On operationalizing syntactic complexity

An experiment comparing three measures of syntactic complexity — node counts, word counts, and a so-called ‘Index of Syntactic Complexity’ — with regard to their accuracy and applicability concludes that since node counts are in most cases unreasonably resource demanding to conduct, researchers can feel safe in using the measure that is most economically to conduct , word counts.

Abstract

In the recent functional linguistic literature, the notion of syntactic complexity and similar concepts have received considerable attention. Yet, when trying to operationalize syntactic complexity as an independent variable in statistical research designs, it soon emerges that the notion is somewhat underdefined. I will report the results from an experiment comparing three measures of syntactic complexity — node counts, word counts, and a so-called ‘Index of Syntactic Complexity ’ — with regard to their accuracy and applicability. While presupposing that node counts are cognitively the most ‘real ’ measure, it turns out that the other two measures are near-perfect proxies of the former. I conclude that since node counts are in most cases unreasonably resource demanding to conduct, researchers can feel safe in using the measure that is most economically to conduct, word counts.