Two languages that never shared an ancestor, and trade words anyway.
Bhashegalu, ಭಾಷೆಗಳು, means languages. Hindi is Indo-Aryan and descends from Sanskrit; Kannada is Dravidian and does not. Follow either family tree back as far as the evidence reaches and it never meets the other. What the two share is not ancestry but contact: a thousand years of living next door, through Sanskrit from above, through Persian and Arabic, and now through English. Bhashegalu is a map of that traffic. Which words crossed, which direction they went, and which resemblances are accidents.
The Persian chapter of that story is usually told about Delhi. The other half happened here. The Bahmani sultans ruled from Kalaburagi and then Bidar, the Adil Shahis from Bijapur, now Vijayapura, and all three cities are in Karnataka. Dakhni is still spoken across the north of the state. So Kannada did not receive its Perso-Arabic vocabulary secondhand from the north. It got it from the next street, and then naturalized it with two rules worth learning on day one: a borrowing that ends in a consonant picks up a final -u, so kharch becomes ಖರ್ಚು kharchu; and a Sanskrit word ending in -aa ends in -e, so bhaasha becomes ಭಾಷೆ bhaashe, which is where this app got its name.
The app
Every entry is a pair, not a translation
A dictionary built the honest way. Each entry holds a Hindi side and a Kannada side, each with its own script, romanization, origin and gender, plus a plain statement of how the two relate. Cousins through a shared Sanskrit parent, the same Persian loan taken twice at opposite ends of the country, or simply two unrelated words that mean the same thing. Instant, offline, and it says so when scholars disagree.
Kannada gender, modeled properly
Kannada does not assign gender the way Hindi does. Hindi decides by convention, so a table is feminine and a door is masculine and you memorize both. Kannada decides by meaning: men are masculine, women are feminine, and everything else is simply neuter, apart from a few personified exceptions such as the sun and the moon. For a Hindi speaker that is a whole category of homework that disappears. In the plural the split moves: masculine and feminine merge into the rational avaru, and neuter alone stands apart as avu. It is also a different split from Telugu’s, which groups she with it in the singular where Kannada gives her a pronoun of her own.
Ask Bhashegalu anything
Beyond the curated set, type any Hindi or Kannada word and the app looks it up with the same full treatment, then keeps your lookups on the device so an answer you liked is still there next week. The bundled dictionary works with the phone in airplane mode; the lookup is a fallback for what is not in it yet, never a dependency for what is.
Two scripts, one chart
The alphabets are relatives even though the languages are not. Devanagari and the Kannada script both descend from Brahmi, which is why the charts line up row for row, and the Kannada script has a sister closer still in Telugu. The app puts the two side by side, letter against letter, marks the rows where they stop corresponding, and includes a practice mode for learning to read either one.
A word every day, and the essays behind the words
A word of the day drawn from the pair set, with an optional morning notification. And for the bigger picture, a set of essays on how two unrelated languages came to share so much vocabulary. The same essays are published here in full.
A reading guide that marks what the spelling hides
Everyday romanization cannot show everything. Kannada ಬೆಳೆ (a crop) and ಬೇಳೆ (a split pulse) are different words that both come out bele. So every Kannada entry carries a second reading, derived mechanically from the script itself. Long vowels double (aa, ee, oo), and a capital marks the two things doubling cannot: a long e or o, and retroflexion. The pair comes out beLe against bELe.
No account, no ads, no tracking
The dictionary is bundled in the app and works entirely offline. There is nothing to sign up for, nothing collected, and nothing sent anywhere. Dark mode throughout. It is built for iPhone, iPad, Android, and Mac.
What an entry looks like
Every pair carries both scripts, both romanizations, both origins, and one plain sentence about how the two words are related. The relation label and the origin colors are the same ones the apps use, so a chip means the same thing here as it does on your phone.
