Only 2 of 154 Most spoken Languages from 23 Language Families Employ a Labial for "Tongue" in 1st Consonant Position Demonstrating Embodied Iconicity for Tongue Vs Mouth in "inverse" Control

Abstract

Class-Level Embodied Iconicity in Words for TONGUE and MOUTH: Evidence From 154 Major Languages, With Genealogical Controls and a Balanced-Sample Replication Author: M. Phillips Affiliation: Independent Contact: mg@mgcode.mozmail.com Version: v6 (September 2026). Supersedes v4 and 5 (Zenodo record 18391312). Concept DOI, always resolves to the latest version: 10.5281/zenodo.17791741 Dataset DOI: 10.5281/zenodo.17791715 (c) 2025-2026 M. Phillips. CC-BY 4.0. Version pointer discipline: this document states its own version number once, in the field above. Every internal cross-reference is to the concept DOI, which always resolves to the current version. Hard-coded version numbers scattered through a body text go stale silently and were a source of contradiction in earlier versions. I welcome your feedback. Please feel free to write at the address above. If you find the article useful please forward it. VERSION NOTE - WHAT CHANGED IN v6, AND WHY This version corrects the scope of the central claim, supplies the statistical test that was missing from the abstract, adds a genealogical control that no earlier version ran, and adds a full disclosure of AI involvement. SCOPE CORRECTION. Earlier versions described the pattern as "Near Universal." That is withdrawn. The 98.6% figure describes the world's most widely spoken languages. It is a major-language vocabulary pattern, not a class-level cross-linguistic universal. The body text was already correctly scoped - section 1.2 has always described the principle as "systematic (though not universal)" and predicted "tendencies, not absolutes" - so this correction brings the title into agreement with the paper. STATISTICAL CORRECTION. A sentence in the earlier abstract ended where a p-value belonged, and the value was never supplied. It is replaced here with a named test, a stated null and a reproducible figure. Two mutually incompatible p-values that appeared in earlier versions are withdrawn; neither was correct and neither is reproduced by the test specified in 3.3. GENEALOGICAL CONTROL. Section 5 is new. It reports the composition of the sample by language family and the result of collapsing the test to one observation per genealogical unit. This is the analysis a typologist would run first, and it is better run by the author than discovered by a reviewer. BALANCED-SAMPLE REPLICATION. Section 6 is new and includes two results that bear against the strength of the claim. DISCLOSURE. A full statement of AI involvement is new, and appears before Data Availability. An appended assessment written by a large language model has been removed in full; the disclosure section explains why. BASIS OF THESE CORRECTIONS, STATED PLAINLY. The defects corrected here were verified against a March 2026 working copy of the published text. The v4 record was not retrievable during preparation of this version. Where a correction listed above has already been made in v4, this version simply carries it forward; no claim is made here about the specific contents of v4. SCOPE OF THIS VERSION. TONGUE and MOUTH only. Results for other concepts exist and are held pending separate disposition. ABSTRACT Across the 154 most widely spoken languages, the word for "tongue" begins with a lingual consonant in 136 cases and a labial consonant in 2. Within the lingual-labial binary (n=138), that is 98.6% lingual against 1.4% labial. Words for "mouth" in the same languages show the inverse tendency, elevated labial selection, giving an internal control drawn from the same speech communities rather than an external corpus. Against a null in which the first consonant is selected without regard to the organ named - a labial share of 28.0% within the lingual-labial binary, derived from 172,877 non-target forms in the WALS-200 sample - a one-sided binomial test on the 154-language binary returns p = 2.9e-17. That figure assumes the 154 languages are independent observations, and they are not: they represent 23 genealogical units, of which Indo-European alone supplies 61 languages. Collapsing to one observation per genealogical unit, the same test on 23 units with 2 labial outcomes returns p = 0.025. The effect survives genealogical collapse at conventional significance, at an effect size orders of magnitude smaller than the naive figure. Independent replication on the Comrie-Dryer WALS-200 balanced typological sample (Lexibank Analysed v2 merged with ASJP v20, 95% coverage, n=191) gives TONGUE first-consonant lingual at 74.3% against a 67.0% baseline, OR 1.43, p = 0.017; and MOUTH first-consonant labial at 38.6% against a 26.1% baseline, p = 3.7e-4. On two further balanced samples - a 292-language one-per-lineage curated set and a 359-language ASJP random sample - the TONGUE effect is not significant (p = 0.29 and p = 0.60 respectively). This is