DiscorraDiscorra

How Discorra Works

From raw text to evidence-backed decisions. Here’s the simple flow that powers Discorra’s discourse analysis—optimized for marketers, researchers, and anyone who needs clarity from language.

How it flows

From Text to Insight in Three Steps

Upload or Import

Bring in corpora, campaign copy, reviews, transcripts, or competitor text. CSV, TXT, and web imports supported.

Analyze

Run built-in analytics—Frequency (Fr), Sentiment (Se), Resonance (Re), Topics (Tp), Concordance (Cc)—no setup required.

Visualize & Share

Turn results into clear dashboards, compare messages side-by-side, and export CSV snapshots or share links.

Under the Hood: The Analysis Pipeline

Discorra turns unstructured language into structured evidence with a balanced pipeline—linguistic tradition + modern ML where it helps.

Pre-processing
  • Language detection & normalization
  • Tokenization & lemmatization
  • Stopword handling
Core Measures
  • Frequency & keyness
  • Sentiment & emotion
  • Concordance & KWIC
Model-Assisted
  • Topic modeling (TP)
  • Resonance scoring
  • Similarity & overlap
PRE-PROCESSING
CORE MEASURES
MODEL-ASSISTED
OUTPUT
Language detection & normalizationClean & standardizeTokenization & lemmatizationBase word formsStopword handlingRemove filler termsFrequency & keynessWhat stands outSentiment & emotionTone & affectConcordance & KWICContext windowsTopic modeling (TP)Unsupervised themesResonance scoringTrust / urgency signalsSimilarity & overlapCompare corporaDashboards & ReportsShareable visuals & exportsPRE-PROCESSINGCORE MEASURESMODEL-ASSISTEDOUTPUT

The analysis pages, explained

Each section blends classic corpus linguistics (frequency, keyness, KWIC) with model-assisted layers (topics, resonance, similarity) to move from patterns → implications → actions. Use these tabs to see how we turn raw text into decisions you can defend.

Overview dashboard screenshot

Overview

High-level scan of frequency highlights, topic mix, sentiment balance, and the Resonance Radar for message pull—perfect to orient stakeholders before deep dives.

  • Keyness with z-scores to flag unusually salient terms.
  • Shared vs unique topics across two corpora at a glance.
  • Micro bars & bubbles for quick volume vs distinctiveness.

Frequently Asked

What formats can I upload?

CSV and TXT are supported out of the box. You can also paste text or import from the web where permitted.

How big can my corpora be?

This depends on your plan. See Pricing for token and project limits; Pro supports substantially larger corpora.

Can I compare two corpora?

Yes—side-by-side comparisons are a core feature. You can assess overlap, resonance, keyness, sentiment and more.

How do I share results?

Export CSVs or share read-only dashboards with your team and stakeholders.