The Science of Words

Miller argues that the word is the fundamental unit of language and that scientific study has wrongly neglected the lexicon in favor of grammar. He proposes that every word synthesizes three facets—utterance, concept, and syntactic role—and that knowing a word means knowing a mapping between form and meaning stored in the mental lexicon. This approach rejects the Chomskyan emphasis on syntax as the primary locus of linguistic science. As Miller states in his preface, the structure and functioning of the human mind are intimately related to the capacity for communication, and this relation is a consequence of an innate human capacity for language . The lexicon, not the sentence, becomes the all-embracing aspect of language deserving scientific treatment.

The 1976 book Language and Perception, co-authored with Philip Johnson-Laird, attempted to ground word meanings in perceptual primitives—predicates like Red(x) or Hold(x,y) abstracted from sensory experience. That ambitious programme never produced the complete inventory of concept atoms it promised. In The Science of Words, Miller shifts strategy. Rather than decomposing meaning into perceptual routines, he organizes the lexicon around the word itself and its three facets. He acknowledges in the preface that the earlier search for a constructive theory of lexical knowledge was less easily satisfied than a differential theory, in which meanings can be represented by any symbols that distinguish them . This pragmatic turn directly anticipated WordNet's synonym-set architecture.

The Science of Words was published in 1991, the same year Miller began WordNet at Princeton. The book functions as WordNet's theoretical prospectus: Miller argues the lexicon deserves primary scientific attention, that words map forms to concepts, and that lexical knowledge can be represented computationally. A 1999 paper by Miller notes that WordNet is organized primarily in terms of lexicalized concepts and secondarily in terms of word forms, and that an elaborate semantic organization is provided for those concepts . The book's emphasis on the word as the key to language structure—and on the mental lexicon as a legitimate object of scientific study—provided the conceptual foundation for the database.

W. N. Francis, reviewing the revised 1996 edition in Language, described it as "an introduction to linguistics for beginners in the field, capable of following and understanding a text written in a style more suited to an advanced reader, presumably at university level" . He praised its step-by-step development but warned that simplicity gives way to complexity as the book progresses, potentially baffling students. Francis called it "authoritative but in a way that challenges both student and instructor" . The book's choice to base exposition on the word rather than grammar distinguishes it from conventional introductions, and its color illustrations—from Linear B to the Rosetta Stone—were noted as an attractive feature.

The exponential principle is Miller's "general design principle" stating that the number N of different strings of length λ formed from a repertoire of A alternative units equals A raised to the λ power (N = A^λ) . The formula shows how a small number of basic units—letters in writing, phonemes in speech, or words in sentences—can generate an enormous number of distinct strings. This principle matters because it explains why language is productive: finite resources yield infinite expression. Miller introduces it in his chapter on units of analysis to justify his focus on the word as a fundamental unit. The exponential relationship between unit types and string possibilities is, in effect, the mathematical reason why the lexicon can be so powerful despite being finite.

The title "The Science of Words" leads some readers to expect a history of word origins, etymology, or a popular treatment of vocabulary. One Amazon reviewer expressed disappointment that the book was "a classroom textbook" rather than a book on word origins . In fact, the book is a synthesis of linguistic science's discoveries about words as cognitive and cultural objects—covering the mental lexicon, categorizing experience, the written word, and the decipherment of ancient scripts. It is neither etymological nor a how-to guide to vocabulary. Miller's purpose is theoretical and scientific: to establish the word as a legitimate object of psychological and linguistic investigation, not to trace where individual words came from.