Actualités, CAT, Langues, Traductions

What is Neural Machine Translation (NMT)

Last year professionals had talked much about NMT. What is NMT?

Neural machine translation (NMT) is a machine translation approach that uses a large artificial neural network to predict the likelihood of a sequence of words, typically modeling entire sentences in a single integrated model.

All of the machine translation products (websites or apps) were based on algorithms using statistical methods to try to guess the best possible translation for a given word. This technology is called statistical machine translation.

However, one of the limitations of statistical machine translation is that it only translates words within the context of a few words before and after the translated word. For small sentences, it works pretty well. For longer ones, the translation quality could vary.

Now we have a new machine learning technology called deep learning or deep neural networks, one that tries to mimic how the human brain works (at least partially).

At a high-level, neural network translation works with in two stages:

— A first stage models the word that needs to be translated based on the context of this word (and its possible translations) within the full sentence, whether the sentence is 5 words or 20 words long.

— A second stage then translates this word model (not the word itself but the model the neural network has built of it), within the context of the sentence, into the other language.

One way to think about neural network-based translation could be to think of a fluent speaker in another language that would see a word, say “dog”. This would create the image of a dog in his or her brain, then this image would be associated to, for instance “le chien” in French. The neural network would intrinsically know that the word “chien” is masculine in French (“le” not “la”). But, if the sentence were to be “the dog just gave birth to six puppies” , it would picture the same dog with puppies nursing and would then automatically use “la chienne” (female form of “le chien”) when translating the sentence.

Because of this approach, sentences that are generated from a neural network based machine translation are usually better than statistical machine ones but also sound more fluent and natural, as if a human had translated them and not a machine.

Source: Microsoft

Publicités
That's interesting !

Are musicians better language learners?

28b44f5

Today’s economic environment demands that our children become the very best they can be. A lot of demands are placed upon us as parents, and whether we like it or not, we need to help our children navigate their way in today’s fast-paced world and build their skills for the future. But not all methods, from flashcards to baby signing, actually boost a child’s intelligence, language skills or other abilities for success. Reading through many research papers from peer-reviewed scientific journals, I discovered that music training is the only proven method to boost the full intellectual, linguistic and emotional capacity of a child. Thankfully, for the sake of the stress levels of parents and children, for the whole-brain boost, there is no need to emulate Tiger Mother Amy Chua who pushed her children to play classical instruments for several hours a day, often prompting tearful tantrums from her daughters. As music training boosts all the language-related networks in the brain, we would expect it to be beneficial in the acquisition of foreign languages, and this is what the studies have found. When children start studying music before the age of seven, they develop bigger vocabularies. a better sense og grammar and higher verbal IQ. These advantages benefit both the development of their mother tongue and the learning of foreign languages. During these crucial years, the brain is at its sensitive development phase, with 95% of the brain’s growth occurring now. Music training started during this period also boosts the brain’s ability assist in the pronunciation of languages – and this gift lasts for life, as it has been found that adults who had musical training in childhood still retain this ability to learn foreign languages quicker and more efficiently than adults who did not have early childhood music training. Humans first started creating music 500,000 years ago, yet speech and language was only developed 200,000 years ago. Evolutionary evidence, as interpreted by leading researchers such as Robin Dunbar from Oxford University, indicates that speech as a form of communication has evolved from our original development and use of music. This explains why our music and language neural networks have significant overlap, and why children who learn music become better at learning the grammar, vocabulary and pronounciation of any language.

The benefits are not just for those of us whose mother tongues are obscure. Even for English speakers, there is a growing interest in the advantages that come with learning foreign languages. There are many languages that can benefit us in immense ways, from culture to trade – Chinese, Russian, Arabic, French and Spanish to name but a few – and what better way to ensure your child can pick up all these languages than by teaching them the master language that transcends all others: music.

Music training plays a key role in the development of a foreign language in its grammar, colloquialisms and vocabulary. One recent study found that when children aged nine and under were taught music for just one hour a week, research concluded that they exhibited a higher ability to learn both the grammar and the pronunciation of foreign languages, compared to their classmates who had learned a different extracurricular activity.Finnish children are commonly musically trained from a young age (up until the age of seven) with the playful Musiikkileikkikoulu method, but they only start school at age seven and start language learning at nine or older. Despite this « late exposure » to everything excluding music skills, they commonly end up speaking three to five foreign languages.

Take Ken Stringfellow, the American singer-songwriter known for his work in the Posies and REM, as an example of the impact of music on the ability to learn foreign languages. Ten years ago, well into his thirties, he married a French woman and subsequently picked up a whole new language from scratch. Recording with him in Paris, at first I was amazed at how he had learned it so well without any prior background, compared with my 12 years spent studying French at school; but the research explains it. As a musician who made music from toddlerhood, he would have significantly boosted his brain’s capacity for the syntax, semantics and pronunciation of learning any new language in adulthood.

We must not forget that our children often learn the most when they are engaging in free play and discovering the world for themselves. The combination of a bit of music training and a lot of free play certainly has not harmed the Finns, who in the OECD’s Pisa tests are among the top students in the world not just in their language skills, but in their abilities in mathematics, literacy and science.

The future and its economic demands may be uncertain, and there may be a lot of pressure on us parents, but one thing is certain: in order for our children to thrive, we need not impose this pressure on them. The brain takes care of its own development, with a little bit of music and a lot of love and free play.

By Liisa Henriksson-Macaulay