First published in: Britain 400-600: Language and History, ed.

by Alfred
Bammesberger & Alfred Wollmann. Heidelberg: Winter 1990 (=Proceedings of
Eichstätt 1988), 263-281 (poor typography, many remaining typos). There was
one common list of abbreviations; the short forms and sigla used here are the
usual ones in Celtic Studies. Addenda and corrigenda are inserted here in
square brackets [ ... ]. SWWF (in addenda) refers to autor, Studies in Welsh Word-
Formation, Dublin: DIAS 2000.

Dating the loanwords: Latin suffixes in Welsh (and

their Celtic congeners)

Stefan Zimmer
1 Introduction
Approximately 350 years of Roman rule in Britain have left clear traces in the
British languages. The details and range of Latin influence on British are,
however, not yet fully clear, and the spread of spoken Latin among the local
population is an unsettled question. The sociocultural process of Romanization
must have at least begun in Britain as in every other province of the Roman
Empire. It concentrated in garrisons, cities and towns, and was carried forward
mostly by commerce and the different military institutions. We know almost
nothing about Britons in the Roman army. The rural population was probably
clinging to their ancestral language and way of life.
One has often asked why Welsh did not become a Romance language.
Rumanian did, after the much shorter Roman occupation of Dacia. The solution
to the problem may be found in geographical and sociological facts. Within the
territory of Wales, there was only one Roman town, and a small one: Venta
Silurum, Caerwent.1 After the withdrawal of the Roman legions, Roman-style
institutions were still alive for a certain time (cf. Gildas’ writings; Latin names of
local rulers;2 Latin funerary inscriptions3). The cities and towns of Britain, centres
of Romano-British civilization and of British resistance against the invading
Anglo-Saxons, were heavily struck by the Yellow Plague of 544, which might well
be regarded as the final stroke for Roman Britain. [264]

[Please consider in reading this article that GPC was only published up to the beginning of M-
early in 1988. Cf. fn. 39.]
1
Three cities are known from Roman Britain: Londinium, Eburacum (York), and Lindum
(Lincoln): their bishops are attested as participants of the Gaulish regional council at Arles in
314.
2
E. g. Agricola in Dyfed, Constantinus and Aurelius Caninius in the Southwest.
3
The latest is the Cadfan-inscription in Llangadwaladr, Môn, from c. 625.
2 The Latin loanwords

All the more precious for us are what the Romans have left to the British. In
the first place, this is the host of Latin loanwords in Welsh. They have been
collected by J. Loth,4 L. Mühlhausen5 and H. Lewis6 (cf. also K. Jackson’s LHEB),7
but it has not yet been possible to use them fully as historical sources because of
chronological problems. It has been extremely difficult to date these loans for
several reasons, among others the following:
1. Words taken over from spoken Latin underwent the usual Late British/Early
Welsh sound changes, but could have been ‘corrected’ at any time by checking
against the Latin literary form, e. g. Lladin instead of regular *Lledin < (lingua)
Latina;
2. So-called learned loans down to the present day can be formed as if taken
over in the Late British period, e. g. rheithor (Latin ending -or restored for regular
-ur < -orem) < E = L rector, as if from British Latin rector(em); frith-iant, 20th c.,
translating E friction, as if < L frictio(nem) + W suffix; ffwyth-iant ‘function’, 20th c.,
as if < L functio(nem) + W suffix;
3. The paucity of the corpus of extant Old Welsh texts and glosses;
4. The date of some sound laws is disputed (see the contribution of John Koch
in this volume), so that e. g. the presence or absence of lenition cannot be used
as a criterium for dating.

Moreover, Latin did not disappear from British soil with the last Roman
soldier. It is a matter of dispute whether Latin was still spoken as an everyday
language by Romano-British urban societies in the 5th and 6th centuries (see e.
g. the contributions by Michael Herren and Alfred Wollmann in this volume).
Anyhow, as the language of the Church, of arts and literature, and to a certain
extent of civil administration, Latin continued to influence [265] the Welsh
language for several hundred years (by transmission of English, it still does).

4
Les mots latins dans les langues brittoniques (Paris, 1892).
5
Die lateinischen, romanischen und germanischen Lehnwörter des Cymrischen, besonders im
‘Codex Venedotianus’ der cymrischen Gesetze (Diss. Leipzig 1914), in: Festschrift Ernst Windisch
zum siebzigsten Geburtstag am 4. September 1914 dargebracht von Freunden und Schülern
(Leipzig, 1914), S. 249–348.
6
Yr Elfen Ladin yn yr Jaith Gymraeg (Caerdydd, 1943; abbreviated ELIG).
7
Language and History in Early Britain (Edinburgh, 1953).
3. Latin suffixes: means for dating
Some light on this special feature of Britain’s Dark Ages may be shed by
analyzing the grammatical remains of Latin in British which are proof of a much
deeper influence than that attested by isolated lexical items only. In what
follows, Welsh suffixes borrowed from Latin are used as a means of dating
(terminus non post quem: 410 AD) 8 of at least some words and word groups. The
argument is based on the assumption that after the disappearance of Latin as a
spoken language, only isolated words could be taken over from literary (mostly
ecclesiastical) sources. It is extremely unlikely that any such later loans were
numerous and semantically coherent enough to allow the abstraction of suffixes
which would have been used further for derivation with genuine W words. As the
British and Latin languages are genetically closely related to each other, the
suffixes in question must be classified into three categories:
I. loan suffixes without Celtic equivalents,
II. loan suffixes which were or became identical with corresponding Celtic
suffixes,
III. Celtic suffixes whose productivity was enhanced by Latin loanwords with
identical or very similar Latin suffixes.
Only the first group will be treated fully here. For the second and third
groups, some examples are given.

