Linguistic historians have long used the comparative method to reconstruct ancient languages not attested in written sources. It consists of a detailed comparison of words in related languages of descendants and allows linguists to draw conclusions about the ancient pronunciation of words that have never been recorded.
Scientists are now using the same approach to predict how an undocumented word will sound in a particular language. To do this, they need basic information about the language and related words that have been preserved in written sources.
Two researchers from the University of London described the results of an experiment in which they applied a traditional comparative method to predict the pronunciation of words in eight Western linguistic varieties of kho-bwa, spoken in India. Belonging to the trans-Himalayan family, these species have not yet been described by researchers.
Researchers have come up with a process whereby an algorithm helps predict missing verbal forms. Classical methods involve manual decryption, and new computational solutions have helped scientists to improve the efficiency and reliability of the process, and all results were subsequently manually verified and refined. To increase the transparency and credibility of the experiment, they then logged their predictions online.
“Registration is incredibly important in many scientific fields because it ensures that researchers adhere to a good scientific base. As far as we know, this has not been done in science yet, ”the researchers note.