Runosong Geography

✕ Close

Runosong Geography

About

Runosongs — the old Finnic oral songs called regilaul in Estonian and runolaulu in Finnish — were written down at named parishes and regions across Estonia, Finland, Ingria and Karelia. This page maps the songs' vocabulary to 803 collection places holding 276,764 poems. It is part of RunoVerse, a free suite of tools for exploring these corpora.

The poems come from four corpora: ERAB (the Estonian runosong database), SKVR and JR (the Finnish-side collections), and KR (Kirjalliset Runot, runosongs from early written sources, Finnish and Estonian), which records no collection place, so it is missing from the maps and counts of where songs were written down; the "Places named in songs" layer does count its songs. "Finnish (FI)" here names the corpus group, not the local language: SKVR and JR include Karelian, Izhorian and Ingrian places such as Vuokkiniemi, Suistamo and Narvusi. The All / Estonian / Finnish buttons above filter the views to one side; the "Places named in songs" layer keeps places named in that side's songs.

Distinctive words, shown throughout, use TF-IDF scoring: a word common in one place's poems but rare at the other collection places of the same language scores high (Estonian places are scored against Estonian places, Finnish against Finnish). A wordform is a word exactly as spelled in the songs — inflected and dialect spellings count separately; "unique wordforms" counts distinct spellings.

Word lists of a place — A place’s panel can rank its words in three ways. Distinctive: by the TF-IDF score described above. In most poems: by how many of the place’s poems contain the word. Words found in more than half of the collection places of the same language are left out, so the list shows what is sung here rather than everywhere. For this count the collection points abroad together make one place, as the archive files them under one heading. Most concentrated here: by the share of all poems containing the word that were collected here, counting only words found in at least 5 poems here. Rare words reach a high share easily, so the count matters as much as the bar. One collector who worked here a lot can shape the list.

The Wordforms / Lemmas switch: with Lemmas, every spelling is counted under the dictionary word it belongs to, so the inflected and dialect spellings of one word count as one. A word the dictionary does not know keeps its own spelling.

The six view buttons

Places — Sortable table of all 803 collection places; click any row to open that place's detail panel (distinctive vocabulary, song types, links). Above the table, an "Overview & top places" section (collapsed by default) holds the stat tiles with the headline counts, three highlight cards (most poems, largest vocabulary, a random place — new on every visit) and a Top-15 bar chart of the most-collected places; clicking a card or bar opens that place in the table. The table pages through the full list.

Word Search — Type a wordform to see which places use it, as a bar chart.

Compare — Pick two places and compare their distinctive-word lists: shared words and each side's own.

Map — The interactive map; its "Show:" menu offers eighteen colourings, listed below.

Toponyms — Place names found inside the verse texts, including mythological, biblical and foreign names. Auto-detected, so the list is approximate; the lowest-confidence matches (often ordinary words) are hidden by default — switch the confidence filter to "All" to include them.

Semantic — Each place's thematic profile, counted over tagged word occurrences. The category filter offers 24 thematic categories. Click a row for the place's top-10 category chart. Also available as a map colouring.

Map colourings (the "Show:" menu)

Statistics — One of seven per-place measures, defined at the end of this section.

Poem uniqueness — How unlike the rest of the corpus each place's poems are: every poem carries a 0–100 uniqueness percentile, and the map shades the average, median or a quartile of them.

Word frequency — Type wordforms and see where they occur; "Relative" shades by uses per 10,000 words instead of raw counts. With several words separated by commas, their occurrences are added together: each place is coloured by all of them combined. With "Compare two words" ticked, the two words are not added together but set against each other.

Translation frequency — Type an English word; the map shows the places using wordforms translated with it. With several words separated by commas, their occurrences are added together: each place is coloured by all of them combined.

Verse search — Search whole verse lines by text or regular expression, with an exclude pattern and a Diminutives preset; "Relative" shades by the share of matching verse lines.

Collector — Where a named collector wrote poems down. The index stores each collector's top 10 places, so the map shows at most ten places per collector.