4. Latin suffixes borrowed into Welsh

4.1 -awd(w)r
The suffix, written -audir in OW, -awdr and -awdur in MW, has been taken over
from the L nomina agentis in -atorem. [Addendum: This is wrong: W -awd(w)r is
from the L nom. -tor, whereas the L acc = VL casus rectus -atore(m) gave W -adur,
cf. SWWF 282, 440.] It is used in the same function in W. The W plural is formed
with -awdwyr, as if the words were [266] W nominal compounds with second
member -wr < gwr ‘man’. The group started with the two old loans
creawdwr < creator ‘creator’, 12th/13th c. +, (not to be confused with
creawdr/creawdur ‘creature’, 16th c. +, a learned loan from creatura!) [Addendum:
cf. W creawdyr celi < L creator caeli in T 67.17, cited in GBCC] and
ymerawdwr/ymherodr/ymherawdr < imperator ‘emperor’.
The first W formations with -awdr are
dysg-awd(w)r ‘teacher, wise man’, c. 1300+, : dysg-u ‘to learn’, c. 1200+, ‘to teach’,
13th c. +

8
Or, if one accepts the idea of a Latin speaking or bilingual urban population of British origin
during the 5th and (perhaps still partly) 6th centuries, not much later.
iach-awd(w)r ‘saviour’, 14th c. +, : iachau ‘to heal, save’, c. 1250+ (denominal verb
of iach ‘sane, safe’, c. 1200+, ‘secure’, 13th c. +, ‘redeemed, delivered’, 14th c. +)
llyw-(i)awd(w)r ‘governor, ruler, pilot’, 12th/13th c. +, : llyw-(i)o ‘to govern, rule’,
12th/ 13th c. +.

4.2 -dod
The suffix, written -daut in OW, -dawt in MW, has been abstracted from the
following old loanwords ending in L -tātem:
awdur-dod < auctoritatem, mf. ‘authority’, 12th c. +; many derivatives
car-dod < caritatem, f. ‘alms, charity’, 12th/13th c. + (also written cer-dod in the
14th and 16th c.); many derivatives9
ceu-dod < cavitatem, mf. ‘thought, mind, heart, bosom’, 10 12th/13th c. + (=Bret.
coudet ‘mind, heart’); ‘cavity, hollow, belly’, 13th/14th c. +
ciw-dod < civ(i)tatem, mf. ‘people, inhabitants, community’, 13th c. + (= Bret,
queudet ‘city, town’)
gwyr(y)f-dod < virginitatem ‘virginity’, c. 1300+11 (= Bret. guerchdet ‘id.’)
trin-dod < trinitatem, f. ‘trinity’
ufyll-dod < humilitatem, m. ‘humility, modesty’
un-dod < unitatem, m. ‘unit, unity, union’.

There are only a few old derivatives from Celtic words. This is why I consider
the existence of a Celtic productive suffixe *-tātan (a parallel to, not a loan from L
–tātem) rather improbable. [267]
cymhen-dod ‘wisdom, knowledge, eloquence’, 14th c. + , 12 : cymen ‘wise, skilful,
proper’, 13th c. +; ‘eloquent’, 16th c. +
diben-dod ‘end, fulfilment’, 15th c. + : 1di-ben ‘end, conclusion’, 13th c. +, ‘aim,
intent’, 16th c. + (with intensifying di-)13 — the antonyms are later: anniben-dod,
16th c. + : anniben, 14th c. +, both with several different meanings because of the
ambiguity of di-.

9
Especially productive are the factitive formations like cardot-a vn ‘to beg’ < ‘looking for alms’,
cardot-ai ‘beggar’.
10
This is clearly the result of a special semantic development in British Latin (for an
interpretation, see the final remarks of this paper). In no Romance language, a reflex of
cavitatem has a similar meaning.
11
But GPC has doubts with regard to gwyryf/gwyrf/gwyryf < virgo.
12
Already once c. 1200 with unclear meaning ‘security, assurance’ (?); see GPC.
13
In N. W. another diben-dod ‘untidyness’ : 2di-ben a. (BV) ‘headless, endless’, 17th and 18th c.,
‘slow witted, stupid’ (modern) (with negative di-), is found.
gweddw-dod ‘widowhood, unmarried state’, 13th c. +; ‘bereavement’, 14th–17th
c., : gweddw ‘widowed, widow’, 12th c. +; ‘unmarried’, c. 1200+; ‘bereft, orphaned’,
c. 1300+.14

From the 16th c. on, the suffix -dod became productive in forming abstract
nouns from substantives (e. g. bachgen-dod ‘childhood’), adjectives (e. g. byddar-
dod ‘deafness’) and verbal stems (e. g. cryn-dod ‘trembling, fear’). Together with -
edd and -rwydd, -dod is today one of the most productive abstract suffixes in W.

4.3 -ell
The suffix goes back to L -ellus, -a, -um. There are quite a number of
loanwords. The older W formations are less numerous. The suffix acquired a
certain productivity in modern words for tools. Nearly all are feminine nouns and
show several plural formations. The oldest of them is the i- Umlaut alone, going
back to the L masculine pl. -elli.
asgell (pl. esgyll) < ascella ‘wing, feather, spear’, 12th/13th c. + (= Corn. ascall, Bret.
askell)15
astell (pl. estyll, modern astell-oedd/-au) < astilla ‘lath, plank, boardʼ, 14th c. + (=
MBret. astell)
cafell (pl. -au/-oedd only) < cavella16 ‘sanctuary, temple, cell’, 14th c. +
1
cangell (pl. canghell-au/-oedd) < cancellus ʻchancel’, c. 1200+; ‘sanctuary’, 17th c. (=
Ir caingel ‘chancel’, Bret. cancell ‘locus’)17 [268]
castell mf. (pl. cestyll, castell-i/-au/-oedd) < castellum ‘castle, stronghold’, 12th c. +
(= MBr. castell, Ir caisel); a corn and land measure, 14th c.
cawell m.! (pl. cewyll, cawell-au/-i) < cavella ‘basket’, c. 1200+ (since the 18th c. also
‘cradle of a scythe’, S. W.) (= OBret. cauell gl. cofinus; Corn. cawel, cawal)
cyllell (pl. cyllyll, cyll-aill/-eill)18 < cultellus ‘knife’, 10th c. + (OW celell gl. culter); the L
plural cultelli provided the model for the usual plural formation of this group, but
the L genus has not been retained by W (= OCorn. collel gl. cultellus, MCorn.
collan, Bret. kountell [dissimilated]).

