This article studies the application of the #BenderRule in Natural Language Processing (NLP) articles according to two dimensions. Firstly, in a contrastive manner, by considering two major international conferences, LREC and ACL, and secondly, in a diachronic manner, by inspecting nearly 14,000 articles over a period of time ranging from 2000 to 2020 for LREC and from 1979 to 2020 for ACL. For this purpose, we created a corpus from LREC and ACL articles from the above-mentioned periods, from which we manually annotated nearly 1,000. We then developed two classifiers to automatically annotate the rest of the corpus. Our results show that LREC articles tend to respect the #BenderRule (80 to 90% of them respect it), whereas 30 to 40% of ACL articles do not. Interestingly, over the considered periods, the results appear to be stable for the two conferences, even though a rebound in ACL 2020 could be a sign of the influence of the blog post about the #BenderRule.

Cet article étudie l’application de la #RègledeBender dans des articles de traitement automatique des langues (TAL), en prenant en compte une dimension contrastive, par l’examen des actes de deux conférences du domaine, TALN et ACL, et une dimension diachronique, en examinant ces conférences au fil du temps. Un échantillon d’articles a été annoté manuellement et deux classifieurs ont été développés afin d’annoter automatiquement les autres articles. Nous quantifions ainsi l’application de la #RègledeBender, et mettons en évidence un léger mieux en faveur de TALN sur cet aspect.

This paper introduces a pretrained word embedding for Manipuri, a low-resourced Indian language. The pretrained word embedding based on FastText is capable of handling the highly agglutinating language Manipuri (mni). We then perform machine translation (MT) experiments using neural network (NN) models. In this paper, we confirm the following observations. Firstly, the reported BLEU score of the Transformer architecture with FastText word embedding model EM-FT performs better than without in all the NMT experiments. Secondly, we observe that adding more training data from a different domain of the test data negatively impacts translation accuracy. The resources reported in this paper are made available in the ELRA catalogue to help the low-resourced languages community with MT/NLP tasks.

In this paper, we introduce a sentence-level comparable text corpus crawled and created for the less-resourced language pair, Manipuri(mni) and English (eng). Our monolingual corpora comprise 1.88 million Manipuri sentences and 1.45 million English sentences, and our parallel corpus comprises 124,975 Manipuri-English sentence pairs. These data were crawled and collected over a year from August 2020 to March 2021 from a local newspaper website called ‘The Sangai Express.’ The resources reported in this paper are made available to help the low-resourced languages community for MT/NLP tasks.

Cet article propose un modèle de réseau de neurones pour la résolution d’équations analogiques au niveau sémantique et entre phrases dans le cadre de la traduction automatique par l’exemple. Son originalité réside dans le fait qu’il fusionne les deux approches, directe et indirecte, de la traduction par l’exemple.

The Hamburg Notation System (HamNoSys) was developed for movement annotation of any sign language (SL) and can be used to produce signing animations for a virtual avatar with the JASigning platform. This provides the potential to use HamNoSys, i.e., strings of characters, as a representation of an SL corpus instead of video material. Processing strings of characters instead of images can significantly contribute to sign language research. However, the complexity of HamNoSys makes it difficult to annotate without a lot of time and effort. Therefore annotation has to be automatized. This work proposes a conceptually new approach to this problem. It includes a new tree representation of the HamNoSys grammar that serves as a basis for the generation of grammatical training data and classification of complex movements using machine learning. Our automatic annotation system relies on HamNoSys grammar structure and can potentially be used on already existing SL corpora. It is retrainable for specific settings such as camera angles, speed, and gestures. Our approach is conceptually different from other SL recognition solutions and offers a developed methodology for future research.

Standard neural machine translation (NMT) allows a model to perform translation between a pair of languages. Multilingual neural machine translation (NMT), on the other hand, allows a model to perform translation between several language pairs, even between language pairs for which no sentences pair has been seen during training (zero-shot translation). This paper presents experiments with zero-shot translation on low resource Indian languages with a very small amount of data for each language pair. We first report results on balanced data over all considered language pairs. We then expand our experiments for additional three rounds by increasing the training data with 2,000 sentence pairs in each round for some of the language pairs. We obtain an increase in translation accuracy with its balanced data settings score multiplied by 7 for Manipuri to Hindi during Round-III of zero-shot translation.

