Phonetic Corpora

Technological advancements in recording, storage, and processing have reshaped the study of spoken language data over the past decades. Phonetic corpora are emerging as a key method to explore and empirically validate assumptions about spoken language, be it the acoustic features of individual sounds or various properties of speech unfolding over the course of a conversation. Phonetic corpora are systematic collections of phonetic data, typically speech recordings accompanied by basic annotations and metadata. There exists a plethora of subtypes of phonetic corpora, which may differ substantially with regard to their size, annotation depth, balance, technical specifications, and accessibility. Some phonetic corpora come with powerful web applications that allow for sophisticated queries and visualization of data, while others let users download files to further process using state-of-the-art phonetics and statistical software. Due to their breadth and scope, phonetic corpora can complement the more controlled methods of laboratory experiments. This is especially useful in the case of endangered and minority languages for which participants cannot always be easily recruited. Likewise, the use of multilingual corpora becomes increasingly important to ensure that results are generalizable beyond individual languages or speaker populations. Phonetic corpora thus contribute to two crucial principles in the social sciences. The first is the FAIR principles of scientific research (Findable, Accessible, Interoperable, and Reusable) guiding proper data handling, which is paramount for open and transparent science. The other is the shift toward recognizing diversity in linguistics, challenging the dominance of WEIRD (Western, Educated, Industrialized, Rich, and Democratic) samples.