windows of three words works best for Finnish, Swiss German and Slovene, whereas the pre-trained byT5 model using full sentences obtains the best results for Norwegian. Finally, we perform an error analysis to evaluate the effect of different data splits on model performance.

The Helsinki-NLP team participated in the NADI 2023 shared tasks on Arabic dialect translation with seven submissions. We used statistical (SMT) and neural machine translation (NMT) methods and explored character- and subword-based data preprocessing. Our submissions placed second in both tracks. In the open track, our winning submission is a character-level SMT system with additional Modern Standard Arabic language models. In the closed track, our best BLEU scores were obtained with the leave-as-is baseline, a simple copy of the input, and narrowly followed by SMT systems. In both tracks, fine-tuning existing multilingual models such as AraT5 or ByT5 did not yield superior performance compared to SMT.

Ce travail présente des contributions récentes à l’effort de doter l’occitan de ressources et outils pour le TAL. Plusieurs ressources existantes ont été modifiées ou adaptées, notamment un tokéniseur à base de règles, un lexique morphosyntaxique et un corpus arboré. Ces ressources ont été utilisées pour entraîner et évaluer des modèles neuronaux pour la lemmatisation. Dans le cadre de ces expériences, un nouveau corpus plus large (2 millions de tokens) provenant du Wikipédia a été annoté en parties du discours, lemmatisé et diffusé.

We present lemmatization experiments on the unstandardized low-resourced languages Low Saxon and Occitan using two machine-learning-based approaches represented by MaChAmp and Stanza. We show different ways to increase training data by leveraging historical corpora, small amounts of gold data and dictionary information, and discuss the usefulness of this additional data. In the results, we find some differences in the performance of the models depending on the language. This variation is likely to be partly due to differences in the corpora we used, such as the amount of internal variation. However, we also observe common tendencies, for instance that sequential models trained only on gold-annotated data often yield the best overall performance and generalize better to unknown tokens.

This paper presents CorCoDial, a research project funded by the Academy of Finland aiming to leverage machine translation technology for corpus-based computational dialectology. In this paper, we briefly present intermediate results of our project-related research.

Pro-TEXT is a corpus of keystroke logs written in French. Keystroke logs are recordings of the writing process executed through a keyboard, which keep track of all actions taken by the writer (character additions, deletions, substitutions). As such, the Pro-TEXT corpus offers new insights into text genesis and underlying cognitive processes from the production perspective. A subset of the corpus is linguistically annotated with parts of speech, lemmas and syntactic dependencies, making it suitable for the study of interactions between linguistic and behavioural aspects of the writing process. The full corpus contains 202K tokens, while the annotated portion is currently 30K tokens large. The annotated content is progressively being made available in a database-like CSV format and in CoNLL format, and the work on an HTML-based visualisation tool is currently under way. To the best of our knowledge, Pro-TEXT is the first corpus of its kind in French.

This paper presents OcWikiDisc, a new freely available corpus in Occitan, as well as language identification experiments on Occitan done as part of the corpus building process. Occitan is a regional language spoken mainly in the south of France and in parts of Spain and Italy. It exhibits rich diatopic variation, it is not standardized, and it is still low-resourced, especially when it comes to large downloadable corpora. We introduce OcWikiDisc, a corpus extracted from the talk pages associated with the Occitan Wikipedia. The version of the corpus with the most restrictive language filtering contains 8K user messages for a total of 618K tokens. The language filtering is performed based on language identification experiments with five off-the-shelf tools, including the new fasttext’s language identification model from Meta AI’s No Language Left Behind initiative, released in July 2022.

Occitan is a Romance language spoken mainly in the south of France. It has no official status in the country, it is not standardized and displays important diatopic variation resulting in a rich system of dialects. Recently, a first treebank for this language was created. However, this corpus is based exclusively on texts in the Lengadocian dialect. Our paper describes the work aimed at extending the existing corpus with content in three new dialects, namely Gascon, Provençau and Lemosin. We describe both the annotation of initial content in these new varieties of Occitan and experiments allowing us to identify the most efficient method for further enrichment of the corpus. We observe that parsing models trained on Occitan dialects achieve better results than a delexicalized model trained on other Romance languages despite the latter training corpus being much larger (20K vs 900K tokens). The results of the native Occitan models show an important impact of cross-dialectal lexical variation, whereas syntactic variation seems to affect the systems less. We hope that the resulting corpus, incorporating several Occitan varieties, will facilitate the training of robust NLP tools, capable of processing all kinds of Occitan texts.

This paper outlines the ongoing effort of creating the first treebank for Occitan, a low-ressourced regional language spoken mainly in the south of France. We briefly present the global context of the project and report on its current status. We adopt the Universal Dependencies framework for this project. Our methodology is based on two main principles. Firstly, in order to guarantee the annotation quality, we use the agile annotation approach. Secondly, we rely on pre-processing using existing tools (taggers and parsers) to facilitate the work of human annotators, mainly through a delexicalized cross-lingual parsing approach. We present the results available at this point (annotation guidelines and a sub-corpus annotated with PoS tags and lemmas) and give the timeline for the rest of the work.

Cet article présente un retour d’expérience sur la transformation de corpus annotés pour l’alsacien et l’occitan vers le format CONLL-U défini dans le projet Universal Dependencies. Il met en particulier l’accent sur divers points de vigilance à prendre en compte, concernant la tokénisation et la définition des catégories pour l’annotation.

Cet article présente une expérience d’annotation morphosyntaxique fine du volet serbe du corpus parallèle ParCoLab (corpus serbe-français-anglais). Elle a consisté à enrichir une annotation existante en parties du discours avec des traits morphosyntaxiques fins, afin de préparer une étape ultérieure de parsing. Nous avons comparé trois approches : 1) annotation manuelle ; 2) préannotation avec un étiqueteur entraîné sur le croate suivie d’une correction manuelle ; 3) réentraînement de l’outil sur un petit échantillon validé du corpus, suivi de l’annotation automatique et de la correction manuelle. Le modèle croate maintient une stabilité globale en passant au serbe, mais les différences entre les deux jeux d’étiquettes exigent des interventions manuelles importantes. Le modèle ré-entraîné sur un échantillon de taille limité (20K tokens) atteint la même exactitude que le modèle existant et le gain de temps observé montre que cette méthode optimise la phase de correction.

In this paper, we present a parallel literary corpus for Serbian, English and French, the TALC-sef corpus. The corpus includes a manually-revised pos-tagged reference Serbian corpus of over 150,000 words. The initial objective was to devise a reference parallel corpus in the three languages, both for literary and linguistic studies. The French and English sub-corpora had been pos-tagged from the onset, using TreeTagger (Schmid, 1994), but the corpus lacked, until now, a tagged version of the Serbian sub-corpus. Here, we present the original parallel literary corpus, then we address issues related to pos-tagging a large collection of Serbian text: from the conception of an appropriate tagset for Serbian, to the choice of an automatic pos-tagger adapted to the task, and then to some quantitative and qualitative results. We then move on to a discussion of perspectives in the near future for further annotations of the whole parallel corpus.