Search the bank
Filter the ready-made pseudowords. Click a pseudoword to swap it for another one.
Which letter model produced the item. Plain trigrams follow the two previous letters; the Kneser-Ney model follows three and smooths unseen sequences.
The frequency band of the real words the generator learned from.
Form
Length (letters)
C consonant · V vowel · ? either · * any run
Real words
2 or more drops items one letter away from a real word.
Exclude items with a real word inside
Word of at leastletters
Words count from Zipf
Avoids items that look like a real word plus letters.
Exclude real words of other languages
None selected
Highlights them; a button above the table removes them.
Measures
OLD20 range
Mean distance to the 20 closest real words.
Coltheart's N range
Real words one letter away.
Wordlikeness range (percentile of real words)
Higher = more typical letter sequences.
Count trigrams by
Wordlikeness only, in all tabs.
Results
Choose a language to begin.
New pseudowords
Upper-case C/V are slots; lower-case letters are fixed. Leave empty to use a length range.
Length range (when no template)
Zipf range of training words
Embedded real words
of at leastletters
Generated items appear here, scored against the reference lexicon.
Match my words
One pseudoword per word. Click a result to swap it.
Your words
Text, CSV/TSV or Excel. Other columns are kept.
Match on
Ticked measures must be within ± the tolerance.
Real words
At 2+, short words are harder to match.
Exclude items with a real word inside
Word of at leastletters
Words count from Zipf
Avoids items that look like a real word plus letters.
Exclude real words of other languages
None selected
Highlights them; a button replaces them.
Paste or upload a word list to get one matched pseudoword per word.
Analyze words
CV, real-word status, IPA, neighbourhood and wordlikeness for any items.
Your items
Text, CSV/TSV or Excel.
Real words
Exclude real words of other languages
None selected
Highlights them; a button removes them.
Results appear here.
Cross-language sets
Matched pseudowords for 2–8 languages at once.
Languages (2–8)
Only items generated from all reference words are used (no frequency bands), so every language is treated alike.
Ranges
Length range
Measures: target ± tolerance
Untick a measure to ignore it.
Seed items
Measured against the seed language's lexicon.
Text, CSV/TSV or Excel; replaces the box above.
Matches in the other languages
Max. length difference
Closest item within tolerance of the seed.
Exclude
Min. distance from real words
Word lists as set in the other tabs.
Output
Same seed + same settings = same sets.
Click a pseudoword to choose a different one for that language. In the yoked design, choosing a new seed item re-matches the other languages of that set.
Choose two or more languages, press “Check candidates” to see how many items fit, then “Build sets”.