The Indo-European and Austronesian language families may share a remote common ancestor—a relationship that would connect enormous regions of human language and history.
The 2022 comparison list gives that idea a finite research program. Source every reconstruction inside its own family, derive regular sound correspondences from one portion, freeze the rules, and use them to predict lexical and grammatical material held aside from discovery.
Human languages reuse compact sound inventories, shift meanings, borrow words, and generate chance resemblances. The comparative method turns those realities into controls and gives the proposed relationship a path toward new knowledge.
Ground the 2022 Seed List in Comparative Dictionaries
The original comparisons:
| Proposed meaning | Indo-European-side form as written | Austronesian-side form as written | Working reading |
|---|---|---|---|
| three | *treys | *telu | Both forms have dictionary trails; test the correspondence |
| hand / five | *ronk | *lima | Remove this row as written |
| flow | *ser | *qalur | Attach exact reconstructions and reflexes |
| skin | *skend | *qanic | Attach exact reconstructions and reflexes |
| walk / foot | *stembh | *qaqay | Attach exact reconstructions and reflexes |
| smoke / ash | *smew | *qabu | Attach exact reconstructions and reflexes |
| watch / day | *serw | *qalayaw | Attach exact reconstructions and reflexes |
| salt | *sal | *qasira | Attach exact reconstructions and reflexes |
| talk / mouth | *bhas | *baqbaq | Attach exact reconstructions and reflexes |
| bow | *bhewgh | *busur | Attach exact reconstructions and reflexes |
| give birth / woman | *bher | *bay | Attach exact reconstructions and reflexes |
| two | *dwoh₁ | *dusa | Attach exact reconstructions and reflexes |
| sky / upward | *dyews | *daya | Attach exact reconstructions and reflexes |
| river / lake | *danu | *danaw | Attach exact reconstructions and reflexes |
| tongue | *dn̥ǵʰwéh₂ | *dilaq | Keep *dilaq as “lick” and rebuild the semantic route |
| wide | *delh₁ | *dempad | Attach exact reconstructions and reflexes |
| I | *eǵoh₂ | *aku | Attach exact reconstructions and reflexes |
| think | *men | *nemnem | Attach exact reconstructions and reflexes |
| name | *h₁nómn̥ | Proto-Polynesian *hingoa | Compare at the Proto-Polynesian level only |
The asterisks preserve the notation of the initial proposal. Each row gains an exact source, proto-language level, daughter reflexes, and within-family derivation.
Modern comparative lexicons immediately improved the seed. The Indo-European Lexicon lists no ronk reconstruction for “hand.” The Austronesian Comparative Dictionary distinguishes PAN lima, “five,” from PAN qalima, “hand,” so the original hand/five row crosses glosses. The dictionary also gives PAN dilaq as “to lick” rather than “tongue,” turning that relation into a separately documented semantic history. The “name” comparison belongs explicitly at the later Proto-Polynesian level rather than Proto-Austronesian depth.
Keeping these corrections visible shows how the research program strengthens itself. Every surviving pair receives provenance and reconstruction depth before cross-family comparison.
The search procedure now begins from a fixed concept, retrieves published reconstructions from each family’s reference works, follows daughter reflexes, and only then compares forms. A bridge such as “lick” to “tongue” becomes a historical claim with its own evidence.
Establish Each Side Within Its Family
Every cross-family row begins with two complete within-family records:
- exact published reconstruction and notation;
- proto-language level;
- daughter forms supporting the reconstruction;
- regular sound changes connecting daughters to the proto-form;
- attested or reconstructed meaning without convenient substitution;
- known loans, competing etymologies, and disputed analyses.
A Proto-Polynesian form stays at its documented level. A “lick” root reaches “tongue” through a sourced semantic path. A reconstruction absent from the literature cannot support a cross-family comparison.
The Austronesian Comparative Dictionary provides particularly strong infrastructure because it distinguishes reconstruction levels and separates canonical comparisons, near comparisons, loans, and chance look-alikes.
This stage concentrates the seed list into its strongest sourced comparisons.
Candidate Sound Laws Create Predictions
The seed list suggested four candidate correspondences:
- Indo-European s with Austronesian q;
- r with l;
- aspirated bh with b;
- d with d.
A remote historical relationship should produce repeated, conditioned sound changes across position, neighboring sounds, stress, and daughter-language chronology.
The candidate s : q came from the seed and now becomes a rule to test on new material. A conditioned correspondence table distinguishes initial, medial, and final position; neighboring vowels and consonants; stress; and sound changes already established within each family.
Every candidate s : q comparison answers six questions:
- How many forms in the same sound environment follow another correspondence?
- Does s map to unrelated Austronesian sounds under changing conditions?
- Does q map to several Indo-European sounds without a conditioning rule?
- Does the proposed semantic path fit known historical change?
- Can contact or borrowing account for the form?
- Do inflection, pronoun, numeral, or other grammatical systems preserve the pattern?
Once frozen, the rules can predict comparisons that played no role in inventing them.
Count the Complete Search
The initial probability calculation examined seven apparent s : q matches among short roots. It produced a low probability by treating the correspondence as though selection occurred before inspection.
The actual discovery process offered multiple sound pairings, semantic paths, and candidate roots. Words also have statistical relationships, and phoneme inventories form structured systems rather than uniform consonant bags.
A predeclared held-out procedure captures that full search:
- Freeze a sourced list of meanings and reconstructions.
- Derive candidate sound laws from one portion only.
- Score predictions on untouched meanings.
- Process unrelated family pairs through the same method.
- Penalize semantic flexibility and one-off exceptions.
- Repeat under competing reconstructions.
The result measures whether Indo-Austronesian rules predict withheld structure more accurately than unrelated family pairs and alternative accounts.
A Family Relationship Predicts More Than Vocabulary
A relationship older than both families can leave several connected traces:
- regular sound changes across a system;
- shared morphology resistant to borrowing;
- pronoun and numeral patterns with consistent transformations;
- reconstructable intermediate forms;
- chronology and migration compatible with archaeology and population history;
- stronger prediction when the two families enter one model.
Controls make the comparison decisive. Correspondences should extend beyond the selected pairs, grammar should add matching structure, and unrelated family pairs should score lower under the same procedure.
Together these outputs can turn a lexical resemblance into a model of deep linguistic history.
The Proposal Invites a Comparative Project
A conjecture creates value when another researcher can repeat and extend its work.
The seed list now performs a clear job. Two rows leave in their original form. Candidate correspondences stay tied to the material that produced them. New lexical and grammatical data remain held out until the rules freeze.
This sequence gives historical linguists more than a resemblance between three and telu. It gives them sourced data, explicit candidate laws, documented corrections, and a prediction procedure.
The Indo-Austronesian hypothesis can now meet specialists as a real comparative-linguistics project with the potential to illuminate one of humanity’s deepest language relationships.