A landmark new paper by Siva Kalyan and Mark Donohue uncovers the main ways in which the grammars of the world’s languages vary, on the basis of a dataset consisting of over 1,000,000 data points, covering 3,089 of the world’s languages. The title of the paper is: “The Dimensions of Morphosyntactic Variation: Whorf, Greenberg and Nichols were right“.

A computational analysis reveals that the dimensions of variation that emerge from the data conform to existing proposals (e.g., the classification of languages according to whether the verb precedes or follows the direct object, or whether they rely more on case-marking or agreement to indicate the relation between the verb and its arguments), while also suggesting previously unsuspected universals (e.g., that gender and plural marking tend not to co-occur with ergativity and clusivity contrasts; see the black and white maps). The same dimensions of variation recur within individual continents, confirming their universality.

A series of maps, allowing us to visualise the different combinations of these dimensions of variation, reveals areas of typological similarity around the world. The coloured map here shows a number of areas defined purely on linguistic typological grounds, based on the first three dimensions of variation; distinctly different types of languages can easily be identified, such as those in Southeast Asia, Australia, or Mesoamerica.

The figure below the map shows the interaction of dimensions 2 and 3, with the same colouring as the map. Some identifiable regions are highlighted (from the article). The appendices provide alternative visualisations of the dimensions, as well as maps of individual features, which will be useful pedagogically as well as for research purposes.

The paper is published in Linguistics Typology at the Crossroads, and is available in open access at: https://typologyatcrossroads.unibo.it/article/view/17482.

The data and code are available at https://osf.io/u9qbe/, as is a link to an interactive globe.