below follows my treatment of paragraph 0 (the preface) of Amrae
Choluim Chille. I am grateful for any correction and addition.
Preface (invocation of God’s protection):
The preface is not written in a syllabic metre. Instead stress seems
to be the main feature. The invocation consists of 4 stanzas (all
starting with día). The stress pattern of a stanza is (with x marking
full stress, the number of unstressed syllables being irrelevant for
line 1: xxx
line 2: xx
line 3: xx
Stanzas 1 and 2 and stanzas 3 and 4 are linked by end-rhyme. The
last word of every stanzas is linked by alliteration to the first or
second word of the following stanza. Within every stanza there is
some alliteration also, but not very consistent (line 6, 9, 10, 12). I
am not sure if line 1 día, día would qualify as alliteration.
nominatiuus pendens, used here to focus the sentence on the
main addressee of the invocation. The ME Irish commentaries note
that the double invocation of God is based on biblical models:
“Domine, Domine, uirtus salutis meae” and “Deus, Deus meus,
respice in me, quare me dereliquisti?” (I guess these examples
must be from the psalter.) Unlike these biblical examples however,
where Domine and Deus meus are vocatives, in the ACC it is
1sg deut augm s-subj of do-guid (S2) “entreats earnestly, asks
pardon”. The position of the augment is preverbal (cf. EIV 134 ff.). I
am not certain about the function of the augment here, but I think it
has potential force, turning a 1sg hortative into an optative “may I
entreat” (cf EIV 109).
The whole being a verb in a nasalising relative clause, nasalisation
being shown by the geminate rr. A nasalising relative clause
because of the object antecedent día.
Reconstruction: CC *to-(ro)-g°es-s-u: < PIE (transponate) *g°hedh-
s-o-h2. The u in -rrogus is not a reflex of the OIr root vowel u in
guidid (which is a very late development), but it is a product of
“infection” of the apocopated Celtic 1sg ending onto unstressed e
(cf OIr tomus < *tomessu- < *to-med-tu-).
The preposition ré^N “before, in front” is here obviously used
instead of the usual temporal conjunction resíu “before”. Neither in
DIL nor in GOI could I find another example for that. ré follows resíu
in that it requires the subjunctive, but unlike resíu not an
independent but a dependent verbal form follows.
resíu really is ré^N plus the petrified dative of the particle se, -se, -
sa “this, that”. It seems as though we had a very archaic use of the
preposition ré^N as temporal conjunction in this case.
Speculation: Alternatively one could speculate that we have the
word ré (io, n) “space of time, period” before us, possibly in the
acc.sg., used as a conjunction. Cp. the conjunction in ta(i)n
“when”, which also is really an acc.sg. of tan “time”. But note that in
tan requires an independent verbal form as well. We might expect
the same for a conjunctional use of ré “space of time”.
1sg. conj. s-subj. of (S1a ii) téit “goes”.
Reconstruction: CC *te:x-s-u: < *tejg’-s-o-h2
preposition i^N “in, into” plus poss.pron. a^L “his”
acc.sg. of gnúis (i, f) “face, countenance”
cp. cymr. gnis “jaw, chin, face”, to PIE g’enh2u- “jawbone, chin” >
metathetised g’nuh2- (Lionel S. Joseph, Problems in the
Development of the IE Laryngeals in Celtic, 91 f.) in CC *gnu:-s(-)ti-
CULU TRE NÉIT:
Well, what shall we do with this unclear passage?
Before we can try to make sense of the whole phrase, we have to
look at the individual words.
The ME and the modern commentators agree in taking culu as a
form of the not very well attested word cul “chariot”
The standard etymology for cul is that it continues the PIE dual
*k°oloh1 of *k°olo- “wheel”. This has some difficulties: the dual of o-
stems in Celtic seems to have been reshaped to -o# instead of -
oh1# (under influence of the numeral duwo “two”). CC *k°olu:
would have been the only word to have retained the old ending -u: <
-oh1. I am always opposed to invoking “archaisms” as explanations
for forms that might also be explained in a manner more
conforming to standard morphology and sound changes.
