Words We Share

Research working paper · v0.2

A Shared Second Language for Humanity

Design principles and a testable research proposal inspired by The Loom of Language

Conscious Observer · 16 October 2026

Download the original PDF

Abstract

This paper proposes a voluntary research program for a shared auxiliary second language: a language people could learn for communication across linguistic communities while continuing to use and develop their existing languages and pass them on to others. Inspired by Frederick Bodmer's The Loom of Language, edited and arranged by Lancelot Hogben, it combines a historical reassessment with a small, testable design. The accompanying prototype contains 120 base entries, a limited grammar, and sample exchanges. Its vocabulary draws on a wider range of proposed language sources than the book's predominantly Greek and Latin approach.

Evidence establishes that language barriers impose real burdens in some settings, including scientific participation. It does not establish that a newly constructed language would secure widespread adoption, be equally easy for all learners, or reduce war. Those outcomes require distinct evidence. The proposal therefore separates design objectives from demonstrated results and evaluates a new language against existing auxiliary languages, multilingual education, and translation tools. The immediate aim is an experiment in direct understanding that people can use and evaluate. Preserving existing languages and ensuring informed participation are part of that aim.

1 Babel and the possibility of understanding

In Genesis 11:1-9, people who share a language undertake a collective building project; the confusion of their speech and their dispersal interrupt it. The story raises a question at the heart of this project: what becomes possible when people can understand one another? It can also invite reflection on the ambitions and responsibilities that accompany collective power. Babel provides a cultural and theological opening here; readers need not share a historical or religious interpretation of the story. [2]

The concern behind this proposal is that failures of understanding can make cooperation harder and leave room for suspicion. That concern motivates the inquiry. It does not establish how much human conflict results from misunderstanding. Understanding words, understanding another person's situation, and agreeing with that person are different achievements. A shared language might help people discover common interests; it might also make an existing disagreement clearer. Institutions, the distribution of power, and what people seek to achieve through communication still matter.

The Loom of Language itself makes this distinction sharply. Its concluding page explicitly rejects the idea that language reform alone can abolish war and identifies powerful social causes beyond linguistic misunderstanding. It nevertheless argues that planned language could assist collaboration on a planetary scale. Preserving both parts of that argument gives the project a hopeful foundation without promising that language design will bring peace. [1, p. 511]

2 What the historical proposal actually says

The book was first published in January 1944. The supplied scan lists a fifth impression in 1949. Its title page credits Frederick Bodmer as author and Lancelot Hogben as editor and arranger. Hogben's foreword describes substantial collaboration, and a footnote to the final chapter identifies its views as the outcome of joint discussion. That footnote also points to Hogben's Interglossa, whose independently consulted front matter confirms publication in 1943. Our prototype is a new proposal inspired by these works; it is not a recovered dictionary from Bodmer or an authorized continuation of Interglossa. [1, title and publication pages, pp. 11-13, 481; 3]

The central social proposal is to add a shared language while retaining existing ones. The opening of Chapter XII anticipates the persistence of regional languages and proposes a common second language for communication among their speakers. It explicitly recognizes the political character of achieving agreement on a shared educational program. This paper retains the additive aim while replacing immediate universal implementation with voluntary, staged experimentation. [1, p. 481]

The book proposes several design principles: avoid irregular word changes; make word order intelligible; use relatively familiar word material; reduce unnecessary memorization through reusable elements; and provide regular spelling. These principles do not amount to a finished language specification. On printed page 493, the book identifies unresolved research questions about qualifier placement, interrogation, emphasis, and grammatical signposts. A rigid subject-verb-object grammar presented as an already settled blueprint would overstate the source. [1, pp. 485-493]

The discussion of vocabulary economy is more careful than a simple demand to eliminate verbs. The book distinguishes reducing the number of different words from reducing the length of utterances. It notes that inventing a cumbersome definition can be harder than learning an additional word. After discussing Basic English's sixteen operators, it warns that a new auxiliary should not import their English idioms unchanged. These qualifications inform our decision to retain 25 useful verbs rather than force every action into a tiny set of general operators. [1, pp. 500-506]

The historical vocabulary recommendation is predominantly Greek and Latin, justified by their circulation in science and technology. The book itself acknowledges this Western origin and argues against a vocabulary assembled merely to give language communities proportional representation. Our approach, drawing on a broader range of roots, reopens that question. Familiarity, sound structure, recognition, semantic clarity, and community acceptability must be investigated together. Including words from several places does not by itself establish fairness. [1, pp. 494-499]

