Studying Networks of Language Families
Language evolution and how languages are connected to one another are commonly studied by focusing mainly on cognates between the two languages (cognates are words in different languages that have the same linguistic derivations as one another). However, one flaw in this method is that most words have evolved very quickly and are unable to preserve their roots for more than 5,000 to 9,000 years. This puts a time constraint on how far back linguists can trace languages to see when they diverged.
A new method of studying relationships among language families focuses specifically on “ultraconserved” words. These words are not necessarily cognates, but they tend to be words of similar meaning that are used more than once per 1,000 in everyday speech across multiple of the seven language families studied (Altaic, Chukchi-Kamchatkan, Dravidian, Inuit-Yupik, Indo-Eurpoean, Kartvelian, and Uralic). Apart from lengthening the time period over which the relationships between languages can be studied, using ultraconserved words does not limit linguists to studying words that sound phonetically similar (cognates sound phonetically similar). From a networks perspective, studying the relationships among language families based off the cognates they have in common is akin to having several networks with numerous weak ties linking them, whereas studying the relationships among languages families based off the ultraconserved words they have among them is similar to having several networks linked by fewer, much stronger ties.
In my mind, cognates between languages are weak ties because it can’t be said with certainty whether a cognate was a word that two languages had in common before they diverged, or a coincidence between two languages that they happen to have a word with the same meaning pronounced in similar ways. While the latter may sound unlikely, it has been statistically shown that it is difficult to prove whether resemblances between certain languages happened by chance or because the two languages were related (see sources 11 and 12 on the linked paper). Overall, using the methodology described in the linked paper (titled “Ultraconserved words point to deep language ancestry across Eurasia), using ultraconserved words has offered an alternative method of making the construction of language trees more reliable.