First of all there is no inherent reason to assume why a word “two
wheels, chariot” should have escaped a general reshaping of the
dual ending -oh1 > -o (or rather -u: > -o). Second, if it had escaped
the reshaping a word CC *k°olu: would have resulted that would
have stood outside all inflectional classes of Celtic - a secondary
reshaping should be expected to integrate it into one of the
I rather think that OIr cul continues a u-stem *k°olu- (this might of
course be a secondarily reshaped k°olu:!). LEIA remarks that
culmaire “chariot builder” might represent an older *culbaire, which
could be a *k°olw-arijo-, a derivative with -aire from a u-stem base
*k°olu-. But we wouldn’t expect a u in the first syllable then, but
rather an o.
(BTW: the root vowel of the ancestor word of cul need not be o. If
we had an e-grade form *k°elu(:)- here we would arrive at the same
result in Old Irish: CC *k°elu- > *k°ilu- > *k°ulu- > k°ul > kul)
However that may be, culu looks like an acc.pl. of cul, if anything.
The ME commentator didn’t understand it and explained it as an
instance of a poetical device of obscuration called dechned “two-
ending”, where the poet puts a (meaningless) ending to an already
meaningful form. I really assume that dechned is simply a practical
way for ME commentators to explain away old forms they no longer
understand. In this case the commentator wanted to read a
nom.sg. and had to get rid of the -u somehow.
From a purely formal point of view culu might also be a 1sg pres.
ind. of a verb *ceilid, going back to *k°el(H)eti. The only root in
question would be PIE k°elh1 “to turn, to turn to(wards)”, but there
is no trace of a thematic present k°elh1eti in anywhere else in
The preposition tre, tri^L “through” takes the acc.
A masculine noun meaning “battle, fight”, according to DIL and
LEIA. Judging by the stem final consonant the word was an i-stem.
Peculiarly though, the gen.sg. of the word seems to have been néit
as well; not *néto as one would expect. GOI and CCCG mention
this endingless gen.sg. only for loan words from Latin, but give no
example for genuine Celtic words. McCone in Stair na Gaeilge
doesn’t mention it either.
Etymology: according to LEIA from *nanti-, related to goth. ana-
nanþjan “to dare”.
Now let’s turn towards the whole phrase:
The ME commentator understands the phrase as “(I will go before
God’s countenance) like a (scythed) chariot through a battle
(against demons)” (the ACC obviously offering the weaponry
against the demonic attacks). This sounds like a good piece of
overinterpretation. This interpretation is not plausible since he has
to assume an “artificial disguise” -u having been stuck unto
CM translate the phrase as “through the chariots of battle”. To do
so they have to take tre as a postponed preposition to culu. There
are indeed other examples for this highly artificial word order in the
ACC. néit has then to be taken as gen.sg. of an i-stem. This is
indeed attested for this word otherwise. From a formal point of view
CM’s translation is acceptable, although the meaning is not so
clear: why should one come before God’s face “through chariots of
battle”? Is there a parking lot for chariots in front of heaven’s gate
which one has to climb over before entering? Does anybody have a
better interpretation? I have none.
gen.sg. of nem (s, n) “heaven”. The raised i of the stressed syllable
shows that a form *nemisos has to be reconstructed as the
IMO in Celtic every unstressed e before s and another vowel was
raised towards i, or rather: e and i fell together in that position in
something very close to i. So the expected gen.sg. *nemesos
became *nemisos. There is a lot of evidence for that from
Continental Celtic, like for example aiuizas (in Botorrita I), IMO from
*h2ejwes-eh2-, to the root h2eju- “life force”, or the very freuqent
Gaulish suffix -isio-, which seems to have developed originally as
an -io-derivative from s-stems, but was then generalised as a
ním = negative particle ní plus 1sg infixed pronoun class A.