3 Evidence and limits of the contemporary case

Language barriers can exclude capable people from useful exchanges. Amano and colleagues surveyed 908 researchers in environmental sciences across eight nationalities and reported greater effort associated with several scientific activities for non-native English speakers, including reading, writing, and conference participation. These findings give us concrete reasons to take linguistic burdens seriously. Its self-reported, disciplinary sample is not an estimate of the burden on every person worldwide, and it does not test a constructed-language remedy. [4]

A shared auxiliary language could redistribute some language-learning burdens by requiring participants to learn a common additional medium. That is a design hypothesis. Learners would still differ in literacy, prior languages, educational opportunity, disability, free time, and familiarity with the chosen vocabulary and script. A language without a national owner can still confer advantages on particular groups. Inventing a language does not make it neutral. We need to define what fairness would require, measure the resulting burdens, and establish ways to address them.

Simpler morphology is a reasonable experimental starting point, but fewer markers do not automatically imply better communication. In two artificial-language experiments with adult native English speakers, Kenanidis and colleagues found that redundant case marking could hinder initial learning relative to a non-redundant condition; the salience of the cue affected the result. The authors also discuss research in which learner age, first language, and learning stage complicate the picture. It supports testing streamlined grammar, not declaring every grammatical redundancy useless. [5]

What counts as familiar word order depends on the learner. The World Atlas of Language Structures documents subject-object-verb, subject-verb-object, and several other patterns, as well as languages without a dominant order. Our choice of subject-verb-object is a practical first specification. It has no demonstrated universal superiority. Testing should measure its costs for learners with different language backgrounds, including people whose usual writing systems are not Latin-based. [6]

Conflict is a separate research question. Fearon and Laitin's influential study of civil war reports that greater ethnic or religious diversity was not associated with greater civil violence after controlling for income in its analysis. This is not a direct test of linguistic mutual intelligibility or of a language intervention, and the accessible publisher abstract cannot settle the larger literature. The study illustrates why an account of conflict needs to explain how a proposed cause operates and consider other explanations. This proposal makes no numerical claim about the fraction of conflict attributable to language. [7]

Multilingual education and translation are essential comparators. UNESCO's 2025 guidance emphasizes education in languages learners understand and the value of multilingualism. The NLLB Team's 2024 research demonstrates machine translation across 200 languages while describing persistent data inequalities. Both sources strengthen the case for access through existing languages and careful evaluation of translation tools. A constructed auxiliary must earn a useful role alongside those efforts. The cited translation study is an illustrative research milestone, not a claim about the best available product in 2026. [8, 9]

4 Design commitments for a shared second language

The first commitment is preservation. Participation should add a communicative option without requiring people to abandon a home, heritage, regional, national, or signed language. The term universal describes an aspiration to broad availability and access. It does not justify mandatory uptake. Schools and public services should not withdraw existing-language support on the strength of an untested prototype.

The second commitment is explicitness. A small grammar should state how participants ask questions, negate a claim, express time, identify who acts on whom, and repair a misunderstanding. Lexical entries need bounded meanings and examples. The goal is successful communication at a manageable learning cost. A smaller dictionary or fewer rules help only if they serve that goal.

The third commitment is empirical fairness. Candidate words may draw on several linguistic traditions, but the selection process should examine how much work those words demand of learners. A mixture of origins is not enough. Loanwords already cross language-family boundaries; crediting an immediate source does not establish an ultimate etymology. Community review is necessary before promotional claims about representation. Meanings that are culturally sensitive or linked to sacred practices require particular care and consultation.

The fourth commitment is revisability. Every published version should identify its vocabulary, grammar, provenance, unresolved questions, and changes from earlier versions. Early learners should have stable teaching materials during a study. Revisions should follow disclosed criteria, with explanations for retaining, replacing, or redefining words. A small founding team can maintain a prototype; it cannot credibly claim to speak on behalf of humanity.

The fifth commitment is accessibility. The first written and spoken prototype uses Latin letters to keep the initial test manageable. That choice creates a literacy burden for some learners. Teaching materials should therefore explain sounds independently of English spelling, use illustrations and audio, and offer accessible text alternatives. Signed languages are full languages, not versions of spoken languages made with the hands; development of any signed counterpart would require its own Deaf-led design and evaluation. The present spoken prototype makes no claim to supply one.