Time period — Poems by collection decade, chosen with the two sliders; pre-1800 records are excluded unless ticked.

Emotion vocabulary — Density of emotion words in 11 unified categories, with positive/negative presets. (A finer emotion split — 24 families — lives in the Semantic tab's category filter; the two counts describe different groupings.)

Verse cluster — Where a recurring verse occurs.

Similar poems — Enter a poem ID; the map shows where its most similar poems were collected.

Semantic category — The Semantic profile as map shading.

Dialect area — Finnish Western/Eastern and Estonian North/South dialect groups; "Other languages" covers Karelian, Ingrian, Votic, Sami and more.

Individual dialect — The same map split into single dialects.

Sarv region — Five broad runosong regions after folklorist Mari Sarv.

SKVR volume regions — Which group of the printed SKVR series (one volume per collection region) most of a place's poems are printed in: Ingria (III–V), South Karelia incl. the Karelian Isthmus (XIII), Karelia (I, II, VII), rest of Finland (VI, VIII–XII, XIV). The mixed 1997 supplement (XV) is not counted, places without SKVR poems stay uncoloured, and Finnish communities abroad are shown as diaspora.

Places named in songs — Dots for the places the verses themselves name, from countries down to single villages. Foreign and biblical places are included; mythic otherworlds such as Tuonela and Pohjola are left out. A circle's size shows how many songs name the place. Clicking a dot draws lines to the places where those songs were written down; zoom out to reach the distant places. Most mentions are found by matching each place's checked spellings in the verse text. In ERAB and in Kalevipoeg, further mentions were found by reading single words in their verses: a language model marked a word as a place name, the place was looked up in place-name registers, and where several places share the name, the one nearest another place named in the same song, or else nearest where the song was recorded, was taken. Uncertain cases were read by two further language models, working separately. No folklorist has checked these mentions, so each place's panel lists them apart. Each panel also says where the dot's coordinate comes from. The List of places ranks the places of the kinds ticked in the key by songs, mentions or name. Its song counts include songs whose collection place is unknown. A ≈ before a mention count means that count comes from an older count of poems, or mixes that with the verse scan, rather than from the scan alone. The word-list counts cover songs whose collection place is known, grouped by that place's side. With Estonian or Finnish pressed, mention counts come from the verse scan alone; at 19 places the two sides therefore do not add up to the count shown under All, which rests wholly or partly on an earlier count of poems. The Toponyms tab lists a larger set of place names that were found automatically and not checked, so it also holds some personal names misread as places; this layer shows only checked places.

Villages, communes and other places — One dot per finer place written on the records; thousands of points, with its own legend on the map.

Custom CSV — Upload your own place-and-value list and map it.

The seven Statistics measures: Poem count — poems recorded at the place (a poem filed under several places counts at each). Unique wordforms — distinct spellings occurring in the place's poems. Verse count — lines of poetry. Avg words/line — word tokens divided by verse lines. Collectors (top 3 recorded) — how many of the place's leading collectors the index stores (at most three; not the full collector count). Alliteration rate — the share of verse lines whose content words begin with the same letter; lines where the match is only that two words start with (different) vowels count at reduced weight. Parallelism score — average word overlap between consecutive verse lines (0–1).

Map extras: the Parishes/Counties toggle regroups the map into counties (colour bands are recomputed, so shades are not comparable across the two views); the search box finds and highlights a place; Share copies a link that reopens the current view — the tab, the colouring with its search or filters, and on the map also the position and zoom; Export downloads the places table as CSV.

Data

803 collection places across 276,764 poems, from the SKVR, JR and ERAB corpora; the fourth corpus, KR, records no collection place. Distinctive words computed with TF-IDF scoring against the other collection places of the same language.

The map's "Villages, communes and other places" view breaks the same records down one level finer — to villages, communes and other named spots — for the records that give one; its place count is therefore in the thousands, not 803, and the two totals measure different things.

Keyboard shortcuts

Close with the button, the Esc key, or a click outside this box.