Searches for your query as a substring within normalized verse text (lowercase, no punctuation).
Browse by place: choose a place filter, then search for * to list every verse recorded from that place.
The primary shard (matching query's first letter) loads first for quick results. Remaining shards load in the background.
The Match menu sets how the query is matched: Anywhere (substring, the default), Whole word (the query is a complete word), or Word ending (a word that ends with the query — e.g. kene or kõnõ matches the diminutive suffix on emakene, armakõnõ without catching mid-word hits such as the speech verb kõnõlda).
Regular expression: the pattern is matched per word (a verse matches if any of its words matches). The POSIX word-boundary tokens [[:>:]] (end of word) and [[:<:]] (start of word) are accepted — e.g. .+k[ao]n[ao][[:>:]]. Matching is case-insensitive over normalized text (lowercase, punctuation removed, whitespace collapsed); hyphens and apostrophes are removed by joining the two halves (so counts can differ slightly from a database that treats them as word breaks). Type a NOT pattern in the Exclude box to subtract matches, and combine with the ET language filter to restrict to Estonian (ERAB).
Lemma search: arriving from the Collocates "Lemma view" links searches every inflected form behind each lemma at once (e.g. kasinam matches kasinamad, kasinamp, kasinem), so results match the collocate count. A blue banner lists the forms; typing a new query returns to normal search.
Searching in English
Set Match to English meaning and type one or more English words (e.g. moon people). A verse has to satisfy every word you typed. There are two ways it can:
Through a dictionary word. The page looks up which Finnic words the annotation ever glossed with your English word and shows them as chips. What it then searches for in the verses are spellings — the spellings that were actually glossed that way, plus each selected word's own base form. So a search for moon looks for kuu, kuun, kuuta and hundreds more, not for the base form alone.
Through the translation of the line. Two thirds of the corpus has an English translation of the whole verse line. A verse also counts if your word appears there, even when no single word in it was glossed that way. The note above the results says how many verses were found only like this, and every result says which route found each of your words.
Two controls change what is searched:
Search as dictionary word widens the first route to every recorded spelling of the selected words, not only the ones glossed with your English word. It finds more, but it also brings in occurrences the annotation read differently — forms of kuu that meant 'month' rather than 'moon'. It does not affect the line-translation route.
English from switches either route off, so you can see dictionary evidence and translation evidence separately.
Two things worth knowing about the chips. Only words that mostly mean what you asked for are ticked to start with: at least a fifth of everything the annotation ever said about a word must point at your English word. The rest are one click away under show weaker matches — that is why kuutamo is not ticked for moon, since it means 'moonlight' nine times out of ten. And the search is by whole English word, not by letters: moon does not reach moonlight, afternoon or month. (The Word Explorer's translation filter accepts *moon* and does reach them, which is why the two pages can disagree.)
Untick a chip whenever a word is there for the wrong sense; the search re-runs immediately.
Reading the Results
Green badge: total occurrence count across all poems.
Language bar: blue = Estonian, orange = Finnish.
Places: number of distinct collection places.
Detail Panel
Word links: click any word in the verse title to look it up in Word Explorer.
Geographic bars: click a place name to view its profile.
Occurrence table: every occurrence with poem and network links.
Cross-links: jump to Similarity, Network, Path Finder, Collocates, and more.