N1904addons - Feature: morph_ttr

Feature group	Feature type	Data type	Available for node types	Feature status
statistic	Node	str	book chapter verse sentence group wg phrase subphrase clause	✅

Feature short description

Type to Token Ratio based on morph-tags for all word nodes under this node.

Feature values

A float number stored as a string representing a ratio in the range 0 to 1 (inclusive) where the dot denotes a decimal point, not a thousands separator.

Detailed feature description

This feature provides the Morph-to-Token Ratio (MTR), which is a measure for morph diversity. It is defined as:

\[\text{MTR} = \frac{|\{\text{unique morphs in the text}\}|}{N}\]

Visualizing

The following plot compares the Type-to-Token Ratios measured over word form (TTR), lemma (LTR), and morphology (MTR) for each book of the New Testament. The image clearly shows that shorter books generaly are resulting in higher ratio, even though TTR is iself already a normalized measure. To account for this length-related bias there are various methods of normalization. A large number of those methods are made conveniently accessible using the Python package lexicalrichness.

See also

Related features:

• text_ttr: Type to Token Ratio based on wordforms for all word nodes under this node.
• lemma_ttr: Type to Token Ratio based on lemma for all word nodes under this node.

Data source

The production notebook can be found on this repository.