14
Another old word could be pererin-dod ‘pilgrimage’ (: pererin < L peregrinus).
15
Cf. the recent formation blaen-asgell ‘wing forward (player)’.
16
For semantic reasons, the interpretation of GPC (as a variant of cawell, see below) seems
unlikely to me. I should rather presume a double meaning of British Latin cavella. The
difference of W -f-(-v-) : -w- (OW kauell) could have been introduced as a means of
differentiation.
17
Cf. MHG kanzel ‘chancel’ and ‘pulpit’.
18
On the plural cyllellawr, 13th c. once, perhaps a collective borrowed from L *cultellorium, see
below.
cymell m., a., vn (pl. cymhell-ion) ‘incitement, urge, compulsion’, 13th c. +; as a.
‘constrained, forced’, 16th c. +, backformation from cymhell-af ‘to incite, etc.’, c.
1200+, < compello19
fflangell (pl. -au/fflengyll) < flagellum ‘whip, scourge; whipping, scourging’, 1588
OT+20
ffynhonnell (pl. ffynonnell-au) < fontanella ‘source, fount’, 1567 NT+, perhaps a late
learned loan; but cf. ffynnon ‘source’, 9th c. + < fontana
gefell mf. (pl. gefe(i)ll-iaid/-ion, gefyll; f. gefe(i)ll-es) < gemellus ‘twin’, 14th c. (= Bret.
gevel)
gradell mf. (pl. -au, gredyll) < gratella (for classical cratella) ‘griddle, gridiron,
bakestone, pan’, 10th c. + (OW gratell gl. graticula)
1
isgell m. (pl. -au) < iuscellum ‘broth, soup, decoction’, 14th c. +
llogell (also llawgell)21 mf. (pl. -au/-i) < locellus ‘small box’, ‘small room, cell, box,
bag’, 9th c. + (OW locell … gl. ferculum) (= OBret. logel gl. loculus, Bret. logel ‘hut’)
mantell (pl. mentyll, mantell-au/-oedd/-i/-ydd) < mantellum22 ‘mantle’, 13th c. + (=
OCorn. mantel gl. mantellum)
pabell (pl. pebyll) ‘tent, pavilion’, < papilio or its pl. papilies ‘tent for military and
civil purposes’ (Thesaurus Linguae Latinae vol. X/l, fasc. ii, 253 f.)
padell (pl. -i/-au, pedyll) < patella ‘pan, bowl’
pibell (pl. -au/-i) < *pipellum23 ‘pipe, duct’
porchell m (pl. perchyll) < porcellus ‘young pig’ [269]
pubell < pupilla ‘eye, look’ (ELIG 46) or ‘pupil of the eye, eye’ (GM)
tafell (pl. -au/-i, tefyll) < tabella ‘slice, slab’
ysgafell (pl. -au) < scabellum24 ‘ledge’
ystafell (pl. -oedd) < stabellum ‘room, chamber'.

From these loanwords, the suffix -ell was abstracted very early to form
derivatives from inherited Celtic words: diminutives, words for tools (similar to
the L models), and other words. Gender and plural formation of these
derivatives usually agree with that of their bases.

Examples for diminutives:

19
Or directly < *compellum? — Cf. further argymell vn ‘to constrain, force’, 14th c. once; ‘to incite,
urge’, 19th c.
20
Cf. further the vn fflangell-u and other derivatives. GPC takes it as a late learned loan because
of the date of the first attestation. But cf. OBret. flagell gl. flagrum.
21
Because of popular etymological connection with llaw ‘hand’.
22
Or from ME = OFr mantel?
23
?, cf. pib ‘id.’ < pipa.
24
H. Lewis, ELIG s. v., derives it from L scamellum. But Fr. escabaut, escabelle (cf. HG [dialect of
Frankfurt am Main] Schavellche) points clearly to L -b-.
both-ell mf., -au/bothyll (also pothell and ffothell) ‘blister’, 14th–18th c.; a cattle
disease, 18th c. + (supposedly < ‘round hillock’) : both f., -au ‘nave (of a wheel),
boss (of a shield), round pot’, 13th c. +25 (cf. Bret. bouzell-en ‘innards’)
crib-ell mf., -au ‘small comb, crest (of a bird), ridge, summit’ (also name of plants),
17th c. +,26 : crib mf., -au ‘comb’, 8th c. + (OW ha crip gl. pectens; MBr. cribell)
gron-ell mf., -au (also cronell)27 ‘fish-spawn, hard roe of fish’, 17th c. + (lit. ‘small
grain’) : grawn c ‘grain’, 12th/13th c. +
hun-ell f., -au ‘nap, wink; sleep of death, grave’, 16th c. + : hun f., -au ‘sleep’,
12th/13th c. +
lin-ell f., -au/-i ‘line, verse, etc.’, 16th c. + : llin < L linea, 12th/13th c. +28

Examples for names of tools:

craf-ell f., -au/-i ‘scraper, grater’ (of several different tools), c. 1200+, [270] : craf-u
‘to scratch’, 13th c. + (= Bret. kravell ‘hoe for weeding’)
chwistr-ell f., -au/-i ‘syringe, water pistolʼ, 16th c. + (onomatopoetic formation?)
gwä-ell mf., pl. gwëill/gweillion ‘knitting-needle, awl, pin, broach, spindle’, c.
1200+; unexplained formation according to GPC, but for semantical reasons not
to be separated from gwa-u, gwe-af ‘to knit, sew’, 14th c. +, gwe mf. ‘tissue’, c.
1200+
gwynt-ell (also gwintell) f., -i/-au, gwyntyll ‘round basket without a handle, of
woven hazel withes or osiers, for potatoes etc.’, 18th c. +; perhaps < OE windel
‘basket’.29

