3b
Both speech recognition and speech synthesis are forms of natural language processing. But what does ‘language’ in natural language processing (NLP) mean? While the ‘natural’ in NLP refers to human language, there are at least four ways in which language can be defined or interpreted in NLP. Language, as Alan Turing’s canonical essay on machine intelligence demonstrates, has been a desirable object to imitate computationally. It is also a formalism with rules that can be decoded for computational purposes. Language can be datafied and statistically analyzed, and last but not the least, it is also a contextual variable owing to the multiple languages spoken and written across the world. The development of speech recognition and speech synthesis systems in Asia, Africa, and elsewhere is driven by rendering a primarily Western technology to fit the requirements of various non-Latin and non-Western languages. While both automatic speech recognition and text to speech engines rely on the phonetic breakdown of speech, many East Asian scripts are idiographic where each letter can have multiple pronunciations. Similarly, India is a multilingual country with more than hundreds of languages and thousands of dialects, which posits different challenges for developing ASR and TTS engines. Teaching natural language processing requires a reflection on the many meanings of language in this domain.
Resources
- Xiaochang Li. ““There’s No Data Like More Data”: Automatic Speech Recognition and the Making of Algorithmic Culture,” Osiris 23 (2023): 165-182.
- Jonathan Sterne and Mehak Sawhney “The Acousmatic Question and the Will to Datify: Otter.ai, Low-Resource Languages, and the Politics of Machine Listening,” Kalfou Vol 9, no. 2 (2022).
- Papa Reo project on data sovereignty and speech recognition in the Māori language
- Alan Turing. ““Computing Machinery and Intelligence.” Mind LIX, no. 236 (October 1950): 433-460.
- Thomas Mullaney. The Chinese Typewriter: A History. Cambridge, MA: MIT Press, 2018.
- Thomas Mullaney. “Typing Is Dead,” In Your Computer Is on Fire. Edited by Thomas Mullaney, Benjamin Peters, Mar Hicks, and Kavita Philip. Cambridge, MA: MIT Press, 2021.