The approach to language typology pioneered by Joseph Greenberg has inspired a large literature and a fruitful search for cross-linguistic generalizations in many different grammatical domains. Only recently, however, have linguists begun to understand some of the theoretical ideas and assumptions underlying Greenberg's conception of language, and their significance for understanding human cognition. After a long period in which the dominant view in American linguistics saw language as determined by innate structural principles, encapsulated from the rest of cognition, and thus largely unaffected by the processes of speaking and understanding, the field has gradually become receptive to a view in which actual language usage, including perception, production, distributional information, and semantic and functional factors, is the key to understanding a great many aspects of linguistic structure. From this standpoint, many of Greenberg's early research questions and findings look extremely prescient. For example, he investigated the relation of frequency to grammatical categories and their function (for example, discovering a relation between frequency of particular semantic classes of nouns in particular case categories in Russian in Greenberg 1979), and he also proposed numerous processing explanations for language universals (e.g., word order universals, numeral systems, and kinship systems, in Greenberg (1966, 1978, and 1980 respectively). Both frequency and processing explanations are fundamental in current usage-based accounts of language structure (cf. for example the work of some of the current leading figures in typology such as Joan Bybee and John Hawkins).
A third type of explanation offered by Greenberg is semantic: he identifies plausible semantic links that motivate typological patterns. In the terms of modern cognitive linguistics, in this work Greenberg establishes conceptual connections between particular semantic categories that justify the recurrent synchronic and diachronic relations found among linguistic categories. In this paper I focus on this last type of explanation, and claim that it is potentially the richest and most significant for understanding human conceptual structure. I will give some examples (including the ergative patterning of collective markers identified by Sapir and others, and of the relation of object marking to transitivity markers) showing how Greenberg's brief account of some particular semantic relation, properly understood, has profound implications for human categorization. Uniting Greenberg's unerring identification of conceptual semantic links among categories with the more elaborated descriptive mechanisms of cognitive semantics, we can integrate his explanations with what is more generally known about human cognition. Moreover, I will show that by using the same basic idea and identifying other such conceptual links, we can ultimately hope to arrive at a broad and detailed picture of what kinds of conceptual distinctions and conceptual groupings are important to human beings in general. Thus, this approach gives us deeper insight into the human mind.