4.4 -or

This suffix is a loan from L -arius, -aria, -arium and -are. It is used to derive
agent nouns, collectives and local nouns from substantives. It is not productive
in modern W, but there are new borrowings from E of the type baglor ‘bachelor’

25
This could be an inherited Celtic word (cf. Gallo-Latin bottos) or a loanword < L butta.
26
In Ceredigion also cripell ‘little ridge, rocky hillock’.
27
This form may be influenced by crwn ‘round’, as GPC proposes, but could also be due to a
general tendency in W not to distinguish between cr- and gr- (cf. already gradell < L cratella).
The phonetic details and possible historical background cannot be discussed here.
28
Cf. further am-linell f., -au ‘outline, sketch’, 18th c. + with derivatives. — hocrell f., -au ‘young
girl’, 14th c. +, may also belong to this group. According to GPC, it is a loan from E hoggerel
‘young sheep’ (cf. the latinized hogerellus, 13th c.). It could however be a case of double
borrowing from W to E and from E back to W (cf. M. Förster, Keltisches Wortgut im Englischen,
Texte und Forschungen zur Englischen Kulturgeschichte, Festgabe für Felix Liebermann … (Halle,
1921), p. 119–242, esp. 132 f.) like hogyn, hogan. Förster did not mention hocrell and hoggerel.
29
Suggestion by GPC; in fact, it cannot be excluded that the word, in spite of its very late literary
attestation, is much older. Potato baskets are rarely mentioned in literature.
(university grade), 18th c., < bachelor (as if < L baccala(u)reus), bordor ‘frontier’,
15th c. +, < border, sgolor < scholar, and many more recent E words (partly
adapted orthographically), e. g. actor, ffactor, sensor, tiwtor. All such words are
not taken into account here.

I regard the following as loanwords from living British Latin:

allor f. < altare ‘altar’, c. 1200+
canghellor m. < cancellarius ‘chancellor’, 12th c. +
callor mf. < caldarius ‘cauldron, kettle’, 9th c. +
catgor m. < British Latin *kattvóres (vel sim.)30 (tempora) 'ember, fast’, 14th c. +
cyllell-awr < *cultellarium/-a ‘a collection of knives, a set of knives’ (in the kitchen)
(?), 13th c. once as pl. of cyllell ‘knife’ < cultellus31
egwyddor mf. < OW agwyddor < *afgwyddor < abecedarium 'principles, alphabet’,
10th c. + (since the 15th c. also without e-) (cf. OIr abbgitir)
hestor mf. < sextarius a measure of quantity, 9th c. + (cf. the OW gloss in
sextario .i. hi hestaur)
Ionawr m. < *Ianarius < Ianuarius ‘January’, 13th c. + [271]
tymor m. < *tempóris < tempor-is32 ‘season’
ysgubor f. < scoparium ‘barn’ (OW scipaur),33 cf. ysgub ‘sheaf’, < scopa.

Welsh formations with the suffix abstracted from these loanwords are the
following:
blaen-or m. ‘leader’, etc., 14th c. + : blaen m. ‘end, front’, 12th c. +
carchar-or m. ‘prisoner’, c. 1200+ : carchar ‘prison’, c. 1200 + , < carcer(em)
cerdd-or m. ‘musician’, c. 1200+; ‘craftsman’, 13th 14th c. : cerdd f. ʻsong, music,
poetry’, c. 1200+; ‘craft’, 13th c. +
corn-or m. ‘bugler, leader, queen-bee’, 14th c. + : corn ‘horn’, c. 1200+
cryth-or m. ‘crwth-player, fiddler’, 13th c. + (= OIr cruittire) : crwth m. ‘a kind of
fiddle’, c. 1200+
drys-or m. ‘porter, doorkeeper’, c. 1200+ : drws m. ‘door’, 9th/11th c. +
ffo-awr m. ‘fugitive’, 13th c. once : ffo m. ‘flight’, 12th/13th c. + < fuga
(g)el-or f. ‘bier’, 12th/13th c. +, lit. ‘lier, something to lie (someone?) on’ (= Corn.
geler, elor, MBr gueler) : *gol- ‘to lie’, cf. gorwedd

30
ELIG: < quattuor (tempora): less probable because of the retained ending, which could,
however, have been restored.
31
On MW melyinawr see I. Williams, CT, ad viii 21.
32
According to ELIG 20 (§74). I propose to start from the nominative/accusative plural tempora
(with an intermediate stage * tempora), because generalization of a genitive is unlikely.
33
Mentioned ELIG 48.
heus-or m. ‘herdsman’, 13th c. +, : *haus ‘sheep’, cf. heus-lau Corn. ‘sheep lice’,
16th c. +34 (= OBret. ousor gl. opilio)
llen-or m. ‘writer, literary man, scholar’, 13th c. + : llên f. ‘literature’, 13th c. +, <
legendum (also c ‘learned/literate people’, 14th/15th c. +)
maen-or f. name for an administrative unit (S. W.), 35 9th–17th c. (OW mainaur),
perhaps : maen ‘stone’ as ‘unit defined by boundary stones’36
MW magl-awr according to GPC a. ‘ensnaring, fig’, 13th c. thrice; perhaps
originally a noun ‘snarer’, : magl ‘snare, trap’, 13th c. + < macula ‘mesh’
perigl-or m. ‘mass priest’,37 : perigl ‘mass’,38 at least since 14th c. (Dafydd ap
Gwilym)
porth-or m. ‘porter’, at least since 13th c. (Mabinogi)
telyn-or m. ‘harpist’, f. -ores, : telyn f. ‘harp’. [272]

4.5 -us
The adjective-forming suffix -us is a loan from L -osus and -usus. It is still quite
productive today, but considerably less so than -og and -ol. There are only very
few old loanwords:

astrus < abstrusus ‘difficult, ambiguous’, 13th–17th c.

dolurus < dolorosus ‘sore, painful’, 13th c. +, 14th c. three times, also doluryus,
probably < doloriosus; cf. the basic word dolur m. ‘pain, hurt’, 12th/13th c. + <
dolorem
llafurus < laboriosus ‘laborious, painstaking’, 13th c. +, 14th c. five times written
llafuryus (-y- perhaps a trace of -i-); cf. the basic word llafur m. ‘labour, hard work’,
12th/13th c. + < laborem.