reilge = 3sg. conj. augm. a-subj. of W2a léicid “lets, leaves” < *-ro-
léice < *-ro-link°ija:t. The ending -e shows that a sequence -ija:t/s
developed into -e in Old Irish. (This is important for the feminine a:-,
ia:- and i:-stems where it is unclear whether gen.sg -e went back to
-ija:s with long a or -ijas with short a.) Later the 3sg. a-subj.-ending
of W2 verbs was reshaped to -(e)a in analogy to the W1 verbs, so -
reilgea would be a later form.
The augment again has here - as above - potential force, turning a
jussive “let him not leave me” into a somewhat milder wish.
From *-ro-léice we would really expect *-roilge, but the
augmentised variant always seems to be -reilic etc. GOI 529
suggests influence by reilic “graveyard”, but this seems a bit
farfetched. Perhaps it is just vowel assimilation like in do-gní and
do-sluindi, where we also get assimilated augmented forms like do-
rigéni and do-ríltiset; there is also a similar variation in deut. do-léici
versus prot. -téilci.
acc.pl. of lorg (o, m) “trace, path, pattern, model, rear, band”.
Léicid “leaves” can either mean to “leave someone in some place”
( + dat.) or to “allow someone to move somewhere” (+ acc.). In this
case we have the latter meaning.Therefore I think that CM’s
translation “let him not leave me (=allow me to go) unto paths” is
Again, as with culu, the ME commentator takes the final -u to be an
example of deliberate obscuration of lurg (dat.sg.), but this is
unnecessary. He understands the passage as meaning “let him not
leave me in the band (of demons)” (with the datival meaning of
léicid), but CM’s translation makes more sense from a
morphological point of view.
When used at the beginning of a relative clause the preposition i^N
never takes the relative particle -a^N, which all other prepositions
take. Instead it simply stays as it is.
3.sg. pres. pass. conj. of W2a éigid “cries out, screams”.
That CM print a trema on -égthiar (and on a few other passives in
the ACC) is a stupid mistake taken over from WS’s edition.
WS (in 1899) thought that the spelling -thiar indeed meant a
disyllabic ending -thi-ar and wrote so in his edition p. 34. He
compared it with Old Cymric forms in -iaur, -ior. He thought he had
to do with a very archaic passive ending, consisting of -thi-
(comparable with Old Greek -tai), unto which an unclear Celtic -ar
Today we know that this is nonsense: First of all disyllabic
pronounciations (=hiatuses) are not possible in unstressed
syllables in Old Irish. Second, Old Greek -tai goes back to PIE -toj,
one of the two possible IE sets for medial endings. The other one -
tor is the only one attested for Celtic (there are nice examples in
Celtiberian). The Celtic passives and deponents all go back to this -
(t)or ending in the 3sg. This is of course also true for -égthiar here.
In standard orthography we would expect the form to be -égther.
For the spelling -ia- two explanations are possible: perhaps at the
time of the composition of ACC (traditionally said to be about 600)
the rules for the spelling of schwa in unstressed syllables were not
yet as fixed as a hundred years later and a schwa between a
palatalised and a non-palatalised consonant could be written -ia-.
Alternatively one could say that the vowel of the final syllable had
not yet developed to schwa, but was still /a/. For the development
of o > a in final syllables, at least before s and n, cp. McCone, Stair
na Gaeilge 78. I don’t no about any counter evidence that o did not
develop to a before final r as well. If the final vowel was still /a/, but
followed a palatalised consonant the most natural thing to do would
be to write an i before the a.
For curiosity’s sake: the ME commentator explains inégthiar to be
“ainm demain duib do-beir múich for cach muintir” “the name of a
black demon who brings gloom on every family”.
CM take this word to be gen.sg. of múchad “gloom, heaviness,
oppression”. It is used here as a preposed genitive of méit, the
usual word order would be ar méit múichtheo.
