Text Analysis¶

Analyze passages for readability across phonological, lexical, semantic, and affective dimensions with interactive color-coded highlighting.

Overview¶

Paste a passage to get aggregate percentile statistics across the ~150-column property spine. Word-level highlighting colors the passage by any single property.

Key Features: - Aggregate percentile statistics for entire passages - Interactive highlighting by property with color gradients - Three preset phonetics passages (Grandfather, Rainbow, Caterpillar) - Coverage tracking (percentage of words in vocabulary) - Unknown word identification with dotted underlines

Use Cases: - Clinical: Assess therapy script complexity, select developmentally appropriate reading materials - Education: Evaluate text readability for students at different levels - Research: Analyze stimulus materials for controlled psycholinguistic properties - SLP: Track phonological, lexical, and semantic complexity of treatment materials

Basic Usage¶

Enter or paste text into the input field (or load a preset passage)
Click "Analyze Text" to compute aggregate statistics
Select a feature from the dropdown to highlight words by percentile
Review aggregate percentiles to understand overall passage characteristics
Hover over highlighted words to see individual percentile ranks

Analyzed Properties¶

The highlight dropdown surfaces the core dimensions below; the underlying data spine carries ~150 property columns.

Phonological Complexity (3 properties)¶

Property	Interpretation	Color Scheme
Syllables	Lower = simpler	Orange gradient (light → dark)
Phonemes	Lower = simpler	Orange gradient (light → dark)
WCM (Word Complexity Measure)	Lower = simpler	Orange gradient (light → dark)

Lexical Properties (2 properties)¶

Property	Interpretation	Color Scheme
Frequency	Higher = more common	Red (rare) → Green (common)
Age of Acquisition	Lower = learned earlier	Orange gradient (early → late)

Semantic Properties (2 properties)¶

Property	Interpretation	Color Scheme
Familiarity	Higher = more familiar	Blue gradient (low → high)
Concreteness	Higher = more concrete	Blue gradient (low → high)

Affective Properties (2 properties)¶

Property	Interpretation	Color Scheme
Valence	Higher = more positive	Red (negative) → Green (positive)
Arousal	Higher = more exciting	Purple gradient (calm → excited)

Preset Passages¶

Grandfather Passage¶

Type: Standard phonetics passage Purpose: Adult speech assessment, voice/resonance evaluation Length: 113 words Source: Standardized diagnostic passage for speech-language pathology

Rainbow Passage¶

Type: Classic phonetics passage Purpose: Voice quality assessment, speech rate analysis Length: 85 words Source: Fairbanks (1960) - widely used in clinical and research settings

Caterpillar Passage¶

Type: Pediatric speech sample Purpose: Child speech assessment, narrative language evaluation Length: 181 words Source: Standardized passage for pediatric speech assessment

Understanding Aggregate Percentiles¶

Aggregate percentiles represent the average percentile rank of all known words in the passage:

0-25th percentile: Very low on this property (e.g., very simple, very rare, very concrete)
25-50th percentile: Below average
50-75th percentile: Above average
75-100th percentile: Very high on this property (e.g., very complex, very common, very abstract)

Example Interpretation:

Syllable Count: 35th percentile → Passage uses words with fewer syllables than average
Frequency: 72nd percentile → Passage uses more common words than average
Concreteness: 20th percentile → Passage uses more abstract words than average

Interactive Highlighting¶

Color Schemes¶

Difficulty/Complexity (Syllables, Phonemes, WCM, AoA): - Light orange → Words at low percentiles (simple/early) - Dark orange → Words at high percentiles (complex/late)

Frequency (Frequency): - Red → Rare words (low percentiles) - Yellow → Medium frequency - Green → Common words (high percentiles)

Semantic (Familiarity, Concreteness): - Light blue → Low ratings (abstract, unfamiliar) - Dark blue → High ratings (concrete, familiar)

Diverging (Valence): - Red → Negative valence (low percentiles) - Gray → Neutral - Green → Positive valence (high percentiles)

Intensity (Arousal): - Light purple → Low arousal (calm) - Dark purple → High arousal (excited)

Unknown Words¶

Words not in PhonoLex's vocabulary (~48K canonical content words) are marked with: - Dotted gray underline - Reduced opacity (60%) - Hover tooltip: "Unknown word (not in vocabulary)"

Words with no data for the selected property are marked similarly: - Dotted gray underline - Hover tooltip: "[word]: No data for this property"

