The vocabulary a learner is taught and the vocabulary people speak are drawn from different halves of the language. Formal Hindi reaches for Sanskrit-derived words, and news bulletins and textbooks select them deliberately, sometimes choosing a word a speaker would never use at home. Everyday speech runs largely on the Persian and Arabic derived layer, on native words, and on an enormous quantity of English absorbed whole.
So a learner working through the obvious material acquires a register rather than a language, and the register is the one furthest from an ordinary conversation.
Six words, and the two layers under the vocabulary
- kal is both yesterday and tomorrow. The tense of the verb tells you which, and the word itself means the day next to this one in either direction.
- parso is the day either side of that, so it means both the day before yesterday and the day after tomorrow, on the same principle.
- pani and jal, both water. The first is everyday and the second is Sanskritic and formal, and this pair is the whole register system in miniature.
- shukriya and dhanyavaad, both thank you, the first from Persian and Arabic, the second from Sanskrit. Which one a speaker chooses carries a social signal.
- the Sanskrit and Persian layers run through the entire vocabulary. Formal and official Hindi reaches for Sanskrit; everyday speech and Urdu reach for Persian and Arabic.
- chai, tea, and the word travelled from Chinese through Persian into Hindi and then into English, which is a useful reminder that the borrowing has gone in every direction.
Three layers, and one of them is the one you were taught
The Sanskrit layer is the formal and official one: administration, literature, news, and anything a committee wrote. It is also where Hindi and Urdu diverge, since the same idea reaches for a Sanskrit word on one side and a Persian one on the other, and that divergence happens above the level anybody speaks.
The Persian and Arabic layer arrived over centuries and is thoroughly domestic. These are not foreign words to a Hindi speaker; they are the ordinary words for ordinary things, and they are shared with Urdu completely. The English layer is the third and it is not a garnish: whole categories of everyday vocabulary are English, spoken with Hindi grammar around them, by everybody, in every register below the formal.
What follows is a practical rule that runs against instinct. When two words exist for the same thing and one of them is Sanskrit-derived and formal, the everyday choice is usually the other one, and a learner reaching for the word their textbook taught will be understood and will sound like a newsreader.
Learning the register that the material selects
Here is the stall, and the material causes it rather than the learner. News Hindi is comprehensible, dignified, easy to find and free, so it becomes the listening and reading diet. Textbooks reinforce the same layer because they are written by people who write formally. A year of it produces somebody with a large vocabulary that is systematically one register too high.
The symptom is specific and recognisable: they are complimented on their Hindi and not understood in a shop, or understood and gently laughed at. The exit is a change of material rather than more study. Film dialogue and unscripted conversation sit much closer to the middle, and the words that turn up there are the ones to collect, which is why the listening argument in film Hindi, news Hindi, and the gap between them is a vocabulary argument as much as an ear one.
One idea, several words
Where each layer sits, from a kitchen to a government notice
An English speaker's instinct about formality transfers, because Hindi built its formal register the same way English did, out of a borrowed classical vocabulary.
Placed by where each layer is normally heard, from casual speech to official writing. The placings are this page's judgement.
- English words, with Hindi grammarEverywhere below the formal, and used by everybody
- The Persian and Arabic layerThe ordinary vocabulary, shared completely with Urdu
- Native and everyday wordsSitting beside the layer above rather than beneath it
- Sanskrit derived, in speechMarked, careful, and occasionally comic in a shop
- Sanskrit derived, in printThe register news and textbooks select deliberately
The rail is about words rather than about grammar, which is shared across all of it. It also runs the wrong way for a learner: the free material clusters at the right hand end and the conversations happen at the left.
Two references for a word with two histories
Platts' dictionary, free online through the Digital Dictionaries of South Asia, is the one that tells you where a word came from. It covers Hindi and Urdu together, as one language with two vocabularies, and marks whether a word is Sanskrit, Persian or Arabic in origin. It replaced guessing at register from feel, which is what learners do and which produces the error above. It is old, its English is Victorian, and for this one job nothing has replaced it.
The other is Hindi subtitles on a film you already know. It replaced trying to learn the everyday register from written material, which does not contain it. Subtitles give you the spoken words in the script you are learning to read, with the meaning already known, so the only new thing being practised is the vocabulary itself. Free wherever you already watch, and the useful setting is Devanagari subtitles rather than English ones.
The mistake: treating the English layer as cheating
Learners frequently avoid the English words, on the grounds that using them is not really speaking Hindi. That instinct is wrong twice. It is wrong descriptively, because the English words are what people say, and reaching for a Sanskrit equivalent is the marked choice rather than the neutral one. And it is wrong practically, because it makes a learner harder to understand rather than easier.
The layer to be careful with is the opposite one. A Sanskrit-derived word used in a casual conversation is the error that gets noticed, and there is no equivalent penalty for using the English word everybody uses. The rule that follows is unglamorous: when in doubt in speech, take the word you would guess, and save the formal register for reading.
What the words need to be usable
Every noun here carries a gender, and the two vocabularies distribute their endings differently, so the guessing rules work better on one layer than the other, which is gender on every noun, and postpositions where English puts prepositions. Reading them at all means reading conjuncts on sight, in forty six letters hanging from a line, and how they join. The Persian layer also carries sounds that the Sanskrit layer does not, which is retroflex and aspirated pairs that English does not distinguish. And which words earn a card at all is the general question in making cards that teach rather than cards that pile up.