Pre-requisites for the Computational Linguistics program

In general, the prerequisites to enter the first year (M1 CL) correspond to the formal courses taught in the L3 LTEI program (linguistics courses plus python programming, algorithmics, linear algebra and probability theory). If you speak French, following this programme is a good way to acquire these prerequisites. For M2, the prerequisites correspond to the courses taught in M1.

Please note: the application to L3 is independent from the application to M1. It is entirely possible to apply to both programmes; you simply need to submit two separate applications.

• M1 CL:
  • Computer Science:
    • Prerequisites: Python programming, object-oriented programming, algorithms. Students are expected to have a good command of Python (in particular: understanding variables, control statements, data structures, file handling, object-oriented programming, recursive programming, and trees).
    • To (self-)evaluate your ability to follow the courses, you will find an example of questions HERE (including the links to the required data). Students in the CL track are expected to be able to solve all these questions from the very beginning of the academic year.
  • Linguistics:
    •  Prerequisites: none
    • Advantages: general linguistics, phonetics, phonology, morphology, semantics
  • Mathematics:
    • Prerequisites: basic notions of probability and statistics, linear algebra
    • Advantages: formal language theory (finite automata and regular expressions)
• M2 CL:
  • Prerequisites:
    • Computer Science: advanced programming (including knowledge of pandas, scikit-learn, PyTorch, HuggingFace 🤗), algorithms
    • Linguistics: notions of general linguistics, fundamental knowledge of phonetics, phonology, grammar, formal syntax, formal semantics
    • Statistical learning: basic principles of supervised classification and regression, neural networks, backpropagation, static and contextual word embeddings, transformer architecture, pretrained language models of the encoder type (such as BERT) and decoder type (such as GPT), design of an LLM (neural scaling laws, instruction fine-tuning, etc.)
  • Advantages
    • Knowledge of the NLP field

Here are some additional tips to help students prepare for the Master’s program so that it runs as smoothly as possible:

• Programming:
  • most lab sessions in M2 use python. Being able to easily handle corpora, data structures in python is required.
  • To (self-)assess your ability to follow the courses, you will find sample questions HERE (including links to the necessary data). Students in the CL program are expected to be able to solve all of these problems from the beginning of the year.
• Maths:
  • basic probability theory, basic linear algebra
• NLP and computational linguistics
  • The Jurafsky and Martin’s notebook provides introduction to major tasks in NLP, and the current methods to achieve these tasks
  • the Natural Language Processing with Python book and package is a good way to exemplify NLP methods
  • machine/deep learning for NLP:  Y. Goldberg “Neural Network Methods in Natural Language Processing”, Morgan & Claypool, 2017
  • these notions will be covered in the master’s program, but it might be good to anticipate.

• for questions concerning the M1 year, contact Guillaume Wisniewski
• for questions concerning the M2 year, contact Marie Candito