Career in Natural Language Processing

I had the pleasure of interviewing my Spanish teacher and the insights into the field of languages were so much more than I expected. He actually started as an engineer, worked in a major IT MNC. He pursued his interest in languages by learning French and Spanish. He had heard about EduFrance in Alliance Francaise, where he was studying French and wanted to explore the field of Natural Language Processing. Research led him to the Master's program in NLP offered by the European Erasmus Mundus program.

Université de Franche-Comté & Universitat Autònoma de Barcelona, Degree Name International Masters, Field Of Study Natural Language Processing and Human Language Technology, 2012 – 2014

• Studied under Erasmus Mundus scholarship by the EU government, they pay for tuition fee and boarding
• Erasmus Mundus – EC pays a certain amount of scholarship to the people who get selected. Funded by 5 years. The course is still offered but without Erasmus Mundus scholarship. Individual scholarships of the universities are still available. Erasmus Mundus can have different courses for a scholarship.
• 2 years program, can be done in up to 4 universities in the UK, France, Portugal, Spain, decided to do from France and Spain
• Entrance requirements
• Certain level proficiency of French and Spanish
• Resume
• Two letters of recommendation
• Mark sheets
• Statement Of Purpose
• Need not have programming background only language background
• The curriculum was not difficult
• With Exercises, assignments, projects, it was more of a practical course – You interact with classmates, professors or research in the library.
• Professors knew English
• Not hard for non-European students
• This course for a programmer to know more about languages, linguist to know more about programming
• He had done internships, however, they were not mandatory – they were offered by the universities.
• One of the projects, related to his internship in Spain. He was given an option to get hired from the same company
• In France, professors asked to join a mailing list through which one would get to know few job openings. Everyone working in NLP in Europe is on that list.

Job options after the Masters

• Certain friends – Europeans working in the European parliament as interpreters
• Phoneticians in companies that deal with text to speech software
• Ph.D. – a lot of good universities offer since it’s a new and upcoming field

IITs, IIIT might have these programs. He met a few people from IITs in certain conferences. But these might be of pure programming, data engineering or machine learning.

Fee for French university for 1 year is 400 euros, certain universities might charge for foreign students. However, this is much less compared to the US. Quality is better in the US. They do better research, they have better grants. Good research opportunities in Paris as well. Other countries are Germany, Spain, Sweden offering NLP programs. They charge around the same with a maximum of  1000 euros.

Job Prospects in Languages

• Interpreter – It's based on the number of words. Even in BPOs, document translation career growth is limited
• Teaching in India - Unless in a school, you do not get paid well, not a secure job.
• Translation/interpretation in India – BPOs – pure call center jobs
• Freelancing of translation – the advantage is when Indian language in involved
• Tourism dept – work as a guide
• Translator in international events for foreign delegates
• If you are really good, you can translate books, you will work in publishing houses
• In the UN, you can get a permanent job for 5 major languages, for others – only part-time jobs.

In companies like Apple or Google, a phonetician is hired but later machine learning takes over. The phonetician would be let go in 1-2 years and the work is continued by programmers. Small companies without these requirements, hire phoneticians. However, quality of work is not that good with Apple or Google in terms of translation, but pretty good in terms of formal language like newspaper articles. These cannot help in terms of Facebook chat or comic book types.

For professions of phoneticians in MNCs, it is better to learn to program so that one need not discontinue after a few years. Programming is very important after taking Masters for good job options.

Since NLP is a new domain, a lot of startups offer services and there is a lot of work in translation.

Pay grades

• In teaching, you might not be paid the same as a native speaker even if you are equally good
• In BPOs, same as other roles
• Tourism Guide pay is good in Europe

Personality trait

You have to learn to be creative. He feels you cannot be straightforward when it comes to language. You have to come up with different ways to deal with language. You have to be updated. You have to see the language the way it is used. You have to know how language works.

They enjoy learning languages and using them. They are open to other cultures – if you have to learn a language, you need to see their movies, read their books, and talk to them. You have to be open to making friends with that person. If you don’t speak Spanish with someone, you forget it. You have to explore to know beyond the classroom to learn a language completely.

Workload

Not much of workload in his current role in Europe, 9-6 work hours. However, he could not comment on the situation in India, since he works in a product based company that is different from service-based company.

In the case of teaching, there is work beyond teaching hours for slides/exams/preparation for class. He says the career growth in this field depends on one’s work. As a researcher – one can become CTO as well if you are good.

Google CTO is one such example. The head of R&D in the company he did his internship in Spain joined after her Ph.D. with Languages background. She grew with the company.

He thinks work does not get monotonous since language keeps evolving, the style of each language is different. Hence there is complexity in software to understand formal vs informal. For example, imagine the complexity in understanding native language written in English. Challenges are different in the translation of different languages. To understand emotions, current tools are not that good. Only a few robots in Japan can do that but again specific to Japanese. There are huge possibilities of research to communicate in sign language, which is upcoming in Europe, gestures to text.

Pain points

The field is not much explored in India but internationally right now, it is boom time with a lot of opportunities abroad. Things might get saturated after 10 years – now you can advantage of the time. In India, depends on the company you work for either in the field of teaching or tourism.

Pursuing NLP

When asked how he would have pursued his interest if he had to do it all over again, he felt it could have been done better. Now, the internet helps, one would have to read more, listen to more music to grow their capabilities. You can contact a person of a foreign language, have an exchange of languages. Research is more accessible now as well. One can read papers on linguistics (language behavior, evolution) and NLP, language acquisition children vs adults – for becoming better teachers.

He suggests if you are interested in pursuing a career in NLP, you can join a good university in linguistics like JNU for formal education and learn programming by yourself. Then you can do a masters in NLP. You will be in a good level in 5 years. You can get better masters opportunities if you go for formal education in linguistics.

"Most of the work is currently being done only in English. That would not be the case in 10 years. NLP will be used to go through social media responses, reviews. The software will be used to make a summary of a 2-page article in newspapers. Anywhere there is text, you can use it."