Looking at a non-Indo-European set of examples, we can imagine receiving the following data from a linguist recently returned from an expedition to a remote region of the Amazon. The examples are a set of cognates from three related languages, but what would the proto-forms have looked like?

Using the majority principle, we can suggest that the older forms will most likely be based on language 2 or language 3. If this is correct, then the consonant changes must have been [p] → [b], [t] → [d] and [k] → [ɡ] in order to produce the later forms in language 1. There is a pattern in these changes that follows one part of the “most natural development principle,” i.e. voiceless sounds become voiced between vowels. So, the words in languages 2 and 3 must be older forms than those in language 1.

Which of the two lists, 2 or 3, contains the older forms? Remembering one other “most natural development” type of sound change (i.e. final vowels often disappear), we can propose that the words in language 3 have consistently lost the final vowels still present in the words of language 2. Our best guess, then, is that the forms listed for language 2 are closest to what must have been the original proto-forms.