5 The 120 word prototype

Prototype A01 accompanies this paper as a small demonstration. It contains 30 function words, 10 digit words, 25 verbs, 35 nouns, and 20 qualities. These 120 base entries are a starter inventory, not an estimate of the vocabulary needed for ordinary adult life or scientific work. Compounds and proper names are not counted as additional base entries. No participant has yet learned or tested this system, and no fluency result is claimed.

Words keep their form across people, number, and tense. A basic clause follows subject, optional time marker, optional negation, verb, object, and any relation phrases. Pa marks past time and fu future time. Na negates the predicate. An unmarked clause takes its time from context or expresses a general statement. Ka introduces a yes-or-no question without changing word order. Every complete predicate includes a verb; isa links a subject to a property or identity.

For example, Mi soma luga means I study a language. Mi pa soma luga places the study in the past. Ka ti voli soma luga? asks whether the listener wants to study a language. The form soma remains unchanged. Mi dona kitabu a ti means I give a book to you; the relation word a identifies the recipient. These are demonstrations of specified rules, not evidence that the rules are easy for all learners.

The starter lexicon explores cues associated with several traditions, including Swahili, Arabic, Indonesian, Mandarin, Japanese, Tamil, and European languages. Some forms are original mnemonic choices; others reshape candidate source forms to fit the proposed sound pattern. Each entry records its design cue and whether that cue has been checked. Most candidate etymologies remain provisional, and none has received a community endorsement. This project assigns each prototype word a fixed meaning; the source word may have a different range of meanings.

The prototype uses five vowel letters and avoids consonant clusters in its base entries. Each written vowel is a separate syllable nucleus, and stress falls on the next-to-last vowel, or the only vowel in a one-syllable word. These choices can be stated compactly, but recognition and pronunciation must still be tested. Short function words are especially vulnerable to confusion. The companion specification documents the sound values, sentence patterns, verb frames, examples, and limitations so that others can examine and test the proposal.

Compounding is controlled. Domu-soma is provisionally registered as a learning building or school, with the generic head first. Its intended meaning must be taught and tested; a transparent-looking combination is not guaranteed to be interpreted correctly. The prototype does not yet define general relative clauses, nested quotations, evidential systems, technical terminology, or a complete arithmetic vocabulary. A successful first test would justify further development, not universal deployment.

6 Testing usefulness before making adoption claims

A learner may remember a rehearsed sentence without being able to use its rule in a new situation. Evaluation needs to distinguish those achievements. The first study should use low-stakes tasks: requesting an object, identifying who gave it to whom, distinguishing a statement from a question, reporting an event as past or future, and correcting an intentionally ambiguous instruction. These demonstrations do not establish suitability for medical, legal, or emergency use.

Before recruitment, define the protocol, instructional time, stimuli, outcome measures, exclusions, and analysis plan. Recruit adult volunteers across several first-language and literacy backgrounds, and record prior multilingual experience rather than treating nationality as a substitute for language. Use an initial feasibility study to estimate variability and practical problems; determine the sample size for a confirmatory study through a prospective power analysis. No particular number of volunteers is defended as statistically sufficient in this paper.

Different comparisons answer different questions. A prototype-versus-Esperanto condition would examine two learning systems under equal instructional time. A prototype-versus-translation-assisted condition would assess practical task performance under specified equipment and connectivity. A comparison with simple existing-language instruction would need careful adjustment for what participants already know. Recruitment, allocation, training materials, and evaluator blinding should be designed to minimize favoritism toward the new system.

Primary outcomes should include accurate completion of unseen communication tasks and retention after a delay. Secondary measures can include learning time, requests for clarification, confidence, pronunciation confusions, and participants' willingness to continue. Report uncertainty, and show outcomes separately for relevant learner backgrounds. Overall averages can conceal a system that works well for one group and poorly for another. Measure errors in negation, participants, quantities, and time explicitly rather than relying on a general satisfaction score.

The social hypothesis needs its own study. After linguistic feasibility is demonstrated, cooperative tasks could examine whether a shared auxiliary changes the quality of interaction, perceived inclusion, or willingness to work together. Such measures would still not establish an effect on war. Participants should be able to disagree with one another and with the project; successful communication includes accurately expressing disagreement.

The program needs clear rules for revision and for deciding when to stop. A candidate word that repeatedly causes confusion should be changed. A grammar that produces larger learning burdens for particular groups needs redesign. If the prototype offers no useful advantage over existing tools, documenting that result is valuable. A credible research program must allow an unsuccessful result to change its direction.

