The paper "Hire a Linguist!: Learning Endangered Languages with In-Context Linguistic Descriptions" by Kexun Zhang, Yee Man Choi, Zhenqiao Song, Taiqi He, William Yang Wang, and Lei Li introduces LINGOLL (LINGuistic descriptions for Ongoing Language Processing), a novel approach to enable large language models (LLMs) to process and translate endangered languages. The authors observe that while many endangered languages lack large corpora for training LLMs, they often have grammar books or dictionaries, which can serve as valuable linguistic descriptions. LINGOLL leverages these descriptions to process input text from endangered languages, including morphological analysis, dictionary lookups, and grammar knowledge, to improve translation and other NLP tasks. The key contributions of the paper include: 1. **LINGOLL Approach**: A training-free method that integrates linguistic descriptions to process and translate text in endangered languages. 2. **Implementation**: The approach is implemented on two models, GPT-4 and Mixtral, and evaluated on five tasks across eight endangered or low-resource languages. 3. **Performance**: LINGOLL significantly improves GPT-4's performance on translation tasks, increasing BLEU scores from 0 to 10.5 for 10 language directions. It also enhances mathematical reasoning accuracy from 18% to 75% and response selection accuracy from 43% to 63%. The paper highlights the importance of linguistic knowledge in the era of advanced LLMs and demonstrates how existing linguistic resources can make endangered languages more accessible. The authors also discuss the limitations and future directions, emphasizing the need for further research to extend the approach to more languages and address challenges such as tokenization and interface compatibility.The paper "Hire a Linguist!: Learning Endangered Languages with In-Context Linguistic Descriptions" by Kexun Zhang, Yee Man Choi, Zhenqiao Song, Taiqi He, William Yang Wang, and Lei Li introduces LINGOLL (LINGuistic descriptions for Ongoing Language Processing), a novel approach to enable large language models (LLMs) to process and translate endangered languages. The authors observe that while many endangered languages lack large corpora for training LLMs, they often have grammar books or dictionaries, which can serve as valuable linguistic descriptions. LINGOLL leverages these descriptions to process input text from endangered languages, including morphological analysis, dictionary lookups, and grammar knowledge, to improve translation and other NLP tasks. The key contributions of the paper include: 1. **LINGOLL Approach**: A training-free method that integrates linguistic descriptions to process and translate text in endangered languages. 2. **Implementation**: The approach is implemented on two models, GPT-4 and Mixtral, and evaluated on five tasks across eight endangered or low-resource languages. 3. **Performance**: LINGOLL significantly improves GPT-4's performance on translation tasks, increasing BLEU scores from 0 to 10.5 for 10 language directions. It also enhances mathematical reasoning accuracy from 18% to 75% and response selection accuracy from 43% to 63%. The paper highlights the importance of linguistic knowledge in the era of advanced LLMs and demonstrates how existing linguistic resources can make endangered languages more accessible. The authors also discuss the limitations and future directions, emphasizing the need for further research to extend the approach to more languages and address challenges such as tokenization and interface compatibility.