Breadth versus depth: Theoretical reasons for system-independent comparison of languages

This paper examines the question of breadth vs. depth of description in theoretical and comparative grammar. I argue that there is no reason why comparison of languages with the aim of finding universal properties should have to be based on wide-ranging (“deep”) distributional descriptions. Large-scale comparisons must first of all be based on uniformly applicable, system-independent concepts for comparison. By contrast, “deep” and wide-ranging language-particular descriptions are ideally based on language-particular categories (p-categories), in the spirit of “Boasian methodological relativism”.