Every nominal paradigm in this series — every declension table on page 5, every stem this whole series has ever declined — comes from one file, shabda/index.json, and every entry in it carries not just a set of case-forms but a gloss: C.gloss(stem, liṅga) returns the corpus's own paraphrase, in Sanskrit and, where the corpus supplies one, in English. Nothing about that changes when the goal changes from "prove a rule" to "build a vocabulary" — the same function, searched by meaning instead of picked by pattern, is what built every list on this page.
Analysis, then use
प्रयोगःA reader who has followed this series from page 1 can already parse why देवेन means "by a god" and why राज्ञः is ambiguous between two cases. What that reader may still not have is a plain answer to "how do I say tree, or I want water, or how much does this cost?" — because every earlier page picked its words to illustrate a rule, not to furnish a vocabulary. This page inverts that: it picks a dozen everyday settings and reports which real words the corpus actually has for each one.
The corpus behind this series is closer to an अमरकोश-style synonym dictionary than to a modern bilingual one: each entry is a cluster of Sanskrit words sharing a sense, harvested from classical verse. That makes it excellent for concrete, ancient things — every tree, animal, body part and season below is genuinely attested — and it makes it patchy for anything institutional or modern: पुस्तकम् "book," विद्यालयः "school" and व्याकरणम् "grammar" — a word this whole site is arguably about — are simply not headwords in it. That is not this page inventing a gap to be honest about; it is what the search actually returns, reported as found.
When the corpus has no word, build one
Page 9 · समासःSanskrit's usual answer to a missing word is not to borrow one — it is to weld two existing words together. Page 9 already worked out the mechanism: two finished padas lose their own case-endings and become a single new prātipadika (2.4.71 ↗), and — the fact this page leans on — that new word is declined exactly as its last member would be. So a learner who knows page 9 does not need पुस्तकालयः to be a headword anywhere; they need only आलयः to be a real, declinable word, and the rest follows.
Pick a setting
InteractiveEvery word below carries its gender and its stem-family, because a vocabulary list without either is close to useless — you cannot put a word in a sentence if you do not know how it declines. Click through to page 5 for the family's full paradigm, or straight to the corpus's own live declension tool for this exact word. Coined words are marked plainly and never carry a form this page did not pull.
Words at work
वाक्यानिOne sentence per setting, built the way page 5's were: every noun form below is pulled straight from that word's paradigm, cell by cell — never typed — and shown at its citation, pre-sandhi boundary, exactly as page 5's words were (page 3 covers what actually happens where two words meet in speech). The verbs are ordinary finite forms, typed directly, exactly as page 5's ददाति and वदति were; this page's whole contribution is which nouns to put in which cases, not the verb conjugation itself, which pages 11–13 already cover. Coined words are never bent into a sentence — an unattested stem has no paradigm to pull a form from, and this page does not guess one.
Addressing someone — the vocative, in passing
Page 5 covered the vocative as a reduced प्रथमा, not a real eighth case. Here it is simply useful: four of this page's most common family and school words, called out to directly.
Every word on this page
Coverage| Setting | Word | Gender | Meaning | Status |
|---|
Check yourself
अभ्यासःSix questions. Every answer names the fact that settles it.