Clinical Applications¶

Therapy Script Assessment¶

Goal: Ensure therapy materials are developmentally appropriate

Steps: 1. Paste therapy script into Text Analysis 2. Review Syllable Count, WCM, and AoA percentiles 3. Target 25-50th percentiles for early intervention (simple words) 4. Target 50-75th percentiles for intermediate stages 5. Adjust script by replacing high-percentile words with simpler alternatives

Example:

Original: "The enormous elephant enthusiastically embraced exercise."
Analysis: Syllables (85th), WCM (92nd), AoA (88th) → Too complex for early intervention

Revised: "The big elephant liked to run and play."
Analysis: Syllables (35th), WCM (28th), AoA (15th) → Appropriate

Reading Material Selection¶

Goal: Match text complexity to student reading level

Steps: 1. Analyze candidate passages 2. Compare Frequency and Concreteness percentiles 3. Select texts with properties matching target level: - Early readers: High concreteness (60-80th) - Intermediate: Mixed concreteness - Advanced: Abstract concepts (20-40th)

Stimulus Control for Research¶

Goal: Match experimental stimuli on psycholinguistic properties

Steps: 1. Analyze Condition A passage 2. Analyze Condition B passage 3. Compare aggregate percentiles across the analyzed properties 4. Adjust passages to minimize differences in confounding variables 5. Re-analyze to confirm matching

Performance¶

Operation	Time	Notes
Text analysis	50-100 ms	Depends on passage length
Percentile lookup	<1 ms	Pre-computed during export
Highlighting render	5-20 ms	Depends on passage length

Scalability: Tested on passages up to 1,000 words without performance issues.

Data Coverage¶

Vocabulary: ~48K canonical content words (content-word frequency mass + WordNet realness, plus a curated deny/allow list; the full ~125K CMU-phonology lexicon backs exact-word lookup)

Property Coverage (% of vocabulary with data): - Phonological properties: 100% (all words) - Phonotactic probability: ~100% - Frequency: 100% (all words - filtering criterion) - AoA: ~100% canonical (PhonoLex, in-house) - Familiarity: ~100% canonical (PhonoLex, in-house) - Concreteness: ~100% canonical (PhonoLex, in-house) - Valence + Arousal: ~100% canonical (PhonoLex, in-house; Warriner D axis not re-derived)

Coverage Impact: Passages with technical/specialized vocabulary may have lower coverage. Unknown words and words without specific property data are clearly marked.

Limitations¶

Vocabulary Coverage¶

~48K canonical content words covers 99%+ of typical English text
Proper nouns, neologisms, and specialized jargon may not be included
Non-English words and code-switching not supported

Property Availability¶

Not all words have every property (see coverage above)
Missing properties marked with dotted underline when selected
Aggregate percentiles computed only from words with data

Interpretation Caveats¶

Percentiles are corpus-relative, not absolute measures
High-frequency words may be simple OR common depending on context
Semantic properties (familiarity, concreteness) are canonical-only, so they have lower coverage than phonological properties on out-of-vocabulary-heavy text

Linguistic Limitations¶

General American English pronunciation only (CMU Dictionary)
Single-word analysis (no multi-word expressions or idioms)
No syntactic or discourse-level analysis

References¶

Data Sources: - Word frequency is a PhonoLex in-house derivation from the FineWeb-Edu educational web corpus (~800M tokens); the norms below serve as scale anchors for the in-house derivations. - Brysbaert, M., Warriner, A. B., & Kuperman, V. (2014). Concreteness ratings for 40 thousand English words. Behavior Research Methods, 46(3), 904-911. - Scott, G. G., et al. (2019). The Glasgow Norms. Behavior Research Methods, 51(3), 1258-1270. - Vitevitch, M. S., & Luce, P. A. (2004). Phonotactic probability for words and nonwords in English. Behavior Research Methods, 36(3), 481-487. (Method origin; computed in PhonoLex from the CMU Pronouncing Dictionary.) - Warriner, A. B., Kuperman, V., & Brysbaert, M. (2013). Norms of valence, arousal, and dominance. Behavior Research Methods, 45(4), 1191-1207.

Standard Passages: - Fairbanks, G. (1960). Voice and articulation drillbook (2nd ed.). New York: Harper & Row. (Rainbow Passage) - Darley, F. L., Aronson, A. E., & Brown, J. R. (1975). Motor speech disorders. Philadelphia: Saunders. (Grandfather Passage)