Same meaning, entirely different words, from entirely different sources. Hindi reached for Arabic, Kannada for Sanskrit. This is the ordinary case between these two languages, and saying so plainly is more useful than hunting for a connection that is not there.
Here the two do match, and the reason matters: neither borrowed from the other. Both inherited the same Sanskrit word independently. In Hindi it is the formal register beside everyday kitaab; in Kannada pustaka is simply the word for book, and it keeps the final a that Hindi drops.
The app is named for this pair. One Sanskrit word, inherited separately by an Indo-Aryan language and a Dravidian one. Kannada handles the Sanskrit ending the way it always does: a final aa becomes e, so bhaasha arrives as bhaashe. That swap alone does not make the word a tadbhava, because the aspirated bh and the retroflex sh are both intact; the worn down baase is the real tadbhava.
A clean tatsama pair, taken essentially unchanged into both languages. Hindi drops the final short a and says desh; Kannada pronounces it and says desha. That one syllable is the most reliable difference between the Hindi and the Kannada version of a shared Sanskrit word.
Another Sanskrit word held in common, with the same final vowel difference: Hindi samay against Kannada samaya. Hindi deletes the inherited short a at the end of a word and Kannada keeps it, which is why so many shared words sound one beat longer in Kannada.
Both languages took guru straight from Sanskrit and left it alone, and the word has since traveled far past both of them into English. The gender is worth noticing: Kannada calls a guru masculine because he is a man, not because of how the word ends, while every table and door in the language is simply neuter. For a schoolteacher a Kannada speaker is likelier to say shikshaka, or the borrowed meshtru.
How words actually crossed
- Sanskrit, from above. Sanskrit stands in a different relationship to each language. For Hindi it is the ancestor; for Kannada it is a donor, roughly as Latin is for English. So when Hindi says pustak and Kannada says pustaka, neither borrowed from the other: both reached for the same Sanskrit word independently. Kannada’s own grammarians sorted that layer into named categories in the thirteenth century, tatsama for a word taken whole, tadbhava for one reshaped on the way in, and achchagannada for native stock. Cousins through a shared parent, not borrowings from each other.
- Persian and Arabic, through courts in Karnataka. The Bahmani sultanate ruled from Kalaburagi and then Bidar; the Adil Shahis held Bijapur, now Vijayapura, until 1686. Persian ran their revenue offices and Dakhni filled their streets, and it is still spoken across North Karnataka. That layer is administrative and commercial almost word for word: sarkaara, vakeela, kaagada, bajaaru.
- English, in the last two centuries. Both languages borrowed heavily, often the same word for the same new thing, and Kannada put its own ending on it: bus arrives as ಬಸ್ಸು bassu and car as ಕಾರು kaaru. It is the one layer nobody finds surprising.
Two rules that hide a loanword. A borrowing that ends in a consonant usually picks up a final -u: kharch becomes ಖರ್ಚು kharchu, jawaab becomes ಜವಾಬು javaabu. Note which way that rule runs. The -u is worn by native Dravidian words too, by ನೀರು neeru and ಹಾಲು haalu, so it never betrays a loan. It naturalizes one. The second rule catches the Sanskrit layer: a word ending in long -aa ends in -e in Kannada, so maalaa is ಮಾಲೆ maale, kathaa is ಕಥೆ kathe, and shaalaa is ಶಾಲೆ shaale. Learn those two and hundreds of words stop looking foreign.
Same skeleton, different handwriting
Both scripts descend from Brahmi, so they are organized by the same phonetic grid and mostly line up letter for letter. The rows outlined in saffron are the ones where they stop corresponding: the five letters Hindi built with a nuqta for Perso-Arabic sounds and the two more it built the same way for its own retroflex flaps, the retroflex ಳ Devanagari never needed, the two letters Kannada has retired, and the short e and o that Kannada writes and Hindi does not distinguish. Twelve letter rows in all; the vowel-sign and conjunct sections further down wear the same saffron wherever those same gaps come round again.
