Home › Language
Type-token ratio (TTR)
Type-token ratio compares how many different words a speaker uses with how many words they use in total. It is quick to calculate, widely cited, and easy to misuse.
The definition
Tokens are all the words in a sample. Types are the different words. The ratio is:
TTR = number of types ÷ number of tokens
The result lies between 0 and 1. A higher value means less repetition and, loosely, a more varied vocabulary.
A worked example
Take the sentence “The cat sat on the mat.”
- Tokens: the, cat, sat, on, the, mat = 6
- Types: the, cat, sat, on, mat = 5 (“the” appears twice but is one type)
- TTR = 5 ÷ 6 = 0.83
Why clinicians use it
TTR and related measures are used in language sample analysis to describe how varied a person’s word choice is, for example in children’s language development (Templin, 1957), in aphasia, and in research on discourse. It offers a quick, objective number from a conversational or narrative sample.
The big problem: sample length
TTR falls as a sample gets longer, because common words such as “the” and “and” keep repeating while new words become rarer. A 50-word sample will almost always have a higher TTR than a 500-word sample from the same speaker. Two consequences follow:
- Only compare TTRs from samples of the same length.
- Do not compare your result with a published norm unless your sample was collected and counted in the same way.
Better alternatives
| Measure | Idea |
|---|---|
| MATTR (moving-average TTR) | Slide a fixed window (often 50 words) along the sample, calculate TTR in each window, then average. Much less sensitive to length (Covington & McFall, 2010). |
| MTLD | Measures how many words it takes before the running TTR drops to a set threshold, then averages across the text (McCarthy & Jarvis, 2010). |
| vocd-D / HD-D | Model-based or sampling-based estimates designed to reduce length effects (McCarthy & Jarvis, 2010). |
Decisions you must make and record
- What counts as a word? Contractions (“don’t”), hyphenated words, numbers and proper names need a rule.
- Repetitions, false starts and fillers. Are “um” and repeated words included?
- Word forms or root words? Is “walk” the same type as “walked”?
- Whose speech? Count only the target speaker, not the partner.
Write your rules down so the same method can be repeated at re-assessment.
Free calculator
Paste a sample below. The tool lower-cases the text, treats letters, numbers and apostrophes as parts of words, and reports TTR and MATTR. Nothing you type leaves your browser.
Interpretation. TTR is a descriptive measure, not a diagnostic test. Use it alongside other evidence such as standardised assessments, parent or carer report and clinical observation.
Sources and further reading
- Templin, M. C. (1957). Certain Language Skills in Children. University of Minnesota Press.
- Covington, M. A., & McFall, J. D. (2010). Cutting the Gordian knot: The moving-average type–token ratio (MATTR). Journal of Quantitative Linguistics, 17(2), 94–100.
- McCarthy, P. M., & Jarvis, S. (2010). MTLD, vocd-D, and HD-D: A validation study of sophisticated approaches to lexical diversity assessment. Behavior Research Methods, 42(2), 381–392.