User:Mellohi!/EDPC errors

Ranko Matasović's Etymological Dictionary of Proto-Celtic, despite its key importance in general Indo-European historical linguistics, is chock-full of errors and omissions. Not even his corrigenda managed to catch many of the mistakes. Due to these errors and gaping holes in the book, many Celticists hold it with a critical or indifferent eye and it is recommended to also study Celtic phonological and morphological history from other sources to avoid reproducing EDPC's problems blindly.

Specific entry errors

• *abalo-
  • EDPC fails to explain the u-infection in Old Irish ubull (“apple”).
  • Any influence of the nominative singular of *h₂ébōl is dismissed. Stifter has used the nominative singular to explain the u-infection via a chain of declension class shifts.
• *ank-o-
  • An EDPC entry that is very infamous among Celticists. In an attempt to avoid deriving it from a Narten present, Matasović tries to explain the controversial vowel in the Old Irish verbal root ·icc as being rebracketed from ní (“not”).
  • The morphological problems in this solution are obvious. The particle ní is not known to cause the elision of the initial vowel of the verb after it, for starters.
• *barag(en)o/ā
  • In an attempt to shoehorn Welsh bara and Old Irish bairgen (“bread”) under the same proto-form, Matasović forgets a very basic Old Irish morphonological rule where consonant clusters caused by syncopated -a- or -o- are depalatalized. Under his reconstruction, the palatalization in bairgen makes zero sense.
• *danto-
  • Erroneously classifies Old Irish dét as a neuter o-stem instead of a neuter nt-consonant stem.
  • This is a really bizarre error from Matasović here; both DIL and GOI will tell you in your face that dét is not a neuter o-stem.
  • Matasović provides no explanation for the Old Irish neuter nt-stem declension; perhaps, given his classification error in the first place, he doesn't know or believes it exists!
• *do
  • Just plain bad. Matasović gets the vowel wrong (should be *dū) and leaves out all Brittonic descendants (even though they very well exist) for no discernable reason.
• *du(s)bwuyo-
  • Typo, should be *du(s)bwiyo-.
• *ēskyo-
  • Matasović has no explanation for why the *ē failed to break into ía in Old Irish éscae despite the -sc- remaining non-palatal.
• *fotlo-
  • Matasović fails to provide a satisfactory explanation of ól's apparent disyllabic alternative form oöl, despite listing it in his entry. His etymology assumes that it was always supposed to be monosyllabic.
• *freswo-
  • Matasović has no explanation for how réud has no f/ph in it despite *-sw- usually yielding this in Old Irish.
• *gʷrīns-/gʷrenso-
  • Erroneously connects Old Irish grían (“sun”) to this reconstruction by reconstructing *gʷrensnā, expecting somehow for this -en- to yield -ía- before broad consonants despite this diphthongization consistently never affecting Old Irish é produced from -en- (see also *sentus yielding sét, kentu- yielding cét-).
  • An extremely bizarre error from Matasović here, as GOI (which is cited dozens of times in EDPC, showing that he read it himself) explicitly states that Proto-Celtic -en- can't produce -ía-. This fact was known a century ago!
• *kanta-bwi-yo-
  • For no reason, EDPC reconstructs the base of this verb as bwiyeti while also reconstructing *buyeti for the substantive verb that this verb is clearly based on. (KPV goes with *bwi-yo-)
• *kasni-
  • It is impossible for -ach in ceinach to come directly from the derivational suffix *-ākos, which instead actually yields -og in Welsh.
• *knokko-
  • Welsh cnwch makes reconstructing -o- impossible. The correct reconstruction is *knukkos.
• *leng-o-
  • Matasović provides no explanation for the -b- in lingid's reduplicated stems.
• *liro-
  • Matasović makes various illogical semantic arguments in support of his novel theory that it must have been borrowed from a substrate language. First, he rejects the usual theory of derivation from *leyH- (“to flow”) because, in his words, the "'sea' is not 'flowing' in any possible sense" (???). Then, he also argues that it must be borrowed just because the native word *mori is a synonym (?????).
• *menman-
  • Deriving Old Irish menmae from a neuter n-stem simply does not work; you'd get *meinm instead.
• *snā-
  • snám (“swimming”) is a u-stem, not an o-stem.
• *trowgo-
  • Matasović constructs a bizarre athematic compound *trowg-karos to account for the /k/ in Old Irish trócar (“merciful”), even though a normal thematic compound *trougo-karos works fine and allows a perfect match with Welsh trugar. The Irish form can be generated via normal syncope rules from a thematic compound, and for Welsh the medial -g- of *trougo- is regularly deleted in Brittonic, leaving no issues with a derivation from a thematic compound.
• *wasto-
  • Matasović has no idea how the -o- generally reconstructed in *uɸostos became -a- in Brittonic and Gaulish. But he fails to notice that Schrijver's Studies in British Celtic Historical Phonology, a book he cited, has an explicit explanation for those vowel changes (in Brittonic, wo- split into wo- when unlenited and wa- when it was lenited; descendants then levelled the two variant initials to one variant; meanwhile unrounding of wo- to wa- is regular in Gaulish).