A Multi-Tier Framework for Understanding Spoken Language

Language can be approached from many different vantage points—neuroanatomy, psychology, phonetics, hearing, vision, physics, information theory, formal logic, and so on. Most scientific and linguistic accounts have focused on defining units of analysis, such as the phoneme, the word, and the phrase. Some of the more ingenious, such as articulation theory (Allen, 1994 ; Fletcher, 1953 ; Fletcher & Gault, 1950 ; French & Steinberg, 1947 ) and the Speech Transmission Index (Houtgast & Steeneken, 1985 ) characterize speech communication mostly in terms of equations. To date, all such efforts have failed, largely because language is extraordinarily multidimensional and not particularly amenable to simplifying assumptions. The great American linguist, Edward Sapir, characterized the problem very succinctly: “All grammars leak” (Sapir, 1921 , p. 39). No language can be captured entirely in terms of a closed system of equations; there are always exceptions to the rules. Humans are particularly adept at learning exceptions and accepting them as “normal.”