This paper is concerned with the task of automatically assessing the written proficiency level of non-native (L2) learners of English. Drawing on previous research on automated L2 writing assessment following the Common European Framework of Reference for Languages (CEFR), we investigate the possibilities and difficulties of deriving the CEFR level from short answers to open-ended questions, which has not yet been subjected to numerous studies up to date.The object of our study is twofold: to examine the intricacy involved with both human and automated CEFR-based grading of short answers. On the one hand, we describe the compilation of a learner corpus of short answers graded with CEFR levels by three certified Cambridge examiners. We mainly observe that, although the shortness of the answers is reported as undermining a clear-cut evaluation, the length of the answer does not necessarily correlate with inter-examiner disagreement. On the other hand, we explore the development of a soft-voting system for the automated CEFR-based grading of short answers and draw tentative conclusions about its use in a computer-assisted testing (CAT) setting.

This study investigates the hypothesis of young people having the multi-skills required to switch between formal and informal communication. We collected samples of the written output of students across different media and communication situations. The results obtained through dictation tests show that the students’ level is relatively low, with a majority of grammatical errors. The analysis of linguistic forms common to the corpora indicates that all the participants use traditional spelling in at least one of them. Lastly, we present a qualitative analysis of spelling variation and an overview of the teenagers’ linguistic representations.

Several automatic phonetic alignment tools have been proposed in the literature. They usually rely on pre-trained speaker-independent models to align new corpora. Their drawback is that they cover a very limited number of languages and might not perform properly for different speaking styles. This paper presents a new tool for automatic phonetic alignment available online. Its specificity is that it trains the model directly on the corpus to align, which makes it applicable to any language and speaking style. Experiments on three corpora show that it provides results comparable to other existing tools. It also allows the tuning of some training parameters. The use of tied-state triphones, for example, shows further improvement of about 1.5% for a 20 ms threshold. A manually-aligned part of the corpus can also be used as bootstrap to improve the model quality. Alignment rates were found to significantly increase, up to 20%, using only 30 seconds of bootstrapping data.

La présente étude s'intéresse à l'existence d'une pluricompétence qui permettrait aux utilisateurs de nouveaux médias de communication de passer de l'écrit traditionnel à la CEMO (communication écrite médiée par ordinateur) de la même façon qu'ils changent de registre. Nous avons récolté les productions écrites de jeunes de 14 à 15 ans à travers deux supports (électronique / papier) et dans trois situations de communication (dictée, activité en classe, Facebook) ai n d'étudier l'inl uence de ces variables sur la gestion de l'orthographe. Les résultats aux dictées indiquent un niveau relativement bas (une erreur tous les 5 ou 6 mots) avec une majorité d'erreurs grammaticales, ce qui est conforme aux études précédemment menées sur le sujet. L'observation des unités communes aux trois corpus montre que l'on retrouve la forme graphique standard dans au moins un des corpus (sinon plusieurs), et ce, chez tous les élèves. Le même type d'analyse d'unités communes menée sur le corpus Facebook uniquement montre que la forme standard est maîtrisée dans un grand nombre de cas (88 % des formes) par les élèves. Eni n, nous observons que la palette de variantes graphiques utilisée dans les conversations Facebook est assez limitée (principalement abréviations, smileys et caractères échos) et que le taux de compression des formes est assez faible, indiquant que la plupart des formes sont respectées dans leur totalité ou réduites d'un seul caractère. Mots clés : CEMO, nouveaux médias de communication, orthographe, Facebook, variation diaphasique, représentations linguistiquesThe present study investigates the hypothesis of a pluri-competence enabling new information and communication technology users to switch between traditional writing and computer-mediated communication as they change from one register to another. We collected young people's (aged 14-15) written production across dif erent media (electronic/paper) and communication situations (dictation, class activity, Facebook) in order to study the inl uence of these variables on the students' spelling. The results obtained through the dictations show that the students' level URL : http://discours.revues.org/9020 4 Lénaïs Maskens, Louise-Amélie Cougnon, Sophie Roekhaut et Cédrick Fairon is relatively low (one mistake every 5 or 6 words) with a majority of grammatical mistakes, which is in line with previous studies on the subject. The analysis of linguistic units common to the three corpora indicates that all the participants use traditional spelling in at least one of the corpora. The same type of analysis conducted on the Facebook corpus shows that the teenagers master standard spelling in most cases (88% of the forms). Finally, we observe only a limited range of spelling variations in the Facebook conversations as well as a low compression ratio, which indicates that the linguistic units are rarely shortened.

No abstract