Computational Linguistics and Natural Language Processing are everywhere around these days. Although, the name sounds like you need knowledge of computer science and linguistics to get into it, most of the people working in this area are computer programmers or engineers.. A lot of students from pure Linguistics background want to enter this field, but they doubt if they can do something good in here. The question boils down to how much linguistics is needed in computational linguistics!
In this blog post, I’d like to answer questions like: What are potential jobs that somebody with a linguistics and some basic computer science background/ programming skills can get? And if students with an engineering or computer science degree have an advantage when applying to such jobs.
AI is a very broad field. It includes robotics, Human-Computer interaction, vision technology, etc. The field which is roughly an intersection between linguistics and AI is known by different names in different departments, each focussing on the over-ambitious goal of “making computers learn/understand/generate natural language”. I have listed some of these names, with what the departments they generally belong to within the brackets:
- Computational Linguistics (Linguistics)
- Semantic Parsing and Natural Language Processing (computer science/programming)
- Speech technology/recognition (electrical engineering and acoustic physics)
- Machine learning (statistics)
- NL semantic logic/predicate calculus (mathematics), etc.
The field is highly interdisciplinary. And the jobs are really cool and high paid. But (and this “but” is important), you wouldn’t do much if you are only a good linguist. You need solid programming skills, preferably along with some basic mathematics (probability, linear algebra, set theory, predicate calculus, etc.), statistics (linear, multiple regression, etc.).
If you are looking for an academic job as computational linguistics professor in a university, then probably basic programming skills will be enough. But if you want to work on real, cutting edge research happening at Microsoft, IBM, Google, Amazon, etc.- you cannot do with basic programming skills. Even at the internship level, these companies demand/ prefer solid programming skills. The industry is more willing to accept people who are exceptionally good in one of the other domains and decently good in linguistics. And, thus, even being an awesome linguist will still take you only so far in this field.
Unfortunately, then, computer programmers are more valued in this field, perhaps even more than mathematicians and statisticians. It is a general conception that working on data does not need too much deep, theoretical linguistic knowledge. For me, that is not true. I do admit that it is easier for programmers to learn surface-level linguistics and do magic with the data, improve machine performance; and really tough for a linguist to make sense of complex language laws computationally. In fact, there are times you will be almost forced to give up a lot of linguistics (as in the way you look at language) to enter this field.
Lastly, a very important thing to consider is your area of interest in linguistics. If you are into syntax, semantics and pragmatics, you will be at an advantage. But I am afraid sociolinguistics, language teaching, historical linguistics, stylistics, etc. aren’t really that hot in this area.
I strongly believe more and more people from linguistics background are needed to widen the scope of research in NLP. But that’s not cake walk. Most of the Linguistics students are from Humanities and Social Sciences departments, with little or no exposure to scripting and data science, which becomes a major weakness in their case.
© 2017 Payal Khullar. All Rights Reserved.