Abstract:

Free online machine translation (MT) such as provided by Google Translate, Bing Translator and Yahoo! Bablefish is now ubiquitous. Web users rely on it whenever they need to gain information from a language with which they are not familiar. So do second language learners - often enough as a shortcut to get their homework done, as a few language teachers have found. Should its use in the classroom context be penalised or can translation automation actually assist with students’ learning?

There is already a small body of research on the role of MT in language learning (LL) dating back to the nineties, but it was linked to a technology which was not as widespread, and addressed only the advanced language learner. New research is needed that takes into account recent developments and the most typical learner: the beginner.

To initiate this research, the paper explores a group of teachers and learners’ reactions towards the following five emerging translation automation technologies:

1. Google Translate Conversation Mode. It combines MT with speech recognition and synthesis. Many students will have this application installed in their smart phone. Rather that supporting language learning one can imagine this doing away with its need: even at its current level of performance (and it’s likely to improve with use) there is no need, i.e., to learn basic Spanish to travel to South America any more.
2. Duolingo (www.duolingo.com). The site aims at simultaneously helping teaching a language and translating the web. It is not clear yet which use it will make of current MT if any, but assumes a model in which translations will be generated automatically from the guesses (and the votes) of the learners. Technology in this case will encourage rather than discourage language learning.
3. Tradukka (www.tradukka.com). The site offers a user-friendly interface in which the language learner can compose into the language being learned from a draft written first in the mother tongue in one box, then translated by MT and finally proofread by the learner at another box on the side. The accuracy of the final output can be self-assessed via its back-translation into the mother tongue box.
4. Gabble-on (www.gabbleon.com). The site started by asking volunteers to evaluate the performance of Google Translate, Bing Translator and Yahoo! Babelfish on small texts and ended up offering general language activities with a focus on maintaining what has been learnt.
5. Microsoft’s Contextual Thesaurus (http://labs.microsofttranslator.com/thesaurus). Is an English-to-English statistical MT system in which you can enter a phrase which generates (as statistical MT does) many paraphrases, some of which are likely to be acceptable and will constitute a thesaurus-like phrase book, more useful for language learning purposes that the typical word thesaurus since words carry meaning best when they are joined together.

We will discuss these teachers and students’ views on whether translation automation will support or undermine language learning. In areas in which the consensus seems to be that it will support learning, we will look at the conditions under which the technology could become most effective.

Keywords:

Language learning, machine translation, computer assisted language learning, second language acquisition.