The contrast between the paucity of old loanwords with -us and the
numerous later formations is best explained by the fact that it was just the
double loan of basic word and derivative in both cases of dolur/dolurus and
llafur/llafurus which showed clearly enough the suffix and its function for its
adoption into the system of W word formation devices.

34
GPC considers an etymology for heusor as stemming from CC *oistāros, obviously meant as
*oi-stā-ro- ‘who is standing with the sheep’, but this seems unnecessary. [Addendum: No,
GPC is right. Cornish heuslau must be analysed as heu-slau. ???]
35
Spelled maenol in the law books of N. W.
36
The later meaning ‘villa, manor’ is due to the influence of Ε manor and its Fr. foreforms.
37
But according to the W paraphrase in GM ‘priest hearing confession’!
38
The older (regular) form of perygl ‘danger’ < periculum is also perigl.
 5 Suffixes of double origin

A number of suffixes are probably inherited from CC and borrowed from L: as

mentioned above, this is partly due to the close genetic relationship of L and
Celtic. Though the languages were no longer mutually intelligible in Caesar’s
times, many words and their suffixes were still very similar to each other.

5.1 -ol
An instructive example of this type is the adjective-forming suffix -ol, today
the most productive of its kind. On the one hand, it is found in a number of old
loans from L words with several different suffixes, and on the other hand in
genuine W formations for which Morris Jones (WG 256) postulated a CC suffix
*ālo-. Unfortunately, his oldest (at least as far as can be seen at present) 39
example, estronol (MW estron-awl, 13th–15th c. five times), is a derivation from
the loanword estron ‘foreign(er)’ < extraneus, [273] and is therefore not strong
enough to establish the reconstruction.40

In two words, -ol is from L -alis (-alem):

canol m., -au < canalis ‘pipe, duct’, MW canawl (1) ‘riverbed, canal, stream’, c.
1300+ (oldest attestation: ar hyd canavl Temys translating alveo Tamensis); (2)
‘middle, centre’, 13th/14th c. +; (3) ‘groove of a blade’, 14th c. three attestations;
(4) a. ‘middle, mediating, medium’, 16th c. + (first in a gloss by W. Salesbury, 1567
NT);41 (5) ‘mediator, umpire’, 16th–18th c.42
pedol f., -au, < pedalis, ‘horseshoe’; this meaning is not attested for the L word, so
that we have here a clear case of a special development in British Latin (the
possibility of a later and/or learned loan can be excluded).
Very probably, these two words canol and pedol were borrowed from L not as
adjectives but as substantives, that is to say, canalis and pedalis were already
nouns in British Latin. The adjectival meaning given to canol by W. Salesbury is
founded on his own learned analysis — he knew of course that -alis is an
adjective suffix in L — or on analogy with genuine Celtic -ol (cf. above).

More loanwords in -ol are based on L words with other suffixes:

39
The third volume of GPC has not yet been published. Morris Jones gives no dates for his
examples.
40
Anyway, the suffix was rare in early MW. In CT, only greidawl ‘ardent, eager’, meidrawl ‘mighty,
royal’, and pressenawl ‘wordly, earthly’ are found (and an unclear uydawl), whereas adjectives
in -awc (Modern W -og) are already fairly common.
41
The older adjective is canol-ig, c. 1400+.
42
Written cenawl 16th c., canawl, canol, 18th c.; clearly substantival use of the adjective.
abostol < apostolus (or rather -um) ‘apostle’, 14th c. + (the form ebostol, with e-
from the pl. ebestyl < apostoli, 12th/13th c. +, is already attested in the 13th c.;
later variants are abostl, apostol). The second meaning of ebostol, viz. ‘story,
epistle, homily’, 13th c. + (first in ebostol pwyll aphryderi), either goes back to L
epistula (the expected form *ebystol could be influenced by abostol) or reflects (in
shortened form) the formula epistola S. Pauli apostoli often to be heard in the
Latin mass. The L genitive apostoli should, however, have given W ebestyl (cf. the
pl. mentioned above).
Aircol < Agricola PN (ruler of Dyfed in the 6th c., cf. CT I 13)
baeol m., -au, < *baiula ‘pot, pitcher, bucket’,43 c. 1200+ (= OCorn. baiol, Bret. béol,
therefore a Common British loan). u-Umlaut also in the following two words:
[274]
rheol f., -au, < regula ‘rule, order’
tafol f. < tabula ‘scales’.

The two words ysgol ‘school’ < schŏla and ysgol ‘ladder’ < scāla became
homonymous only in Modern W (MW ysgol ‘school’, but ysgawl ‘ladder’).

5.2 Other suffixes

Other W suffixes belonging to this category will only briefly be indicated
here.44
5.2.1 -ain(t)

is found in a variety of loanwords:

cenfain(t) < conventio(nem), the W meaning being ‘herd, swarm (of pigs, bees,
etc.), multitude’, 12th/13th c. +; MW Cysteint < Constantinus PN; Ger-aint < brit.
*Gerontos < Gerontius PN; gwain ‘sheath’, 13th c. + < vagina(m) (= OIr faigen);
Owain < Eugenius PN (disputed); plygain ‘early morning’ < pullicantio(nem); putain
‘whore’ < putania(m),

and in genuine Celtic words:

rhiain ‘queen’ < British rīganī < CC *rēg-ən-ī ‘she who belongs to the king’ (= OIr
rígain); elain ‘hind, fawn’ < British *elanī ‘the red-brown (vel sim.) female’;

