📄 A Celtic Etymology for Old English claedur, 'clapper'

Summary

Andrew Breeze takes up the unclear origin of the Old English word claedur. It survives only in the oldest English glossaries: Épinal has claedur, Erfurt cledur, and Corpus cleadur. There it translates Latin crepacula and is explained as a wooden board for driving birds away from crops. Earlier scholars sought Germanic connections, such as links to clatter, to the glosses clidrenn, or even to Gothic klismo, but these remained unconvincing.

Breeze proposes that the word was borrowed from Welsh cledr/cledyr, meaning 'stave, rod, rail'. He traces this word through Breton and Irish relatives, early Welsh verse, medieval law texts, the Peredur tale, Dafydd ap Gwilym and a sixteenth-century dictionary. Across these sources the sense of a wooden bar or handle stays consistent. He then shows that the sound changes involved are plausible.

His conclusion is that claedur has nothing to do with clatter. It first meant a stave fitted with clappers, and it adds another Celtic loanword to the earliest English glossaries.

Key points

• The word occurs in three spellings in three glossaries. Related compounds, such as cladærsticca and claderstecca ('clapper stick'), gloss anate, which is possibly a corruption of Latin amite.
• Welsh cledr is related to Middle Breton clezr, Modern Breton klerenn and Middle Irish clithar. In Welsh it is attested from verse of about 800 down to William Salesbury's 1547 dictionary.
• Proposed phonology: Welsh e sometimes entered Old English as æ (compare Æsce, the river Axe). Corpus cleadur shows back mutation, and the compound developed as clædur > *cladur > cladær-.
• Other Welsh loans in the early glossaries are trymide (Corpus 577), sercae (Épinal-Erfurt 18), loerge (Épinal 1) and uuannan (Épinal 576). They come mostly from military and tool vocabulary.
• The text itself cites Selim 5 (1996), pp. 119–121, which does not match the 2019 date in the library entry.

Interesting ideas

• I1. Breeze supports a word origin by laying out dated attestations across many centuries, checking that the core meaning holds steady rather than relying on surface resemblance.
• I2. Spelling variants across manuscripts are tied to one lexeme through named, explicit sound rules (back mutation, retraction) rather than loose similarity.
• I3. The author states the strongest counterexample to his own theory (the a in cladær-) and handles it directly instead of hiding it.
• I4. He treats the original glossator as a witness who can be wrong, since amite (a bird-catching rod) was taken to mean a bird-scarer.
• I5. He orders word forms by derivation, taking the simple form as original and the compound with -sticca as a later stage.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (world): When linking a word to a WordNet/Wikidata sense, prefer candidates whose meaning stays consistent across several independent sources, and record that evidence so the choice can be explained.
• I2 (tokens): The lemmatizer could map variant or nonstandard spellings to a lemma only through a named rule, so every merge is traceable and can be undone.
• I3 (learn): Self-induced repair rules should carry a list of their hardest known counterexamples, used as negative controls during evaluation.
• I4 (learn): When building silver data from several witness parsers or LLM teachers, model each source's typical misreadings instead of trusting any single annotation.
• I5 (world): A "simpler form came first" heuristic could help order derived entities or events when no explicit time markers are present, with the assumption logged as defeasible.