We present CHARCUT, a character-based machine translation evaluation metric derived from a human-targeted segment difference visualisation algorithm. It combines an iterative search for longest common substrings between the candidate and the reference translation with a simple length-based threshold, enabling loose differences that limit noisy character matches. Its main advantage is to produce scores that directly reflect human-readable string differences, making it a useful support tool for the manual analysis of MT output and its display to end users. Experiments on WMT16 metrics task data show that it is on par with the best “un-trained” metrics in terms of correlation with human judgement, well above BLEU and TER baselines, on both system and segment tasks.

fast align is a simple and fast word alignment tool which is widely used in state-of-the-art machine translation systems. It yields comparable results in the end-to-end translation experiments of various language pairs. However, fast align does not perform as well as GIZA++ when applied to language pairs with distinct word orders, like English and Japanese. In this paper, given the lexical translation table output by fast align, we propose to realign words using the hierarchical sub-sentential alignment approach. Experimental results show that simple additional processing improves the performance of word alignment, which is measured by counting alignment matches in comparison with fast align. We also report the result of final machine translation in both English-Japanese and Japanese-English. We show our best system provided significant improvements over the baseline as measured by BLEU and RIBES.

Unlike European languages, many Asian languages like Chinese and Japanese do not have typographic boundaries in written system. Word segmentation (tokenization) that break sentences down into individual words (tokens) is normally treated as the first step for machine translation (MT). For Chinese and Japanese, different rules and segmentation tools lead different segmentation results in different level of granularity between Chinese and Japanese. To improve the translation accuracy, we adjust and balance the granularity of segmentation results around terms for Chinese–Japanese patent corpus for training translation model. In this paper, we describe a statistical machine translation (SMT) system which is built on re-tokenized Chinese-Japanese patent training corpus using extracted bilingual multi-word terms.

This paper is a partial report of an on-going Kakenhi project which aims to improve sub-sentential alignment and release multilingual syntactic patterns for statistical and example-based machine translation. Here we focus on improving a sub-sentential aligner which is an instance of the association approach. Phrase table is not only an essential component in the machine translation systems but also an important resource for research and usage in other domains. As part of this project, all phrase tables produced in the experiments will also be made freely available.

Nous montrons dans une série d’expériences sur quatre langues, sur des échantillons du corpus Europarl, que, dans leur grande majorité, les trigrammes inconnus d’un jeu de test peuvent être reconstruits par analogie avec des trigrammes hapax du corpus d’entraînement. De ce résultat, nous dérivons une méthode de lissage simple pour les modèles de langue par trigrammes et obtenons de meilleurs résultats que les lissages de Witten-Bell, Good-Turing et Kneser-Ney dans des expériences menées en onze langues sur la partie commune d’Europarl, sauf pour le finnois et, dans une moindre mesure, le français.

L’alignement sous-phrastique consiste à extraire des traductions d’unités textuelles de grain inférieur à la phrase à partir de textes multilingues parallèles alignés au niveau de la phrase. Un tel alignement est nécessaire, par exemple, pour entraîner des systèmes de traduction statistique. L’approche standard pour réaliser cette tâche implique l’estimation successive de plusieurs modèles probabilistes de complexité croissante et l’utilisation d’heuristiques qui permettent d’aligner des mots isolés, puis, par extension, des groupes de mots. Dans cet article, nous considérons une approche alternative, initialement proposée dans (Lardilleux & Lepage, 2008), qui repose sur un principe beaucoup plus simple, à savoir la comparaison des profils d’occurrences dans des souscorpus obtenus par échantillonnage. Après avoir analysé les forces et faiblesses de cette approche, nous montrons comment améliorer la détection d’unités de traduction longues, et évaluons ces améliorations sur des tâches de traduction automatique.

