Skip to main content
. 2020 Jan 21;117(5):2347–2353. doi: 10.1073/pnas.1910923117

Fig. 4.

Fig. 4.

Predictability and parseability of the real word-order grammars of 51 languages (red), indicated by International Organization for Standardization codes, compared to baseline word-order grammars (blue distribution). Predictability and parseability scores are z-scored within language, to enable comparison across languages. The gray curve indicates the approximate Pareto frontier of computationally optimized grammars, averaged over the 51 languages, with dashed SDs.