Explaining universal semantic properties in the adjectival/adverbial domain

Native speakers would generally agree that certain patterns in the language are preferable to others. The order of several adjectives modifying the same noun is the best example: various potential sequences of adjectives are all equivalent in meaning, yet speakers often prefer specific orders. Recently, with the availability of large quantities of data and significant development of processing tools and statistical methods, researchers have started to study word order preferences using quantitative approaches based on “efficiency-based” communication. In our work, we test three of these theories: pointwise mutual information (PMI) (Futrell, Qian, et al. 2019), integration complexity (IC) (Dyer 2017), and information gain (IG) (Futrell, Dyer, and Scontras 2020). We start by replicating the results obtained by (Futrell, Dyer, and Scontras 2020) on adjective order preferences in English. To verify the cross-linguistic consistency of the results, we analyze adjective order tendencies in Russian. Since the theories in the study provide insights into base word order, not limited to adjectives only, we continue by investigating in detail if the predictions made by these theories can generalize and be applied to explain the order of adverbs.