G-LexAr est un analyseur morphologique de l’arabe qui a récemment reçu des améliorations substantielles. Cet article propose une évaluation de cet analyseur en tant qu’outil de pré-traitement pour la traduction automatique statistique, ce dont il n’a encore jamais fait l’objet. Nous étudions l’impact des différentes formes proposées par son analyse (voyellation, lemmatisation et segmentation) sur un système de traduction arabe-anglais, ainsi que l’impact de la combinaison de ces formes. Nos expériences montrent que l’utilisation séparée de chacune de ces formes n’a que peu d’influence sur la qualité des traductions obtenues, tandis que leur combinaison y contribue de façon très bénéfique.

In this paper, we present a simple protocol to evaluate word aligners on bilingual lexicon induction tasks from parallel corpora. Rather than resorting to gold standards, it relies on a comparison of the outputs of word aligners against a reference bilingual lexicon. The quality of this reference bilingual lexicon does not need to be particularly high, because evaluation quality is ensured by systematically filtering this reference lexicon with the parallel corpus the word aligners are trained on. We perform a comparison of three freely available word aligners on numerous language pairs from the Bible parallel corpus (Resnik et al., 1999): MGIZA++ (Gao and Vogel, 2008), BerkeleyAligner (Liang et al., 2006), and Anymalign (Lardilleux and Lepage, 2009). We then select the most appropriate one to produce bilingual lexicons for all language pairs of this corpus. These involve Cebuano, Chinese, Danish, English, Finnish, French, Greek, Indonesian, Latin, Spanish, Swedish, and Vietnamese. The 66 resulting lexicons are made freely available.

Les définitions des paraphrases privilégient généralement la conservation du sens. Cet article démontre par l’absurde qu’une évaluation uniquement basée sur la conservation du sens permet à un système inutile de production de paraphrase d’être jugé meilleur qu’un système au niveau de l’état de l’art. La conservation du sens n’est donc pas l’unique critère des paraphrases. Nous exhibons les trois objectifs des paraphrases : la conservation du sens, la naturalité et l’adaptation à la tâche. La production de paraphrase est alors un compromis dépendant de la tâche entre ces trois critères et ceux-ci doivent être pris en compte lors des évaluations.

In this paper we explore the contribution of the use of two Arabic morphological analyzers as preprocessing tools for statistical machine translation. Similar investigations have already been reported for morphologically rich languages like German, Turkish and Arabic. Here, we focus on the case of the Arabic language and mainly discuss the use of the G-LexAr analyzer. A preliminary experiment has been designed to choose the most promising translation system among the 3 G-LexAr-based systems, we concluded that the systems are equivalent. Nevertheless, we decided to use the lemmatized output of G-LexAr and use its translations as primary run for the BTEC AE track. The results showed that G-LexAr outputs degrades translation compared to the basic SMT system trained on the un-analyzed corpus.

This year’s GREYC translation system is an improved translation memory that was designed from scratch to experiment with an approach whose goal is just to improve over the output of a standard translation memory by making heavy use of sub-sentential alignments in a restricted case of translation by analogy. The tracks the system participated in are all BTEC tracks: Arabic to English, Chinese to English, and Turkish to English.

Nous présentons anymalign, un aligneur sous-phrastique grand public. Ses résultats ont une qualité qui rivalise avec le meilleur outil du domaine, GIZA++. Il est rapide et simple d’utilisation, et permet de produire dictionnaires et autres tables de traduction en une seule commande. À notre connaissance, c’est le seul outil au monde permettant d’aligner un nombre quelconque de langues simultanément. Il s’agit donc du premier aligneur sousphrastique réellement multilingue.

This paper describes a new alignment method that extracts high quality multi-word alignments from sentence-aligned multilingual parallel corpora. The method can handle several languages at once. The phrase tables obtained by the method have a comparable accuracy and a higher coverage than those obtained by current methods. They are also obtained much faster.

