Measuring vocabulary richness
Lexical diversity describes how many different words a text uses relative to its length. It is used in language assessment, child-language research, authorship studies and editing.
- Type-token ratio (TTR) is distinct words (types) divided by total words (tokens). It is intuitive but falls steadily as a text gets longer, because common words repeat. Only compare TTRs of texts with similar length.
- Root TTR (Guiraud’s index) divides types by the square root of tokens, which reduces (but does not remove) the length effect.
- MTLD (Measure of Textual Lexical Diversity), introduced by McCarthy and Jarvis in 2010, walks through the text and counts how many times the running TTR drops to 0.72. The average length of those segments, computed forwards and backwards, is the score. MTLD is far less sensitive to length, which is why researchers prefer it.
- Hapax legomena are words that appear exactly once.
Reading the result
Higher MTLD means more varied vocabulary. Rather than comparing with a universal “good” number, compare like with like: two drafts of the same essay, chapters of the same book, or a student’s writing over time. Very high diversity is not automatically better (technical writing repeats terms on purpose, because consistency aids understanding).
Words are lower-cased and stripped of surrounding punctuation before counting; inflected forms (“run”, “runs”) count as different types.
Worked example
Here is what the Lexical Diversity Calculator returns for its built-in example. Click Try an example in the tool above to load it, then change it as you like.
Good writing is rewriting. The first draft gets the ideas onto the page; the second makes them clear; the third makes them sing.
Most readers skim before they commit. Short paragraphs, specific verbs and a clear opening sentence give them a reason to stay. Long, winding sentences that bury the point in the middle make them leave.
So count your words, but count what they do, too.| Result | Value |
|---|---|
| MTLD | 89.8 |
| Type-token ratio | 0.791 |
| Root TTR | 6.47 |
| Words (tokens) | 67 |
| Distinct words (types) | 53 |
| Used once (hapax) | 46 (68.7%) |
Questions
What is a good type-token ratio?
There is no single target because TTR depends heavily on text length. Use MTLD for comparisons across lengths, and compare texts of a similar type.
What is MTLD?
The Measure of Textual Lexical Diversity: the average length of word sequences that keep a type-token ratio above 0.72. It is much less affected by text length than TTR.
Last reviewed · How Textkiln tools work