7 Adoption and public communication

People need occasions and partners for using what they learn. Adoption therefore depends on coordination as well as education. The first plausible communities are small groups with a shared activity, such as an international reading group or a collaborative learning exercise. Their needs should guide vocabulary growth. Claims about governments, schools, or international institutions adopting the language would be premature.

Governance can begin with a public revision log, documented selection criteria, multilingual feedback, and a clear distinction between proposals and accepted entries. As participation grows, meaningful representation will require language educators, speakers from a wider range of backgrounds, accessibility specialists, and people willing to criticize the project. Contributors should be credited for actual work; a geographically diverse word list cannot substitute for participation in decisions.

8 A defensible ambition

The strongest case for this project is practical and open-ended. Language barriers create real difficulties; a shared second language is one possible response worth investigating alongside alternatives. The Loom of Language contributes a valuable insistence that language learning and language planning should serve ordinary people. Its specific historical assumptions remain open to criticism and revision.

The next step is an experiment that others can examine: an intelligible grammar, a documented vocabulary, accessible teaching material, and evidence about who benefits and at what cost. If people with different backgrounds can use a modest common language to cooperate more directly, that would be meaningful. A broader promise must be earned through wider participation and stronger evidence.

Methods and acknowledgments

Conscious Observer is the pen name used for this working paper. The project acknowledges Frederick Bodmer as author of The Loom of Language and Lancelot Hogben as its editor, arranger, and substantial collaborator. It also acknowledges the researchers, lexicographers, and institutions cited below. Attribution does not imply their endorsement of this proposal.

The external research addresses particular claims; it is not a systematic literature review. The review used retrieved full articles or relevant passages for the studies on scientific participation, learning, and translation; the civil-war discussion is limited to its publisher abstract. Interglossa verification covers front matter only, not its complete grammar. The initial lexical audit checks a subset of cues; all remaining entries are labeled pending. This is a research proposal and working prototype, not a peer-reviewed paper or a validated international language.

References

  1. 01

    Bodmer, F. (1944; supplied scan lists fifth impression, 1949). The Loom of Language: A Guide to Foreign Languages for the Home Student. Edited and arranged by L. Hogben. George Allen & Unwin. Local primary source: 2015.283471.The-Loom_text.pdf. Printed page numbers are used in citations; printed pp. 481-511 are PDF pp. 495-525. View source

  2. 02

    Genesis 11:1-9. New Revised Standard Version Updated Edition (2021). Passage consulted through Bible Gateway. Used as cultural and theological framing. View source

  3. 03

    Hogben, L. (1943). Interglossa: A Draft of an Auxiliary for a Democratic World Order, Being an Attempt to Apply Semantic Principles to Language Design. Penguin Books. Digitized facsimile hosted by Glosa; first eight PDF pages inspected. View source

  4. 04

    Amano, T., et al. (2023). The manifold costs of being a non-native English speaker in science. PLOS Biology, 21(7), e3002184. Survey of 908 environmental-science researchers; self-reported burdens. View source

  5. 05

    Kenanidis, P., Llompart, M., Fernandez Santos, S., & Dabrowska, E. (2024). Redundancy can hinder adult L2 grammar learning: evidence from case markers of varying salience levels. Frontiers in Psychology, 15, 1368080. Artificial-language experiments; scope is early adult learning. View source

  6. 06

    Dryer, M. S. (2013). Order of Subject, Object and Verb. In M. S. Dryer & M. Haspelmath (eds.), The World Atlas of Language Structures Online, chapter 81. Max Planck Institute for Evolutionary Anthropology. View source

  7. 07

    Fearon, J. D., & Laitin, D. D. (2003). Ethnicity, Insurgency, and Civil War. American Political Science Review, 97(1), 75-90. Publisher abstract consulted; no direct auxiliary-language intervention. View source

  8. 08

    UNESCO (2025). Languages matter: Global guidance on multilingual education. Official publication overview consulted; not the full report. View source

  9. 09

    NLLB Team et al. (2024). Scaling neural machine translation to 200 languages. Nature, 630, 841-846. Technical research milestone; not a 2026 product comparison. View source

  10. 10

    Zamenhof, L. L. Fundamento de Esperanto, English grammar. Online primary text maintained by Akademio de Esperanto. Relevant to the claim audit: regular inflections, compounding, and international vocabulary. View source

  11. 11

    NSW Department of Education and Training (2008). Bilingual Dictionary for ESL Beginners: English/Swahili. ISBN 9780731386451. Selected lexical entries consulted; not an etymological dictionary. View source