Vowels11 of 13 line up
Marks2 of 2 line up
Consonants33 of 43 line up
Vowel signs12 of 14 line up
Conjuncts (ottu)8 of 9 line up
Rows where one script has no counterpart. The full chart, with a note on every row and a practice mode, is in the app.
From the articles
Unrelated languages, familiar words
Hindi is Indo-Aryan. Kannada is Dravidian. No family tree joins them, and yet a Hindi speaker starting Kannada keeps meeting words they already know. Both facts are true, and this whole project lives in the gap between them.
The grammarian who named the layers
Kannada never descended from Sanskrit, but its formal register is saturated with it, and Kannada’s own grammarians sorted the layers in the thirteenth century. On tatsama, tadbhava, achchagannada, and the vowel that ends almost every borrowed noun.
Persian on Kannada ground
Delhi is the famous half of the Persian story. The other half happened in Kalaburagi, Bidar, and Vijayapura, which is why Kannada settles a bill in kharchu and waits for a javaabu. Also: the vowel that hides a loanword.
Same skeleton, different handwriting
Set the Kannada and Devanagari charts side by side and they match row for row, because both descend from Brahmi. Then look closer: Kannada stacks its clusters downward, and keeps two vowel letters Hindi never needed.
When p became h
Kannada milk is haalu and Tamil milk is paal. That is not a coincidence, it is a regular sound change, and it is the best tool anyone has for telling an inherited Kannada word from a borrowed one, or from an accident.
All articles →Get the app
Bhashegalu is built for iPhone, iPad, Android, and Mac. The Android build is ready to download now. The iOS version has been submitted to the App Store and is waiting for review; this section will say otherwise the day that changes rather than a day before it.
SHA-256 of the file above:e5ec3a8670b5c6caf0fef2fc455b6627d9144d088e0d14a7fb1581c213e565a7
Your phone will ask you to allow installs from your browser, because this does not come from the
Play Store. Verify the download with shasum -a 256 Bhashegalu.apk and check it
matches the hash printed here.
On accuracy
A wrong etymology is far worse than an honest “scholars disagree”. Where a word’s origin is genuinely unsettled, Bhashegalu says so rather than inventing a tidy story, and popular folk etymologies are debunked rather than repeated.
The harder discipline is refusing a pattern that is not there. Kannada ಮೀನು meenu is a fish, and Sanskrit has meena, the word behind Hindi meen. A Dravidian source for the Sanskrit word has been proposed, the resemblance has been argued over for more than a century, and no account is agreed, so the entry says that instead of picking whichever story reads better. The reverse trap is just as common: Kannada ಹಾಲು haalu is milk and Hindi haal is a condition or state, and the two have nothing whatever to do with each other.
About the author
Nipun Dubey builds things at the intersection of technology and language. By day he is the CEO & Co-Founder of HappiNest.AI, an AI leasing copilot that helps property managers significantly reduce costs and improve operations for all things leasing. Before HappiNest he led AI and B2B startups bringing transformational change to legacy industries, worked at leading technology companies, and is a graduate of the Wharton School of the University of Pennsylvania.
Bhashegalu comes out of the other thread, and the older one: a lifelong fascination with languages and linguistics, history, and cultural anthropology, and a conviction that showing people how much they already share is the fastest way to connect them. It runs in the family. His grandparents grew up in pre-Partition colonial India, and one grandfather, a renowned professor of Hindi and Indian languages, spent his life teaching the very words this project now traces.
Bhashegalu is the third of a set. Zubane covers Hindi and Urdu and argues that two things people think are separate are really one language. Bhashalu covers Hindi and Telugu and argues the opposite and more interesting case: that two genuinely unrelated languages can still hand each other a thousand words. Bhashegalu makes that second argument on Kannada ground, where the Deccan story is not a distant frontier but the local history.
Get in touch
hello@bhashegalu.com. Corrections and counterexamples especially welcome, and from Kannada speakers doubly so.
