Computational linguistics - Topics

How does dependency grammar differ from phrase structure grammar? What is a formal grammar and how is it used to parse sentences? How do corpus tools help date a new word? Can the grammar of a natural language be fully formalized? What tools do etymologists use besides dictionaries? Why is speech recognition still bad at overlapping conversation? How does a parser decide which words attach to which phrase? Do neural machine translation systems actually learn grammar? How do linguists measure variation in real speech? What is the real difference between rule-based and statistical parsing? Why does tokenization break down for languages written without spaces? Can large language models be used for linguistic fieldwork? How do you formalize the difference between 'and' and 'or'? How much annotated data does a dependency parser actually need? At what point does machine translation beat a human translator? How are word embeddings supposed to capture meaning? How do researchers test whether a language model understands syntax? Do machine translation systems compete against human translators? Why can't a search engine answer a question the way a person would? What can computational methods reveal about how languages change over time? What makes a corpus representative enough to base research on? Why do part-of-speech taggers still choke on tweets and forum posts? How do linguists build a typological database?