Canonical edges, each carrying form and meaning history.
THE CHANGE ATLAS
WORD FAMILIES ↗EVERY REVIEWED EDGE / TWO HISTORIES EACH
FORM → MEANING → EVIDENCEHOW WORDS MOVE.
HOW MEANINGS MUTATE.
A global atlas of every reviewed transformation in the graph—from lost vowels and borrowed spellings to meanings narrowed, inverted, elevated, and remade.
EVERY TRANSFORMATION
IS A CLAIM.
Every edge names the linked forms, relationship, sound history, semantic history, scholar, confidence, chronology when known, and the sources that warrant it. Search the language of change without flattening it.
No reviewed edge is allowed to omit what changed in form.
No reviewed edge is allowed to omit what changed in meaning.
515 distinct sources support the edge ledger.
Named analyses remain attached instead of becoming anonymous fact.
Borrowing, inheritance, derivation, uncertainty, and more stay distinct.
SEARCH THE TRANSFORMATIONS
FOLLOW CHANGE ITSELF.
The atlas searches the exact reviewed prose and authoritative edge labels. It does not manufacture universal “sound laws” or semantic categories by guessing from descriptions.
SEMANTIC HISTORY
1025 EVIDENCE-BACKED CHANGES
No deeper parent is promoted for the Latin prefix in this dossier.
The morphological branch stops at the limit of the cited headword analysis.
CONFIDENCE NOTEA source can identify a component without supplying its complete prehistory.
No direct parent is promoted beyond Old English wīf in the reviewed dossier spine.
The trail stops while probable cognate comparisons remain labeled as relatives.
CONFIDENCE NOTEMerriam-Webster calls the Tocharian comparison probable rather than supplying a secure direct ancestor.
No older form is promoted beyond reviewed *man-¹.
The human-being trail ends before unsupported root-merging.
CONFIDENCE NOTEAmerican Heritage prints man-¹ and men-¹ as distinct entries.
No older form is promoted beyond reviewed *dheh₁(i)-.
The trail ends where the published comparative root ends.
CONFIDENCE NOTEThe root is reconstructed rather than an attested first word.
No older form is promoted beyond Latin mās in the reviewed headword sources.
The trail stops without inventing a reconstruction or linking unrelated mal-.
CONFIDENCE NOTEA missing deeper parent is represented explicitly.
No older form is promoted beyond reviewed *dhghem-.
The trail ends where the published comparative root ends.
CONFIDENCE NOTEThe root is reconstructed rather than an attested first word.
No older form is promoted beyond reviewed *h₃ekʷ-.
The visual-root branch ends where the published comparative analysis ends.
CONFIDENCE NOTEThis deep boundary does not erase the earlier uncertainty between Latin, Etruscan, and Greek.
No deeper parent is promoted for pros- in this reviewed component branch.
The side trail stops at the current evidence limit.
CONFIDENCE NOTEThe person dossier reviewed the headword analysis, not the full independent history of Greek pros-.
No deeper parent is promoted for Old English hlāf in this reviewed dossier.
The loaf branch stops while the separately modeled guardian branch continues.
CONFIDENCE NOTEThe headword sources establish the compound without supplying a reviewed deeper loaf ancestry.
No deeper parent is promoted for Old English hlāf in this reviewed dossier.
The loaf branch stops while the separately modeled kneading branch continues.
CONFIDENCE NOTEThe headword sources establish the compound without supplying a reviewed deeper loaf ancestry.
No older form is promoted beyond reviewed *bheuhₓ-.
The trail ends where the published comparative root ends.
CONFIDENCE NOTEThe root is reconstructed, not an attested first word.
No direct parent is promoted beyond Old English wīf.
The trail stops while a probable Tocharian comparison remains explicitly non-parental.
CONFIDENCE NOTEThe reviewed sources do not supply a secure direct ancestor.
No established parent is promoted beyond Middle English girle.
The secure trail ends before an uncertain Old English and garment comparison.
CONFIDENCE NOTEAmerican Heritage labels the origin unknown; the Middle English Dictionary marks its proposals with question marks.
No older form is promoted beyond reviewed *gʷou-.
The proposed trail ends where the published cattle root ends.
CONFIDENCE NOTEThe final boundary does not resolve the disputed connection back to boy.
No older form is promoted beyond reviewed *wer-³.
The guardian trail ends where the published comparative root ends.
CONFIDENCE NOTEThe root is reconstructed rather than attested.
No older form is promoted beyond reviewed *dheig̑h-.
The kneading trail ends where the published comparative root ends.
CONFIDENCE NOTEThe root is reconstructed rather than attested.
No deeper parent is promoted for Old Norse hūs in this reviewed dossier.
The house component stops while the householder component carries the reviewed deep trail.
CONFIDENCE NOTEThe headword research did not review the full independent history of house.
No shared reconstructed parent is promoted beyond the wīf and kwīpe comparison.
The probable comparison remains open at the present evidence frontier.
CONFIDENCE NOTEThe reviewed sources do not supply a direct ancestor.
No older form is promoted beyond proposed *gyrela.
The reconstructed predecessor remains an open hypothesis.
CONFIDENCE NOTEThe branch is tentative from its first edge.
No secure relationship is promoted beyond the Old English garment comparison.
The garment hypothesis stops at its evidential limit.
CONFIDENCE NOTEThe comparison is not an established ancestor.
No securely established parent is promoted beyond reconstructed *hlaibaz.
The Germanic bread word reaches an evidential boundary; a possible substrate comparison remains separately graded proposed.
CONFIDENCE NOTEAmerican Heritage says perhaps when comparing a common European substrate source.
No reconstructed direct parent is promoted beyond attested Old High German wīb.
The historical woman and wife word reaches an unresolved origin rather than a fabricated proto-form.
CONFIDENCE NOTEDuden explicitly labels the origin unclear and offers two possibilities only as perhaps.
No older form is promoted beyond reviewed *dhghem-.
The humble trail ends where the published comparative earth root ends.
CONFIDENCE NOTEThe root is reconstructed rather than attested.
No older form is promoted beyond reviewed *dhghem-.
The humus trail ends where the published comparative earth root ends.
CONFIDENCE NOTEThe root is reconstructed rather than attested.
READ THE EDGE, NOT JUST THE WORD
TWO HISTORIES TRAVEL TOGETHER.
Pronunciation, spelling, morphology, loss, insertion, adaptation, and regular historical development.
Narrowing, broadening, metaphor, social reanalysis, technical specialization, and meaning retained.
Attested forms, reconstructed forms marked by *, and explicit knowledge boundaries never share one certainty label.
Scholar, confidence, chronology, source stance, locator, and evidence note remain attached to the atomic edge.