43
This meaning is not attested in literary sources, but safely established as Vulgar L by
Romance, Germanic and the Celtic reflexes mentioned above (see W. von Wartburg,
Französisches Etymologisches Wörterbuch, vol. I (Bonn, 1928), p. 205–6.). bajula aquae is first
found in Medieval Latin texts.
44
I hope to publish the details elsewhere.
ysgyfaint ‘lungs’, lit. ‘the two light ones’ (= OCorn. sceuens, Bret. skevent, all <
British *skam-ant-ī, dual).45
The host of abstracts in -aint (of the still very productive type enn-aint ‘ointment’
< British *an-neg-antio- < CC *an-nig-antio- from the root *neg- ‘to wash,
annoint’, cf. OIr nigim ‘I wash’; and hen-aint < CC *sen-antio-) need not to be
discussed here.46 [275]

5.2.2 -od

loanwords:
arawd f., arodion < oratio(nem)47 ‘prayer’, 10th c. + (OW araut)48 (= OIr oróit);
bydysawt mf. < ba(p)tizatio(nem) ‘baptism’, 12th/13th–15th c.;49esgusawt m., -
odion/-odau < excusatio(nem) ‘excuse’, 12th c. + (cf. further vn esgus-o and
esgusod-i, both 14th c. +); ffawt mf. < fatum or fata50 ‘fate’, 12th/13th c. +); parod a.
‘ready, prepared’ < paratus; pechod m., -au < peccatum ‘sin’; priod mf., a. < privatus
‘property’,51 ‘proper’ and ‘husband, wife, married’; pysgod c < piscatum (‘what has
been fished’) ‘fish(es)’; besides pysg c < piscis ‘fish’; and the W singulative pysgod-
yn ‘a fish’; traethawd m. < tractatum52 ‘essay, treatise’; ysbawd m. < spatum? (cf.
ysbodol ‘ladle’ < spatula (a modern learned loan?) ‘shoulder, piece of meat from
the shoulder’.53

Genuine Celtic words ending in W -od stem from a variety of sources. Most
examples are CC formations with dental suffixes (-to-, -ti-, -tu-), e. g. defod f. < CC
*dem-āto- ‘rite, ceremony, custom’, c. 1200+ (= OBret. domot gl. ritum); diod f. <

45
OIr has a different formation from the same basic adjective scam (masculine o-stem) ‘light’ (=
W ysgafn with prothetic y- and formans -no, Corn. scaff, Bret. skañv): its pl. scaim means ‘lungs’,
in opposition to tromm ‘heavy’, trom-chride (lit. ‘heavy heart’) meaning ‘liver’. (Lungs float when
washed.) Semantic parallels are e. g. E lights ‘lungs of slaughtered animal’, Russian lëgkie
‘lungs’, sg. lëgkoe ‘lung’ (: lëgkij ‘light’), Portuguese leve ‘lungs’ (VGK I 76 and LEI A S-31).
46
Loans from languages other than L may also have contributed to the productivity of -ain(t), e.
g. distain < OE disc-þegn ‘chamberlain’, yswain < ON sveinn ‘comrade, mate’.
47
But see LEIA 0-31: oróit < orate (2PlImp); oratio > OIr ortha.
48
Later also other meanings: ‘speech, hymn’.
49
Later also ‘the whole world’, probably < ‘Christianity’.
50
Cf. anffawd ‘misfortune’ (= Corn. anfus).
51
This meaning is obsolete today.
52
Could be a modern learned formation (pseudo-loanword).
53
ffosawd m. ‘battle, wound, stroke’, 13th–17th c. (also written ffosod) is usually explained as a
loan from L *fossatum. To my mind, it is rather a (hybrid) W formation with the old
composition member -od- : ‘a stroke, a push to the trench (fossa, viz. the mass grave)’. There is
another argument against the explanation as a loanword: L fossatum means ‘ditch, trench,
grave’. Ennius has a verb fosso ‘to stick, pierce’, but this has nothing to do with W ffosawd.
CC *dī-āti- ‘drink’, c. 1200+ (= OCorn. diot gl. potus, MBr diet; cf. Ir deog/deoch with
a guttural suffix); aelod mf. < CC *(p)agl- ātu- ‘member’, c. 1200+ (= Mir áige). With
other words, -od serves as a plural marker and has developed into a collective
forming suffix. This started from old consonantal stems of the type llyg < CC
*lukōt-s (sg.): llygod < CC *lukot-es (pl.) ‘mouse, mice’,54 14th and 13th+ c.
respectively. Cf. further euod ‘worms’, malwod ‘snails’, pilcod ‘minnows’, tywod
‘sand’, etc. [276]

5.2.3 -ig
forms derivative adjectives of pertinence from substantives (seldom from
adjectives) and diminutives (special case of substantivized adjectives of per-
tinence). It goes back to CC *-īk- as well as to L -īc(i)us/-īc(i)a/-īc(i)um.55
loanwords:
chwysig c < vesīca, singulative -en ‘blisters’, 13th c. + (= Corn. gusigan, MBr
huysiguenn); lleithig f. < lectīca ‘bench, couch’, 13th c. +; llurig f. < lōrīca ‘cuirasse’,
12th/13th c. + (= OIr lúirech); Meurig < Maurīcius PN; Nadolig < Natalīcia
‘Christmas’; Seisnig < Saxonicus, -um ‘English’ (not for the language!); selsig f. <
salsīcia, pl. -od ‘black pudding, (American E) blood sausage’.

Some old words of Celtic origin:

adwyth-ig ‘deadly, evil, malignant’ : adwyth m ‘destruction, evil, disease’, both 13th
c. +; bonhedd-ig ‘noble, gentle’ : bonedd m. ‘noble descent, nobility’ , both c.
1200+; brenn-ig c ‘limpets’, first in dictionaries of the 18th c., but much older (=
Corn. brennik, Bret. brennik, brinnik, OIr bairneach);56dines-ig ‘urban, civic’, c.
1200+, : dinas ‘fortified town’ < CC dunon; ew-ig f. ‘hind, roe’, 12th c. + (OW eguic,
OCorn. euhic gl. cerva) (must have designated originally ‘sheep’, because of its CC
ancestor *ou̯īka);57 etc.

