Benchmarking zero-shot and few-shot approaches for tokenization, tagging, and dependency parsing of Tagalog text.