Text Analysis for All

An R 'shiny' app designed for diverse text analysis tasks, offering a wide range of methodologies tailored to Natural Language Processing (NLP) needs. It is a versatile, general-purpose tool for analyzing textual data. 'tall' features a comprehensive workflow, including data cleaning, preprocessing, statistical analysis, and visualization, all integrated for effective text analysis.

TALL - Text Analysis for ALL

TALL (Text Analysis for ALL) is an interactive R Shiny application designed for exploring, modeling, and visualizing textual data. It provides a comprehensive, code-free environment for Natural Language Processing, enabling researchers without extensive programming skills to perform sophisticated text analyses through an intuitive graphical interface.

TALL integrates state-of-the-art NLP techniques — tokenization, lemmatization, Part-of-Speech tagging, dependency parsing, topic modeling, sentiment analysis, and more — into a unified, reproducible workflow.

Reference Paper

Aria, M., Spano, M., D’Aniello, L., Cuccurullo, C., & Misuraca, M. (2026). TALL: Text analysis for all — an interactive R-shiny application for exploring, modeling, and visualizing textual data. SoftwareX, 34, 102590.

Read the full paper (Open Access) | Supplementary material

⚠️ Citation policy. TALL is open source and distributed under the MIT license. However, whenever results obtained with TALL are used in a publication, proper citation of the reference paper above is mandatory. Failure to properly cite the software is considered a violation of the license.

Setup

System Requirements

Before installing TALL, ensure you have:

1. R version 4.2.0 or higher — Download from CRAN
2. RStudio (recommended) — Available at Posit
3. Active internet connection for downloads and dependencies
4. Additional tools for the development version:
   • Windows: Rtools
   • Mac: Xcode Command Line Tools

Stable Version (CRAN)

install.packages("tall")
library(tall)
tall()

Development Version (GitHub)

First, verify your build tools:

if (!require("pak", quietly = TRUE)) install.packages("pak")
pkgbuild::check_build_tools(debug = TRUE)

Then install from GitHub:

if (!require("remotes", quietly = TRUE)) install.packages("remotes")
remotes::install_github("massimoaria/tall")
library(tall)
tall()

The development version includes the latest features but may contain occasional bugs.

For detailed installation instructions, visit: Download & Install

An interactive tutorial is also available: View tutorial

Overview

Researchers across disciplines face the challenge of analyzing large volumes of textual data — research articles, social media posts, customer reviews, survey responses, legal documents, and literary works. While programming languages such as R and Python offer powerful NLP capabilities, not all researchers have the time or expertise to use them effectively.

TALL bridges this gap by providing a general-purpose, code-free text analysis platform built on the R ecosystem. It combines the statistical rigor of established R packages with the accessibility of a modern web interface, enabling researchers to conduct reproducible analyses from import through visualization without writing a single line of code.

Workflow

TALL follows a structured analytical workflow that guides users from raw text to interpretable results:

The workflow consists of three main stages:

1. Import and Manipulation

TALL supports multiple input formats (plain text, CSV, Excel, PDF, Biblioshiny exports) and provides tools for corpus splitting, random sampling, and integration of external metadata. Analysis sessions can be saved and reloaded as .tall files for full reproducibility.

2. Pre-processing and Cleaning

Linguistic pre-processing is powered by UDPipe with updated Universal Dependencies v2.15 language models, supporting 60+ languages. The pre-processing pipeline includes:

• Tokenization — splitting raw text into words and sentences
• Lemmatization — reducing words to their dictionary base form
• Part-of-Speech tagging — assigning grammatical categories (noun, verb, adjective, etc.)
• Dependency parsing — identifying syntactic relationships between words
• Special entity detection — tagging emails, URLs, hashtags, mentions, emojis, and IP addresses
• Multi-word expression extraction — six methods including syntactic dependency parsing, RAKE, PMI, Mutual Dependency, Log-Frequency Biased MD, and IS Index
• Custom term lists and synonym merging — domain-specific vocabulary management

3. Statistical Text Analysis and Dynamic Visualization

TALL offers a rich set of analytical methods organized across three sections: Overview, Words, and Documents.

