Measuring Linguistic Complexity: The Morphological Tier

Abstract How to measure morphological complexity is itself an issue of some complexity. (Nichols, 1992) This article develops an information‐theoretic and functionally motivated method of measuring “linguistic complexity”; from corpora that can in theory be applied to any definable substructure. This method is further extended into a way of objectively and numerically measuring the morphological complexity of a given language sample, avoiding the typical difficulties of focusing on only a few unrepresentative types of constructions. By selectively altering the morphological information present in a sample, the complexity can be measured as the change in overall informativeness of a text. This claim has been tested in a small‐scale cross‐linguistic experiment; the results agree well with both intuitions and existing measurements.