This year's GREYC machine translation (MT) system presents three major changes relative to the system presented during the previous campaign, while, of course, remaining a pure example-based MT system that exploits proportional analogies. Firstly, the analogy solver has been replaced with a truly non-deterministic one. Secondly, the engine has been re-engineered and a better control has been introduced. Thirdly, the data used for translation were the data provided by the organizers plus alignments obtained using a new alignment method. This year we chose to have the engine run with the word as the processing unit on the contrary to previous years where the processing unit used to be the character. The tracks the system participated in are all classic BTEC tracks (Arabic-English, Chinese-English and Chinese-Spanish) plus the so-called PIVOT task, where the test set had to be translated from Chinese into Spanish by way of English.

The GREYC machine translation (MT) system is a slight evolution of the ALEPH machine translation system that participated in the IWLST 2005 campaign. It is a pure example-based MT system that exploits proportional analogies. The training data used for this campaign were limited on purpose to the sole data provided by the organizers. However, the training data were expanded with the results of sub-sentential alignments. Thesystemparticipatedinthetwoclassicaltasks of translation of manually transcribed texts from Japanese to English and Arabic to English.

On se place ici dans la tendance actuelle en traitement automatique des langues, celle à base de corpus et aussi dans une perspective que l’on peut qualifier d’approche à moindre effort : il s’agit d’examiner les limites des possibilités de traitement à partir de données textuelles brutes, c’est-à-dire non pré-traitées. L’interrogation théorique présente en arrière-plan est la suivante : quelles sont les opérations fondamentales en langue ? L’analogie proportionnelle a été mentionnée par de nombreux grammairiens et linguistes. On se propose de montrer l’efficacité d’une telle opération en la testant sur une tâche dure du traitement automatique des langues : la traduction automatique. On montrera aussi les bonnes conséquences de la formalisation d’une telle opération avec des résultats théoriques en théorie des langages en relation avec leur adéquation à la description des langues. De cette façon, une opération fondamentale en langue, l’analogie proportionnelle, se verra illustrée tant par ses aspects théoriques que par ses performances en pratique.

We designed, implemented and assessed an EBMT system that can be dubbed the “purest ever built”: it strictly does not make any use of variables, templates or training, does not have any explicit transfer component, and does not require any preprocessing of the aligned examples. It uses a specific operation, namely proportional analogy, that implicitly neutralises divergences between languages and captures lexical and syntactical variations along the paradigmatic and syntagmatic axes without explicitly decomposing sentences into fragments. In an experiment with a test set of 510 input sentences and an unprocessed corpus of almost 160,000 aligned sentences in Japanese and English, we obtained BLEU, NIST and mWER scores of 0.53, 8.53 and 0.39 respectively, well above a baseline simulating a translation memory.

We inspect the possibility of creating new linguistic utterances (small sentences) similar to those already present in an existing linguistic resource. Using paradigm tables ensures that the new generated sentences resemble previous data, while being of course different. We report an experiment in which 1,201 new correct sentences were generated starting from only 22 seed sentences.

La construction de banque d’arbres est une entreprise lourde qui prend du temps. Pour faciliter cette construction, nous voyons la construction de banques d’arbres comme une série d’opérations d’édition et de recherche. Le but de cet article est d’estimer l’effort, en nombre d’opérations d’éditions, nécessaire pour ajouter une nouvelle phrase dans la banque d’arbres. Nous avons proposé un outil, Boardedit, qui inclut un éditeur d’arbres et des aides a l’analyse. Comme l’effort nécessaire dépend bien sûr de la qualité des réponses fournies par les aides a l’analyse, il peut être vue comme une mesure de la qualité de ces aides. L’éditeur d’arbres restant indispensable a notre outil pendant l’eXpérience, les aides a l’analyse seront donc toujours associées a l’éditeur d’arbres. Dans l’eXpérience proposée, nous augmentons une banque d’arbres de 5 000 phrases par l 553 nouvelles phrases. La réduction obtenue est supérieure auX 4/5 de l’effort.

L’argumentation générativiste contre l’analogie tenait en trois points: l’hypothèse de l’inné, celle du hors-contexte et la surproduction. Des résultats théoriques et expérimen- taux reposant sur une formulation calculatoire nouvelle de l’analogie contribuent de façon constructive a la réfutation de ces points.