Wordform Hub ? Help

What is the Wordform Hub?

The Wordform Hub is a single-page dashboard that aggregates four analytical dimensions for any wordform in the Finnic runosong corpus. Type a word and instantly see how it behaves across collocations, substitutions, morphological similarity, and n-gram patterns — all in one place.

How to search

Enter any wordform (not a lemma) into the search box and press Explore or hit Enter. The search is case-insensitive. Try common runosong words like kulla, neiu, laulu, or venda. You can also arrive here with a ?q=word URL parameter from other pages.

Identity card and quick links

The card at the top shows the searched wordform and provides direct links to related explorers: View in Lexicon opens the main lexicon filtered to that word, Full Collocates and Full Substitutions link to the dedicated explorer pages with richer detail, and Geographic Distribution shows where the word appears on a map.

Morphological Neighbors (BERT)

Shows wordforms that are morphologically similar based on BERT contextual embeddings trained on the runosong corpus. These are not semantic synonyms but words that share similar inflectional or derivational patterns. Higher scores indicate closer morphological resemblance. Click any neighbor to explore it.

Collocates (PMI)

Lists words that co-occur with the searched wordform more often than chance would predict, ranked by Pointwise Mutual Information (PMI). The table shows each collocate word, its PMI score, and co-occurrence count. High PMI values indicate strong formulaic or poetic associations.

Substitution Partners

Displays words that appear in the same metrical position across parallel verses — words that singers could swap in while preserving the verse structure. The count shows how many verse pairs contain this substitution. Click any chip to explore that partner. Example verse pairs are shown below the chips to illustrate the substitution in context.

N-gram Patterns

Shows the most frequent bigrams (2-word sequences) and trigrams (3-word sequences) containing the searched wordform. Patterns marked ALIT exhibit alliteration — a hallmark of runosong poetics where adjacent words share initial sounds. This reveals the formulaic phrases and poetic conventions the word participates in.