The Warming Gradient

The Warming Gradient Lexical Temperature Analysis Across Nine Curated Word Lists Saxon Ventura Research Ltd., CC BY licence 16th of January, 2026 Abstract We analyse lexical temperature (Z-score relative to the alphabetic midpoint 13.5) across nine word lists ranging from a baseline corpus of 455,247 English words to specialized vocabularies. A consistent warming gradient emerges: curated word lists selected for human use exhibit systematically higher mean Z-scores (warmer temperatures) than the general corpus. The gradient spans from -17.0 (baseline) to -6.64 (Swadesh universal vocabulary), with children's first words (-7.80) and Swadesh (-6.64) representing the warmest lists. Unexpectedly, spiritual vocabulary (-11.61) is colder than financial vocabulary (-10.25). The word LOVE appears at perfect equilibrium (Z = 0) in four of six lists where it occurs. These findings suggest that humans unconsciously select words closer to alphabetic equilibrium when curating meaningful vocabulary, and that the temperature of specialized vocabularies may reflect their accessibility or psychological distance from everyday human experience. 1. Introduction In prior work (Statistical Analysis of 455,247 English Words, DOI: 10.5281/zenodo.18189640), we established that the English lexicon exhibits non-random structure when analysed through positional ordinal values. Each letter is assigned its alphabetic position (A=1, B=2, ... Z=26), and for each word, we compute the simple mean M/n where M is the sum of letter values and n is word length. The alphabetic midpoint is 13.5 = (1+26)/2. Temperature (Z) is defined as Z = M - 13.5n, measuring a word's deviation from perfect alphabetic balance. Words with Z = 0 sit at exact equilibrium (M/n = 13.5). The baseline English corpus has mean Z = -17.0, indicating systematic weighting toward early-alphabet letters. This paper extends the analysis to eight additional curated word lists, testing whether specialized vocabularies differ systematically in their lexical temperature. 2. Methods 2.1 Word Lists Analysed Nine word lists were analysed, selected to represent different domains of human vocabulary:...