Frequently-asked Questions

First, ask yourself why you want to do LO. If you want to do LO for the following reasons, I suggest that you don’t:

• I want to gain admission to top universities. If that is what you really want, then in my opinion the most effective way is to do maths olympiad. But I strongly advise against doing anything for the sake of university admission. High school is probably the last moment in your life to live freely from the burdens of adulthood, don’t waste it.
• I want to learn a new language. You’re not going to learn any new language by participating in LO. A much better way to learn French is to go to France or, failing that, enrol in a French course with a heavy element of interaction; basically just use it as much as possible.

The following is a semi-good reason:

• I want to study linguistics in university. There is much more to linguistics that what appears on LO. Things like phonology, sociolinguistics, pramatics, and field study skills can never be tested in an exam like this. If you’re not very good at LO, keep in mind that there’re many more facets of linguistics, so don’t get discouraged 🙂

On the other hand, these are good reasons to do LO:

• I enjoy solving puzzles. Although again be mindful that LO problems are just one kind of puzzles, if you don’t do well it doesn’t mean that you’re bad at other kinds of puzzles.
• I want to study mathematical/computational linguistics in university.

If you satisfy yourself that you have a good reason to pursue LO, read the next question.

In a rough order of difficulty:

• IOL Sample Problems – for some reason people miss it, but this is the best problems to try to see if LO is a good fit for you. If you can solve the Advanced level problems you should try a national olympiad
• UKLO – one nice thing about UKLO is the range of problems it has
• NACLO
• APLO
• Onling – the style here is different from the in-person exams, but the difficulty is close to the IOL. What is great is that it is open to everyone, so you could try an IOL-level contest without having to go to the IOL
• IOL Individual.

Vlad Neasçu wrote Linguistics Olympiad: Training guide which is probably the first English book on LO. There are a lot of ideas about how to solve each kind of problems, which you might find useful to adapt.

Most people in the LO sphere classify LO problems into 6 types, roughly ordered by frequency in the IOL (though note that some problems cannot be clearly classified into a single category):

• Morphosyntax (structure), see IOL2023 P3 Coastal Marind
• Semantics (meaning), see IOL2022 P2 Alabama
• Phonology (pronunciation), see IOL2022 P5 Proto-Chamic, Phan Rang Cham and Tsat
• Wildcards (kinship, direction, writing system, politeness, etc.), see IOL2024 Komnzo
• Numbers, see IOL2021 P1 Ekari

My advice for beginners is to master morphosyntax first. At least in Thailand, you could literally get to the IOL just by solving morphosyntax problems. On average, morphosyntax questions compose about 40-55% of the IOL, and they are even more frequent in lower level olympiads. Another great thing about morphosyntax is the grading scheme. If you get past a certain level, you’re almost guaranteed to get at least 8-12 points in any morphosyntax problem, whereas most numbers/semantics problems are 20/0, all-or-nothing.

Although this is by no means one-to-one, comparing to maths olympiad, I think semantics = combinatorics, numbers = number theory, morphosyntax = algebra, and phonology = geometry. This is also ranked in order of how much I enjoy doing each kind of problems, both in maths and linguistics olympiad.

I’ll tell you what works for me, but it doesn’t mean it will work for you as well. As Deng Xiaoping said, “It doesn’t matter whether a cat is white or black as long as it catches mice.” If a method works for you, stick with it.

First I will screen through the entire problem, paying special attention to all the descriptions (the intro, the figures, the diacritics). Then I will look at the translations to understand what features are being tested, or more accurately, what are not being tested. For example, if all the translations are in past tense, then tense is not tested. After you isolate those features, I find that the best way to see the pattern is to abstract everything.

What do I mean by abstraction? At the most simple level, it is to write down all the translations to another notation that doesn’t require you to read the entire translation again, e.g., writing personal pronouns as \(pqr\), where \(p\) indicates grammatical person, \(q\) number, \(r\) gender (and of course more features if the pronouns have more variations). But there are higher levels of abstraction that you could do, like drawing graphs in semantics problem and find an isomorphism between the graph for the unseen language and the translations.

Once you are satisfied with your model, it’s time to write your solution. Solutions come in 2 parts: (i) answers and (ii) explanations. A lot of beginners make mistakes with part (ii), by either explaining too little or too much, but mostly too much. The best solutions in IOL rarely exceed 2 pages, yet many people mistakenly write 5-6 pages solutions. You should aim to explain just enough for a linguist to be able to construct your answers from your explanation. It is a plus if you can do this by using diagrams or abstraction (which if not common you should explain how it corresponds to the translations) because it increases clarity. Don’t ever write how you came up with the model. You will never get any point from it, and the markers will be really mad.

In terms of how I write the solution, I like to write part (i) before part (ii), because it allows me to update my model if I see something fishy in my answers to the tasks. Some people do the opposite, they will try to have a concrete model first before they answer the tasks. Try to write legibly. You don’t need to have perfect handwriting, but make sure that different characters are distinguisable.

Do past papers of exam X.

Read this advice from UKLO: 20 tips by Ellie Warner.

I can’t remember the IPA table. At IOL level, it is good to know the most frequently used symbols, especially the voiceless-voiced pairs of consonants and the vowel chart (for vowel harmony). At lower levels, you don’t even need to know IPA exists. Of course knowing more stuff is good, especially if you want to study linguistics, but there is a tradeoff between the time you spend learning IPA and the time you spend solving actual problems.

No. Although you need to know at least one language, and know it well, in the sense that you should at least understand parts of speech in that language and how the grammar of that language interacts. Otherwise there is literally no way you can solve an LO problem.

An exposure to other olympiads would be advantageous. The most important translational skill is learning how to get stuck and unstuck, and how to do hard exams under tight time conditions. However, there is no singular pathway to LO. People at the IOL are the most diverse group of people I’ve ever met, not just in terms of culture but also their educational background.

No. Throughout my LO journey I have always tried to think of a way to “teach” LO to a general audience, but I failed. I believe that the most time-effective way to improve your LO skills is to solve the problems.

I am not involved in the administration of any LO, so please contact the organisers directly.

Sadly due to the lack of time I cannot help solving individual problems, though if you speak Thai, the people at ภาษาศาสตร์โอลิมปิก LOs Today might be willing to help. If you choose to contact them, tell them I say hi!

You may also want to consider joining an LO Discord server, the link of which I believe can be found on the internet.