reported because it bears against the strength of the claim. The methodological contribution is the unit of analysis. Prior cross-linguistic work tests individual phonemes; this work tests an articulatory CLASS. The distinction is demonstrable rather than asserted: Blasi et al. (2016) report a lateral-tongue association and name English as an exception to it, since English "tongue" begins with T. A lingual-class analysis captures English, and equally captures German Zunge, Turkish dil, Greek glossa and Quechua qallu, all of which a single-phoneme analysis misses. The class has strictly greater coverage than the phoneme on the same concept. The complete 154-word list is reproduced below so that any reader may check it. AI involvement in this work is disclosed in full before the Data Availability statement. SECTION 1: INTRODUCTION 1.1 Theoretical Foundation For over a century, Saussure's principle of arbitrariness has dominated linguistic theory, positing that the relationship between sound and meaning is fundamentally random. This framework challenges that orthodoxy through systematic demonstration of embodied articulatory reenactment - the principle that speakers engage specific articulatory organs and physiological processes due to cognitive resonance seeking a match between word and meaning, when naming concepts related to those organs. The Saussurean paradigm treats sound-meaning relationships as arbitrary convention. This framework proposes an alternative: when naming body parts involved in speech, speakers systematically engage those very organs - the tongue "speaks" about itself using tongue-produced sounds. This work represents 25 years of independent research development, integrating classical linguistic philosophy with modern cross-linguistic analysis. The intellectual lineage traces from Plato's Cratylus dialogue, examining sound-meaning relationships in Greek, through Gesenius's Hebrew philology documenting systematic patterns in Semitic languages, to contemporary embodied cognition research. While Plato raised the question philosophically, and Gesenius documented patterns philologically, this framework provides a systematic cross-linguistic empirical test with explicit semantic-class definition plus first-position salience, along with cumulative measurement methodology. 1.2 Core Hypothesis The central hypothesis proposes that embodied articulatory reenactment operates as a systematic though not universal principle in language formation. Specifically: when concepts relate to articulatory organs or physiological processes, languages show statistical tendencies to employ consonants produced by or associated with those organs. This represents cognitive selection for resonant forms rather than pure arbitrariness. The hypothesis is falsifiable. Languages should show elevated frequencies of tongue-produced consonants in "tongue" words and lip-produced consonants in "mouth" words. This is not mystical correspondence; it is measurable statistical tendency reflecting cognitive preference. Critically, this framework does not claim universal patterns across all languages for all concepts. Rather, it proposes a cognitive ideal model - if language were designed for optimal mind-body integration, these would be the operative principles. Actual languages may show varying degrees of adherence due to historical, social, and contact-induced changes. This is a "Weak Saussure" position: not replacing arbitrariness entirely, but demonstrating systematic non-arbitrary undercurrents. The framework predicts tendencies, not absolutes. Languages exist on a spectrum between arbitrary and embodied, with different historical pressures producing different degrees of alignment. Section 5 and Section 6 report how far the tendency holds once genealogical relatedness is controlled. The answer is: it holds, and it is much smaller than the raw figure suggests. 1.3 Answering the Call for a Systematic Framework The present work does not emerge in a vacuum. Over the past two decades, multiple scholars have explicitly called for exactly the kind of systematic theoretical framework this research provides. At least seven major research teams have identified gaps that ad-hoc studies of individual sound-meaning correspondences cannot fill. 1.3.1 Johansson, Anikin, Carling, and Holmer (2020) Johansson, N., Anikin, A., Carling, G., & Holmer, A. (2020). The typology of sound symbolism: Defining macro-concepts via their semantic and phonetic features. Linguistic Typology, 24(2), 253-310. They write that despite progress in the field, understanding of sound symbolism and its mechanisms "remains patchy," and that one way of bridging the gaps is large-scale cross-linguistic comparison. What they provided: analysis of 344 concepts across 245 language families; 20 macro-concepts as organizational principles; four types of sound symbolism. The gap that remains: their typology categorizes types of sound symbolism but does not provide a cognitive mechanism explaining why speakers prefer