Crash Course Linguistics 15 (Computational Linguistics)

How do computers do language?

Week 15’s Crash Course Linguistics video explores how we can make computers engage with natural human language.

Topics include the steps computers take when interacting with human language, as well as bias and ethics in natural language processing. This week's Thought Bubble addresses the many linguistic limitations of sign language gloves. Closed captions are available in English.

Crash Course has previously touched on computational linguistics from a different angle: their 12-minute video Natural Language Processing is part of the Crash Course Computer Science series. Topics include parsing, text generation, speech recognition, and speech synthesis, with some detailed content about spectrograms that is not included in the Crash Course Linguistics series. Closed captions are available in English and Korean.

For a fun conversation about how computers can learn to invent names for ice cream and craft beers, as well as some of the possibilities and limitations of working with artificial intelligence, check out Making machines learn language, a 44-minute Lingthusiasm episode interviewing Dr. Janelle Shane. This episode is available wherever you prefer to find podcasts, and a transcript is available here.

Problems from multiple past International Linguistics Olympiad national contests highlight some challenges of having computers handle natural language, like coping with words with multiple meanings (answers here) and finding the ends of sentences (pp. 12-13; answers here, pp. 12-14). An exercise used in the 2008 competitions in North America and Australia, aw-TOM-uh-tuh, provides practice in using a finite-state automaton to identify possible words in Rotokas (answers here, p. 7).

Finally, see our newsletter 3 Links for Natural Language Processing for more video resources as well as an easily understood article.

Coming next week: resources about writing systems!

Liz, Gretchen, and Lauren

About Mutual Intelligibility

Mutual Intelligibility is a project to connect linguistics instructors with online resources, especially as so much teaching is shifting quickly online due to current events. It's produced by Lauren Gawne and Gretchen McCulloch, with the support of our patrons on Lingthusiasm. Our editor is Liz McCullough.

Mutual Intelligibility posts will always remain free, but if you have a stable income and find that they’re reducing your stress and saving you time, we're able to fund these because of the Lingthusiasm Patreon and your contributions there.

For the 16 weeks of Crash Course Linguistics, Mutual Intelligibility will be sharing the video weekly, along with supporting resources. We will resume our regular link request and recommendations after the Crash Course Linguistics series. For more on how we usually operate, check out our about page.

If you have other comments, suggestions, or ideas of ways to help, please email mutual.intellig@gmail.com.