54
Later, the paradigms were restructured: llyg acquired a pl. llygon, llygod was understood as a
collective noun, and assumed a singulative llygoden.
55 55
There is at least one instance of L -ĭcus/-um: Saesnig (see above). The feminine Saxonica gave,
with regular ā-Umlaut, W Saesneg ‘English (language)’.
56
According to LEIA B-9 < *ber-īnā ‘belonging to the rocks’.
57
Belonging to L ovis, etc. OCS ovĭca ‘sheep’ (against ovĭnŭ ‘ram’) has short -i- in the middle
syllable. The W and Corn. words are therefore not inherited parallels, but a British formation
with the suffix *-īkā.
6 Celtic suffixes supported by loanwords

To the third category, that of Celtic suffixes which were supported by

loanwords with etymologically identical or phonetically similar suffixes, may
belong the following:

6.1 -es < *-issā

old W formations:
athraw-es ‘female teacher or tutor, mistress, governess’, 13th/14th c. +; [277]
brenhin-es ‘queen’, c. 1200+; Cymra-es < British *Kombrogissā, today ‘Welsh
speaking woman’, c. 1200+; duwies/dwywes/dwyes ‘godess’, c. 1300+; etc.58

I suppose that some loanwords from L, and much later from Anglo-French
(type meistres) have contributed to the still living productivity of the type (cf.
modern hybrids like plismones).

Of L origin could be:

abadess < abatissa ‘abbess’, 13th c. + (= Bret. abadez), and Saesnes < British Latin
*Saxonissa ‘Englishwoman’; less certain is lladrones < latronissa (see foregoing
footnote).59

6.2 -in
The suffix -in forms adjectives mostly from substantives, usually names for
materials in the case of the oldest examples. It goes back to different foreforms,
but in general, these words continue the IE type of -no- adjectives. The suffix is
practically no longer productive today,60 but there are a number of MW
formations. I suppose that some loanwords from L enhanced its popularity
during the Middle Ages:

cegin f. < British L cocina(m) < coquina ‘kitchen’, c. 1200+; Cystennin < Constantinus,
PN; dewin m. < British L devinum < divinus ‘wizard, sorcerer’, 13th c. +, cf. further
sêr-ddewin ‘astrologer’; Lladin < latinus, -um, -e ‘Latin’, 13th c. + (with learned

58
Many words of this type are first found in Dafydd ap Gwilym: bugeiles, ellylies, hodoles,
lladrones.
59
The other types of formations in -es are no longer productive: collectives (branes ‘flight of
raven’, 13th c. +; llynges ‘fleet’ = OIr longas; etc.), and place terms (buches ‘milking-place,
cowshed’, c. 1200+; llawes ‘sleeve’, c. 1200+; lloch(w)es ‘hiding-place’, 14th c. +).
60
The recent Ε words in -ing taken over with the W-looking ending -in are excluded from
discussion here.
restitution of -a-; the regular reflex of L latine would be *Lledin61); melin f. <
molina(m) ‘mill’; pererin m. < peregrinus (or rather -um) ‘pilgrim’.

Some of the oldest W words are:

brenin m. (MW breenin trisyllabic) ‘king, ruler’, 12th c. + (= Corn. brentyn, bryntyn),
< *brigantīno-, cf. braint ‘privilege’; byddin f. ‘army, host’, c. 1200+ (= OBret. bodin,
MIr buiden), < *budīnā;62 cennin c ‘leeks, yellow narcisses’, 10th c. + (OW cennin gl.
cipus) (= OCorn. kenin, Bret. kinnen, [278] MIr cainnenn), : can(n) ‘white’; elin mf.
‘ell, elbow, corner’, 9th c. + (= OCorn. elin, Bret. elin, ilin, OIr uilenn), < *olīnā, cf.
(with different ablaut grade) Gk ὠλένη, L ulna; ewin mf. ‘(finger-/toe-)nail, claw’,
9th c. + (OW eguin) (= OCorn. euuin gl. unguis, OBret. eguin gl. adungem, Bret.
ivin), < British *aŋīnā which, together with OIr ingen, stems from IE *ghīnā;
cf. further cyfewin ‘exact, precise’, lit. ‘with complete claws’ (of cats, in the Welsh
law-books), 13th c. +; gorllewin m., a. ‘west(ern)’, 9th c. + (OW 63guolleuin), < *upo-
leug-īno- : lleu ‘light’?; gwerin mf. ‘people, group’, 9th c. + (OW guerin translating
factio) (= OBret. guerin gl. in duas factiones, OIr foirenn gl. factio), < *arīnā : √
*(s)er-; etc.

6.3 -yg and -eg

These two suffixes, originally belonging to one paradigm as

masculine/neuter and feminine endings, are inherited from CC *-iko-, -ikā and
borrowed from L -icus, -a, -um. -yg is much more rarely found in W than -eg which
became very productive in words for languages, sciences, and all kind of terms
from the range of speech and communication; but there are also formations
from other semantic fields.