In the MSS it is always written muichthea or muichdia (only in YBL
with fada). The -o that CM restitute is an attempt at archaisation
which is not really necesary.
Múchad is the verbal noun to W1 múchaid “presses down,
suffocates, obliterates etc.”, a denominative verb to múch “smoke”.
Verbal nouns of denominative verbs are usually masculine u-
stems, the reconstructed stem being *-a:tus. The gen.sg. of such a
stem would be *múchtho. The apparent palatalisation of
mu/úichtheo is problematical. It points to a reconstructed stem
*mu(:)ketus, which should give a nom.sg. *mu/úchud in Old Irish
(but it is múchad, cp. DIL s.v. múchaid).
Maybe mu/úichthea is the gen.sg. of an otherwise unattested word
*muchud, a derivative of muich (f) “gloom, dejection, sadness”
(also spelled múich and muích), which might be a word
independent from múch “smoke”.
dat.sg. of méit (i:, f) “magnitude”. It is taken to be i:-stem because
of cymr. maint “quantity”, corn. myns, bret. ment < *manti:-. In the
Milan glosses the nom.sg. is mét. I don’t know if this is a different
formation (a:-stem), or a lautgesetzlich development of *-nti:.
The reconstruction is probably *mh1-nt-ih2- > CC ma(:)nti:- >
manti:-, which belongs to the root meh1 “to measure” (Schrijver,
Studies in British Celtic Historical Phonology, Amsterdam 1995, p.
This word belongs to a group of i:-stem words which have, unlike
e.g. rígain, no i-ending in the dat. and acc.sg. (rígnai) , but rather a
short ending. Since however the historical explanation of the
inflection of Celtic a:-, ia:- and i:-stems belongs to one of the most
complicated chapters of Celtic historical linguistics I won’t go into
This is the older form of the adjective mór (o, a:) “big, great”.
Reconstruction: moh3-ro-, cf. cymr. mawr, bret. meur. Irish mór
has its ó through analogy from the comparative mou and the
nom.sg. of anacul (o, n) “protection”, verbal noun of aingid “saves,
protects”. The Celtic root is aneg-, attested also in Gaulish
Anextlomarus “big in protection, giving great protection”.
The value of the c in anacul is a bit difficult to determine. From the
evidence of modern Irish anacal the c stands for /k/, but there are
also spellings like anaghal. This -gh- is most probably in analogy to
the lenited g of the verbal forms.
GOI explains anacul as coming directly from anextlo-, with CC -xtl-
giving -kl- in Irish. This seems a bit ad hoc to me, since there
doesn’t seem to be another example for that development; and I
don’t see a reason why anextlo- shouldn’t give *anachtul in Old
Irish, as one would expect.
DÍA MÁR MO ANACOL:
CM translate this phrase as “Great God, protect me”, but this
certainly isn’t literal. In Old Irish there is no vocative involved. The
most straightforward translation would be “The great God is my
protection”. But given the fact that there are no commas in the MSS
and that word order is sometimes very free in ACC we could also
translate the phrase in one of the following ways: “God, he is great
as my protection (=protector)”, “God is my great protection”.
The preposition di^L “from” is one the prepositions taken by aingid
“protects”. The others are ó^L and ar^L.
dat.sg. of múr (o, m) “wall, rampart”. múr sometimes seems to
mean “abundance”, but LEIA explains this as a development of múr
“rampart”. Anyway, “wall of fire” or “abundance of fire” both make
sense in this case.
This is a loan from latin murus “wall”.
dat.sg.masc. of teintide (o, a:) “fiery”. One would really expect
lenition theintide here after dat.sg. múr, but this isn’t always
This is a derivative adjective from tein or teine (d, m) “fire” with the
secondary adjectival suffix *-ode < CC *-odijo- (further
reconstruction unclear). Reconstruction: CC *tenetodijo-.
