A complete corpus workbench in the browser
Upload your texts and work through them the way you actually would — frequencies and concordances, collocations and topics, sentiment and AI-written analysis — without moving between half a dozen tools or writing a line of code.
- languages
- 23languages
- file formats
- 7file formats
- AI, on our own servers
- Self-hostedAI, on our own servers
Everything in one analysis environment
From raw files to a finished argument — the tools stay in one place and share the same corpus, so nothing has to be exported and re-imported halfway through.
Describe
Understand the shape of your corpus before you form a question.
Token and type counts, vocabulary richness, sentence-length distributions, keywords, named entities, emojis and hashtags — computed on upload.
Frequency lists by word form or lemma, with dispersion across the corpus so you can tell genuinely widespread items from a few dense pockets.
Recurrent multi-word sequences with frequency and range, for the fixed expressions and phraseology that single-word lists miss.
Find
Get to the passages that matter, by form or by meaning.
KWIC concordances with a visual query builder, or search by meaning using BGE-M3 embeddings. Describe the pattern in plain language and let the AI write the query for you.
Statistically significant co-occurrences using mutual information, t-score and log-likelihood, with adjustable windows and filters.
BERTopic over sentence embeddings, so topics come from semantic clustering rather than bag-of-words counts. Relabel, split and merge topics as you refine them.
Interpret
Turn measurements into something you can write about.
Segment-level polarity, plus emotion scoring across the six Ekman categories and neutral, tracked over the corpus so you can see how tone develops through a document or along a timeline.
Fifteen analytical angles — summary, register, rhetorical style and devices, discourse markers, subjectivity, opinions, irony and more — written up as prose you can export.
Partition by metadata or by hand, then compare two subcorpora — or two whole corpora — across statistics, topics and affect, side by side.
Ask fifteen different questions of the same corpus
Choose an angle and a scope — the whole corpus, a subcorpus, selected files, or the results of a search — and get a written analysis grounded in the actual text, with the passages it drew on. Export it as Markdown or PDF.
The language model runs on our own infrastructure. Your texts are never sent to a third-party AI provider.
- Summary
- Readability
- Register
- Rhetorical style
- Rhetorical devices
- Discourse markers
- Subjectivity
- Sentiment
- Opinions
- Emotions
- Sarcasm
- Humour
- Sexism
- Hate speech
- Profanity
How it works
From upload to interpretation in three steps.
Upload your texts
Plain text, CSV/TSV, XML, HTML, PDF or Word — individually or as a ZIP. Metadata in your files becomes filters you can slice by later.
Let it process
Tokenising, tagging, parsing and embedding run in the background. You keep working; results appear as each stage finishes.
Explore & export
Move between tools without losing your place, compare subcorpora, and export findings as Markdown or PDF for your write-up.
Scale with your research
Every plan includes the full analysis suite. The limit is corpus size, not features.
Need more, or an institutional arrangement? Get in touch.
Ready to analyse your corpus?
Create a free account and upload your first corpus in minutes.