Proto-Indo-European is the key to understanding why words like "mother" and "father" are so similar in so many different languages.
Reconstructing Ancient Words
Let’s focus just on the word for “mother” at this point. We can be sure that the ancestral word began with M, because all the modern equivalents do. We can also be sure that the ancestral word had an R in it, because three out of the four modern equivalents do: the words in Spanish, Italian, and French. Next, we can be pretty confident that the ancestral word had a D between the M and the R, since the Spanish and Italian words do. You might be wondering at this point how we can be so sure, since the other two languages, Portuguese and French, don’t have a D in their words for “mother.” The reason is that certain kinds of sound changes are more likely to happen over time. For a D-R cluster to lose the D is a simplification, a natural kind of change. On the other hand, to insert a D before an R is a complication, and a less-common kind of change. So in the absence of further evidence, we conclude that the ancestral word for “mother” for these languages had a D-R cluster, and that two of those languages simplified it.
Now, what about the vowels? The first vowel was most likely A, since that’s what we have in Portuguese, Spanish, and Italian. The second vowel was most likely E, since that’s the one that appears more than any other. So the ancestral word was probably similar to the modern Spanish and Italian words: “madre.” Similar reasoning lets us conclude that the ancestral word for “father” was “padre.”
Of course, we don’t have to go to all that trouble, because we know from written history that all the places in Europe where these languages were originally spoken were part of the Roman Empire, and that Latin was the language spoken across the Empire. That’s actually why they are called Romance languages--because they were spoken in the Roman Empire. In fact, we know from written documents that the Latin word for “mother” actually had a T in the middle instead of a D--it’s “māter,” remember?--but now we can conclude that somewhere along the way from Latin to its daughter languages, that T turned into a D.
However, the surprising thing about the techniques for reconstructing ancestral languages is that they can work even when we don’t have written records of the ancestral language. How is that even possible?
How Do We Know Our Reconstructed Words Are Real?
It’s one thing to reconstruct ancient words when you have written texts that can confirm your reconstructions, as Latin texts can do for the Romance languages. But how can linguists know that their reconstructed words in Proto-Indo-European are anything close to reality? Well, in many cases, they disagree. For example, they still haven’t agreed on a precise reconstruction of the Proto-Indo-European word for “horse,” even though it’s clear that Proto-Indo-European had such a word. However, a striking and unexpected discovery in the early 20th century gives linguists confidence in the method. (1)
In 1879, the linguist Ferdinand de Saussure was reconstructing a set of verbs in Proto-Indo-European that formed different tenses by using a process called ablaut. We actually have ablaut in English, with verbs that change their vowels such as “sing-sang-sung,” and “write-wrote-written.” Anyway, de Saussure found that he could collapse two families of Proto-Indo-European ablaut verbs into a single family if he assumed that there used to be a consonant in a certain position in some of these verbs. Unfortunately, whatever this consonant was, it did not exist in any of the known Indo-European languages. His analysis didn’t go over well at all, especially since he actually proposed not just one hypothetical, mystery consonant, but two of them! Some reconstructed verbs had one, and some had the other. If he could do that in his reconstruction, what couldn’t he propose?
Had de Saussure not died in 1913, he might have had the last laugh. Two years after his death, Bedrich Hrozny [sorry, I have no idea how to pronounce this name] deciphered the writing system for Hittite, the language spoken almost four millennia ago in what is now Turkey, and it turned out that Hittite was an Indo-European language--nearly the oldest one to have been written! (2) Moreover, it was later shown that Hittite verbs had consonants exactly where de Saussure had put them, confirming his outlandish reconstructions. (3) We still don’t know exactly what those consonants sounded like, though.