The Panini Linguistics Olympiad: An Attempt
What is it?
The Panini Linguistics Olympiad is the Indian Selection Test to select eight students to represent India as two different teams (Saffron and Green) at the International Linguistics Olympiad. It is organised every year by the Language Technologies Research Center of IIIT Hyderabad. From the First Round, around 30 to 40 students are selected for the Invitational Round held at IIITH for the final stage of team selection.
The IOL is one of 13 International Science Olympiads for secondary school students, and has been held annually since 2003. Each year, teams of young linguists from around the world gather and test their minds against the world’s toughest puzzles in language and linguistics.
No prior knowledge of linguistics or languages is required: even the hardest problems require only your logical ability, patient work, and willingness to think around corners.
What does Panini even mean?
While it does make sense for the exam to be known as The Indian National Linguistics Olympiad, there is another reason for the seemingly absurd name choice. According to Wikipedia:
Pāṇini (Sanskrit: पाणिनि) (pronounced [paːɳɪnɪ], variously dated between fl. 4th century BCE; and “6th to 5th century BCE” was an ancient Sanskrit philologist, grammarian, and a revered scholar in ancient India. Considered “the father of linguistics”, after the discovery and publication of Pāṇini’s work by European scholars in the nineteenth century, his influence on aspects of the development of modern linguists is widely recognized in the profession; his grammar was influential on foundational scholars such as Ferdinand de Saussure and Leonard Bloomfield.
The Pre Exam Anxiety
My biggest fear leading up to the exam was that I had taken the “no prior knowledge is required” a little too seriously for my own good, and the fact that I had bombed the National Stage of the Informatics Olympiad only a few weeks back. As I waited outside the examination hall, I saw students whom I had read about before, including a person who had been to the Invitational Round last year,a silver medallist at the International Mathematical Olympiad 2019 and a member of Team India at Intel ISEF 2019. To say that I felt like I did not belong there and would mess up my exam would a gross understatement.
The Examination Hall
The actual examination itself was very poorly organised in my city in my opinion. It was a 4-hour long examination (9AM — 1PM) where I forgot to bring in my breakfast or lunch in my infinite wisdom. There were approximately 20 students giving the exam from the Kolkata centre and we were placed in a comparatively small room.
The problem was:
Throughout the four hours of the examination, there were no invigilators present in the room (apart from distributing the question paper, coming in to click photographs, to announce that 30 minutes are left and in the end to collect our answer scripts). We were asked to walk up to the first bench if we wanted any additional A4 sized papers for rough work. Within the first five minutes I walked up to collect two pages for myself when one by one, every student started requesting for pages. Obviously, I could refuse to do that. But I didn’t. And this is how I spent my first 10 minutes distributing paper among all the students.
The constant lack of an invigilator naturally meant certain students would engage in dishonest academic practices (from where I come from, It’s called cheating). A verbal rebuke from my end put a temporary stop to it, although they naturally denied their actions.
P1: Estonian Word Forms
The problem consisted of a table containing groups of words in their various forms in English accompanied by their Estonian translations. There were however 14 blank cells which we were expected to fill up accompanied by an explanation of how we would form an Estonian word from a given English phrase.
P2: Stressed Manabo
A list of words were given in the language Sarangani Manobo and we were supposed to analyse the locations of stress(i.e. emphasis marked with `) in the words and deduce where to put the emphasis on a new set of words
P3: Tenji Braille
The Problem involved interpreting and translating words in Japanese Braille to English. I personally feel this was the easiest problem since I scored close to perfect in it.
P4: Bahinemo
For this problem we were given the first 10 multiples of a number less than 10 written in Bahinemo in ascending order. Bahinemo speakers often use a shortened form form for some numbers which may lead to some ambiguity. In the given numbers one was written in its short form and we had to guess the the number from the given ones. We also had to translate a couple of Bahinemo numerals to Hindu-Arabic numbers. I did not attempt this question at all.
P5: Altai-ed up
This was a fairly straightforward problem. We were given 12 sentences in Altai and their English translations. After that we were supposed to translate some English sentences to Altai and vice-versa.
P6: Twisted Translations
This was a very interesting problem dealing with Word Embeddings. Word Embeddings are representations of words as points in an n-dimensional space. They are very useful representations of words for training machine learning models to perform natural language processing tasks. The embedding for a word is computed by looking at the word in a large corpus of text, and looking at what words occur in the context of that word.
In the problem,2 dimensional word embeddings were in two languages, English and Hrangkhol and the word embeddings of 7 English and 10 Hrangkhul words were provided. The question asked for translating some words and for assigning the 2D coordinates to the translations.
Post-Exam Regret
The exam had a cutoff of 49.25 out of 100 and I scored a total of 44.58. I missed being in the top 27 by 4.67 marks. With the benefit of hindsight, there are several things I would have done differently. Preparing with more problems,reading up on basic linguistic theory, bringing snacks and not turning into a paper distributor would be on my priority list. But hey, that’s what next year is for.
