Readability Benchmarks by Text Type (20 samples: academic, technical, casual, formal, mixed) — Flesch, Flesch-Kincaid, SMOG, Gunning Fog

An open benchmark dataset of computed readability scores for 20 short text samples drawn from common real-world text types: academic research abstracts, journal introductions, law reviews, software documentation, medical guidance, scientific methods, blog posts, social media captions, news briefs, emails, children's books, government notices, corporate memos, speeches, marketing copy, textbooks, release notes, recipes, and legal disclaimers. For each sample the dataset records the genre, sample topic, the source excerpt, and the computed Flesch Reading Ease, Flesch-Kincaid Grade, SMOG Grade, Gunning Fog index, average sentence length, average word length in syllables, passive-voice ratio, complex-word ratio, target audience, and a plain-language interpretation. This makes it possible to calibrate a readability tool against known-difficulty reference texts rather than only against formula reference bands. An interactive readability scorer built on this benchmark set is available at LikeNative, which computes these same metrics on arbitrary submitted text. Columns: benchmark_id, text_type, genre, sample_topic, sample_excerpt, flesch_reading_ease, flesch_kincaid_grade, smog_grade, gunning_fog, avg_sentence_length_words, avg_word_length_syllables, passive_voice_ratio_pct, complex_word_ratio_pct, target_audience, interpretation. License: CC0 / public domain.