False friends and true cousins
Hindi phool and Telugu puvvu both mean flower and look like an obvious match. They are unrelated, and the proof runs through Kannada. A field guide to telling inheritance from borrowing from coincidence.
Every language pair has look-alikes, but Hindi and Telugu have one that deserves a plaque. Hindi phool means flower. Telugu puvvu means flower. Both are short, homely words that begin with p and carry a long vowel. If any pair in this dictionary had earned the right to be cognates, surely this one. It has not. The entry gives the verdict: "The two words look alike and are not related, which is exactly the trap this dictionary exists to catch."
Pull on the two histories and they run in opposite directions. Phool is Indo-Aryan: the dictionary traces it as a worn-down form of Sanskrit pushpa, ground smooth by the same centuries of ordinary speech that turned so much Sanskrit into everyday Hindi. Puvvu is Dravidian: it belongs to a native root for flower, and its relatives are Tamil pu and Kannada huvu. Two family trees, two separate flower words, one accidental resemblance.
But how do we actually know that puvvu was not simply borrowed from some Indo-Aryan flower word along the way? This is where Kannada becomes the star witness. In medieval Kannada, an old Dravidian p at the start of a word regularly became h. Words that were already in the family before that shift show h in Kannada today; words borrowed later kept their p. So an inherited Dravidian word tends to appear with p in Telugu and Tamil but with h in Kannada, and that pattern is a timestamp no borrowing can fake.
Watch the test run. Flower: Telugu puvvu, Tamil pu, Kannada huvu. Ten: Telugu padi, Tamil pattu, Kannada hattu, and the entry for ten spells out the logic: "The Kannada form is the proof that padi is inherited and not borrowed, because old Dravidian p regularly became h in Kannada while later loanwords kept their p." Tiger: Telugu puli, Tamil puli, and Kannada huli, which the entry says "shows the regular change of older p to h that marks a native word rather than a loan." Pig: Telugu pandi, Tamil panri, Kannada handi. Sour: Telugu pullani beside Kannada huli, tamarind and sourness. Five words, one regular correspondence. That regularity is the whole difference between evidence and impression. Relatedness is never two words sounding similar; it is whole families of words changing in step.
The same discipline has to cut the other way, and this dictionary keeps two famous temptations permanently on ice. Hindi kutta and Telugu kukka both mean dog and practically beg to be connected. The entry refuses: the expected Dravidian dog words are the Tamil and Kannada nay type, which kukka does not match, and a shared source in Sanskrit kukkura has been proposed for both sides. "Neither derivation is settled, so this entry claims no relationship in either direction." Hindi billi and Telugu pilli, cat, get the same treatment: billi has a Sanskrit source in bidaala, but pilli matches neither the Tamil poonai type nor the Kannada bekku type, and "the billi and pilli resemblance is unexplained", with borrowing in one direction possible. An honest unknown beats a tidy story, every time, and especially here.
So what do true cousins look like? Inside a family, they look like naam and name, Hindi and English meeting through their shared Indo-European ancestor, or like daant beside Latin dens, or like kannu and neeru sitting among their Tamil and Kannada kin. Across the two families there are no cousins in that sense at all. What exists instead are shared words with a traceable lender: pustak and pustakam both taken from Sanskrit, kharch and kharchu both taken from Arabic through Persian, bas and bassu both taken from English. The same word met twice, by two unrelated languages standing in the same weather. Why Telugu sounds so Sanskrit and The Deccan doorway walk through the two oldest lenders in detail.
Out of all this falls a small field guide, the one this project actually uses. First, look for family: does the word have relatives inside its own family, connected by regular sound correspondences, the way the Kannada h stamps inherited Dravidian words? Second, look for a route: is there a documented channel for borrowing, a Sanskrit register, a Persian court, an English classroom, and a shape that fits the channel, like the -u on kharchu? Third, respect coincidence: languages build thousands of words from a few dozen sounds, so accidental look-alikes are not merely possible but guaranteed. Fourth, when none of that settles it, say so and stop. Of the 5,000 pairs currently in Bhashalu, 3,638 are marked unrelated, and 1,814 of those carry an origin the dictionary marks uncertain on one side or both. Every one of those labels is a small act of restraint.
Puvvu and phool can share a vase. They cannot share a family tree, and no amount of squinting will graft one onto the other. The good news is that the truth is better than the illusion. A false cognate is a dead end, but a borrowed word with a real route behind it carries actual history, courts and classrooms and coastlines, into everyday speech. This dictionary would rather hand you one true story than ten plausible ones. Corrections and counterexamples are welcome, and Telugu speakers doubly so.