Analytical Methods

Overview

Corpus-level descriptive statistics provide a quantitative profile of the text collection:

• Corpus metrics — document count, tokens, types, lemmas, sentences, with averages and standard deviations
• Lexical richness indices — Type-Token Ratio (TTR), Hapax Legomena, Guiraud Index, Yule’s K
• Stylistic indices — Lexical Density, Nominal Ratio, Gini Index
• Frequency distributions — word frequency, TF-IDF rankings, Zipf’s law visualization, word clouds
• Morphological features — distribution of grammatical features (Tense, Mood, Number, Person, VerbForm, Degree, Gender, Case, Voice) extracted from Universal Dependencies annotation
• Dependency tree viewer — interactive visualization of syntactic parse trees for individual sentences, with color-coded Part-of-Speech tags and labeled dependency arcs

Words

Word-level analyses reveal the internal structure and thematic organization of the corpus:

• Keyness analysis — statistical comparison of word frequencies across groups using Chi-squared and Log-Likelihood Ratio tests, with frequency-context plots and word clouds
• Words in Context (KWIC) — concordance analysis with customizable context windows, co-occurrence network visualization, and distribution insights
• Correspondence Analysis — factorial mapping of word-document associations with hierarchical clustering and dimensional interpretation
• Co-occurrence Network — weighted word co-occurrence networks with multiple normalization indices (Association, Jaccard, Inclusion, Salton, Equivalence), Louvain community detection, and adjustable community repulsion for visual cluster separation
• Thematic Map — strategic diagram positioning topics along Callon’s Centrality and Density dimensions (motor themes, basic themes, niche themes, emerging/declining themes)
• Word Embeddings — Word2Vec model training with similarity network analysis, UMAP dimensionality reduction, and cluster-based semantic exploration

Documents

Document-level analyses operate on entire texts and their structural properties:

• Topic Modeling — Latent Dirichlet Allocation (LDA), Correlated Topic Models (CTM), and Structural Topic Models (STM), with automated K selection using four complementary metrics (CaoJuan 2009, Arun 2010, Deveaud 2014, Perplexity/Lower Bound), consensus recommendation, model diagnostics, and covariate effects analysis
• Syntactic Complexity — dependency-based metrics including Mean Dependency Distance (Liu, 2008), parse tree depth, clauses per sentence, subordination ratio, and branching factor, computed per document with corpus-level summaries and distribution plots
• SVO Triplet Extraction — Subject-Verb-Object relationship extraction using dependency parsing, with frequency tables, Sankey flow diagrams, and verb frequency analysis for content and narrative structure analysis
• Polarity Detection — lexicon-based sentiment analysis with contextual valence shifters (negators, amplifiers, diminishers), supporting Hu & Liu, Loughran & McDonald, and NRC lexicons. Document-level polarity scoring with distribution visualization
• Emotion Analysis — eight-emotion detection (Anger, Anticipation, Disgust, Fear, Joy, Sadness, Surprise, Trust) using the NRC Word-Emotion Association Lexicon (EmoLex), with emotion distribution charts, radar plots, per-document heatmaps, and top contributing words per emotion
• Text Summarization — extractive summarization via the TextRank algorithm (graph-based sentence ranking with PageRank scoring), and abstractive summarization powered by Google Gemini AI

TALL AI

TALL integrates Google Gemini AI as an intelligent assistant that provides automated interpretation of analytical results. Available across most analysis tabs (Overview, KWIC, Correspondence Analysis, Co-occurrence Network, Thematic Map, Word Embeddings, Topic Modeling, Polarity Detection, Emotion Analysis, Syntactic Complexity, SVO Triplets), TALL AI examines the visual and numerical outputs and generates contextual, academically-grounded interpretations. AI calls run asynchronously, keeping the application responsive during processing.

Reporting

All analyses can be exported to an Excel workbook with embedded plots, enabling reproducible reporting. Individual plots can be exported as high-resolution PNG images with configurable DPI settings. Network visualizations use native canvas capture for crisp, DPI-aware rendering.

Screenshots

Import text from multiple file formats

Edit, divide, and add external information

Automatic Lemmatization and PoS-Tagging

Language, Model, and Analysis Term Selection

Special Entity Tagging

Multi-word Expression Extraction

Overview — Descriptive statistics, concordance analysis, word frequency distributions

Words — Topic detection, correspondence analysis, co-occurrence networks

Documents — Topic modeling, sentiment analysis, syntactic analysis

Authors

Creators

• Massimo Aria — University of Naples Federico II
  • email: [email protected] | ORCID: 0000-0002-8517-9411
• Maria Spano — University of Naples Federico II
  • email: [email protected] | ORCID: 0000-0002-3103-2342
• Luca D’Aniello — University of Naples Federico II
  • email: [email protected] | ORCID: 0000-0003-1019-9212
• Corrado Cuccurullo — University of Campania Luigi Vanvitelli
  • email: [email protected] | ORCID: 0000-0002-7401-8575

Contributors

• Michelangelo Misuraca — University of Salerno
  • email: [email protected] | ORCID: 0000-0002-8794-966X

Maintainer

Massimo Aria ([email protected])

License

MIT License. Copyright 2023-2026 Massimo Aria.

See LICENSE for details.