VEJE
Slovenian Dialect Treebank Demo
- —
- Documents
- —
- Sentences
- —
- Tokens
One whole word, or two whole words at a chosen distance. Matches the original word and the normalized word. Tick “can be a part of a word” to allow a match inside a longer word.
Metadata
Applies to Collocation, Features, CQL, and tree search. Leave a field on Any to ignore it.
How keyword / collocation search works
Each box is one whole word. A hit must be that word on the original line or on the normalized line, not a piece of a longer word and not a lemma that merely contains the letters.
- One term — sentences that contain that word.
-
Two terms —
both words in the same sentence. The distance field sets how many tokens may sit between them:
2is exact,0–5is a range, and an empty field allows any distance in the sentence. -
Exact surface / richer morphology — use the Features tab
(
formis whole-token equality there). - Ordered patterns / gaps / regex on feats — use the CQL tab.
Pick a mode, run a search — or open Assistant to generate CQL and patterns from a short description.