OIr. tein, teine, brit. tan “fire” is usually compared with avestan
tafno: “heat” < tepnes-. If this is correct it would show that in a
sequence PIE *-epn- p disappeared completely in Celtic, whereas
in a sequence *-opn- it developed to Celtic *-own-.
By the ME Irish commentators this is taken to be a hybrid
compound of Latin diu “a long time” and derc (a:, f) “eye” = “a long
LEIA (s.v. diuderc; following Binchy Ériu 26) however says that this
is a vernacular formation di-us-derk- “long observation, long look”
which was later reinterpreted as consisting of Latin diu and Irish
derc. DIL adopts this view as well.
CM however take derc in the meaning “pit, trench, hollow, hole” and
understand the whole passage as “God, the protection from the
deep pit of tears”. They think that here a contemporary vision of hell
as a deep pit of tortures is envisaged.
If diuderc really is from *di-uss-derk- this would IMO - because of
the two preverbs - speak for the interpretation of derc as verbal
noun “looking” and of the whole phrase as “a long look of tears”.
gen.pl. of dér (o, n) “tear”. Later the word becomes fem.
The Irish word seems to go back to an o-stem *dakro-, whereas
most IE languages (including cymr. sg. deigr < dakru:, pl. dagreu <
dakrowes) point to a IE u-stem.
This adjective has three variants: fírián, fír(i)én, fírión (o, a“just,
righteous”. It is a loan from British, cp cymr. gwirion “just”, itself a
compound of gwir “true” and iawn “sound”.
A compound of fír (o, a:) “true” and ocus (u) “near”. The spelling of
ocus with unorganic (=historically unjustified) f- may be due here to
attempted graphic alliteration. In Middle and Modern Irish the word
really develops an f-: cp Mod Ir fogas “near”.
The conjunction ocus^L “and” (with palatalisation ocuis in the
Cambrai Homily) is perhaps a case form of ocus “near”. But
certainly both are derivatives of oc “at, close by”. This is most
probably cognate with cymr. wng “close, near” < *onk-. The
relationship of ocus to cymr. agos, corn. ogas, bret. hogos “near,
close” is very difficult.
Reconstruction: In view of palatalised forms like comparative oicsiu
we would perhaps reconstruct ocus as CC *onkestu- (ultimately
belonging to the root h2nek’ “reach”?).
3sg. pres. abs. rel. of S3 ro-cluinethar. Neither in GOI, nor in DIL,
nor in EIV, nor in LEIA is this strange active 3sg. relative form of a
deponent verb mentioned. We would really expect ro-chluinethar
The lenition which CM restitute for chluines is unnecessary.
According to GOI 315 the special relative forms of simple verbs are
not lenited in leniting relative clauses in the oldest period.
acc.sg. of donúall, donál (a:, f) “noise, cry, wailing, yelping”.
This is really a compound of the pejorative prefix do-, du- and núall
(o, n or m) “noise, shout”. That a neuter or masc. o-stem word
should become a feminine a:-stem in composition (as the entry in
DIL suggests) seems really incredible. What I would rather think is
that the compound do-núall- changed to i-stem inflection, which is
very frequent (cp do + cenél > dochenéuil, do + adbar > daidbir
etc), and what we have before us here is a substantivised masc. i-
The ME commentator (and with him WS) took mo dó nuaill to
mean “my two cries”, which is of course wrong.
Both the ME commentator and CM understand do^L as de^L “from”
in this case. do for de is of course a very frequent substitution in
Irish, but I had thought that both had not fallen together at an early
date like that and that they are well separated in ACC. Therefore I
would rather take it to mean “to, towards”, the whole phrase: “who
hears my wailing to the cloudy heaven”.
A compound of nem (s, n) “heaven” and íath (u, n) “land, country”
(< PIE pejtu-). Why the composition form of nem is nim- in this
case I do not know. Otherwise it always seems to be nem-.
gen.pl. of nél (o, m) “cloud” (probably < PIE *nebhlo-).