Loanwords:
benffyg/benthyg m. < beneficium ‘loaned/borrowed thing’,64 10th c. + (the old form
— cf. OW binfic gl. beneficium — is still alive in S. W., the new one, due to
regressive assimilation of the labial spirant to the foregoing nasal, is attested
from the 15th c. on); diffyg m. < dificium ‘lack, want’, 9th c. + (OW dificiou gl.
dispendia); ‘decrease, gap, eclipse’, 13th c. +; lleyg m. (f. lleyges, m. also llyg) <
laicus ‘layman, laic’, 9th c. + (OW leeces gl. maritae); ‘profane, unprofessional’,
15th c. +; meddyg m. < medicus ‘surgeon’, 13th c. +; Saesneg < Saxonica <lingua>
‘English’.
61
Which is perhaps reflected by OE læden, se the contribution of A. Wollmann, in this volume.
62
For the etymology, see LEIA B-114.
63
The -r- appearing since the 14th c. may be due to a suffix doublet *upo-ro- besides *upo-.
64
Since the 16th c. also as a ‘borrowed, loaned’.
Old W formations:
helyg c ‘willows’, 12th c. +, < British *saliko-. OIr sail, genitive sailech, continues
the consonantal stem (like in L salix); in British, the thematization should be quite
early. I further postulate an old Celtic derivative *meddyg a. ‘of mead, with mead’,
in meddyg-lyn m. ‘drink of mead’, 14th–17th c. six attestations (+ dictionaries) :
medd ‘mead’; < CC *meduko (cf. Vedic madhuka-, first as PN in
Śatapathabrāhmaṇa).65 [279]

The oldest W adjective formations designating languages are Cymraeg, 1188+

(Giraldus Cambrensis, Descriptio Cambriae, cap. 7: Kembraec, linguam
Kambricam);66 Ffrangeg, 13th c. +; Groeg, 13th c. +; Gwyddeleg ‘Irish’, 13th c. +;
Brythoneg ‘British, Welsh’, 15th c. +, may be older, cf. Bret. Brezonek.

Very early, adjectives in -eg became substantives, e. g. cas-eg, pl. cesig ‘mare’,
12th c. + (= OCorn. cassec gl. equa, MBret. casec), < CC *kankst-ikā ‘she who
belongs to the stallion (or: the studfarm)’; cam-eg (: cam ‘bent, round’), name of
several different tools (e. g. ‘fellow of the wheel’) and also of fishes, c. 1200+;
gwarth-eg c ‘cattle’, c. 1200+ (= MCorn. guarthec).
The use of -eg for forming verbal nouns is extremely seldom. There is only
one early example: (e)hed-eg ‘to fly’, c. 1200+.
All the many terms for sciences and communication current today in W are
considerably later (mostly 19th c+). They are formed on analogy with gramadeg
‘grammar’, a learned67 loan from L grammatica, 14th c.+, and areitheg ‘speech’,
16th c. +, : araith ‘speech, utterance, phrase’(= Corn. areth; OIr airecht ‘assembly’).

6.4 -os
This rarely found suffix forms collectives (with diminutive connotation in
modern formations). It is difficult to assume that all words are formed in analogy
to the only loanword known to me, viz. ceirios < cerasia ‘cherries’, 14th/15th c. +.
Therefore, I propose to start from a British suffix *-āsia, which became MW -aws,
Modern W -os, and was perhaps assisted in developing a certain productivity by
loanwords like ceirios.68

65
The slightly older meddyglyn ‘medicinal draught, medicine’, 13th c. +, is to be connected with
meddyg ‘surgeon’ and should not be taken together with the word above under one lemma,
as GPC does.
66
OIr Combrec is a loan from British *Kom-brog-ikā or from this OW form at the very latest.
67
Because of the ā-umlaut which was working in the 5th c.
68
Other loanwords like achos ‘cause’ < occasio were too far away semantically.
The W collectives in -os are:
careg-os and cerig-os ‘pebbles, small stones’ (in dictionaries of the 17th c. only) :
carreg, pl. cerrig ‘stone, rock’; drein-(i)os ‘thorns’, 16th/17th c.: drein, pl. of draen
‘thorn, prickle’; gwerin-os mf., c ‘common people, poor people, mob, populace’,
17th c. + : gwerin (see above); gwern-os ‘stunted alder-trees, aldergrove’, 15th c.
once, also in place names, : gwern c ‘alders’; helyg-os ‘willows’, 14th c. once, also in
place names, : helyg (see above); [280] llinos f., Pl. -od / -aid / -ion / -ydd ‘linnet,
greenfinch’, 13th c. +, probably < ME < OFr + W suffix; marwydos ‘embers’,
14th/15th c. +; plant-os ‘little children’ : plant c ‘children’; tei-os ‘huts, small houses’
: tai, pl.69 of tŷ ‘house’; ŵyn-os ‘lambs’ : ŵyn, pl. of oen ‘lamb’.70

7 The loanwords as historical sources

The results of this short survey agree with the known historical facts (cf.
introduction).
There is a fair number of words which came into Welsh due to the effect of the
Christian mission in Roman Britain: abadess ‘abbess’ [Addendum: rather a medieval
formation], allor ‘altar’, arawd ‘prayer’, bydysawt ‘baptism’, cangell ‘chancel’, cardod
‘charity, alm’, catgor ‘fast days’, creawdwr ‘creator’, gwyrfdod ‘virginity’, Nadolig
‘Christmas’, pechod ‘sin’, pererin ‘pilgrim’, trindod ‘trinity’, ufylldod ‘humility’, undod
‘unity’.
The majority of the loanwords were borrowed from the language of the
Roman Army, cf. the administration terms awdwrdod ‘authority’, benffyg ‘loan’,
castell ‘fortified place’, ciwdod ‘city’, canghellor ‘chancellor’, tymor ‘season’,
ymherawdr ‘emperor’, ysgub ‘sheaf’ and ysgubor ‘barn’ (supply planning), all
names of the days, of the week and of the months. Terms for military equipment
are fflangell ‘whip’, llurig ‘breastplate’, mantell ‘mantle, cloak’, pabell ‘tent’, pedol
‘horseshoe’. Other words reflect the not always easy life of the legionaries: astrus
‘difficult, ambiguous’, cymell ‘incitement, urge, compulsion’, dolurus ‘painful’,
diffyg ‘difficult’, llafurus ‘laborious, painstaking’; from their manœuvres, the word
asgell ‘wing’ became notorious. Medical care in the army camps and the Roman
towns was probably better than traditional British healing practices, but nothing
very sophisticated survived into Welsh: the meddyg’s (‘surgeon’) diagnosis seems
to have been mostly chwysig ‘blisters’. Roman construction craft must have

69
If teios is an old formation (which is unknown to me because the word has not yet been
published in GPC), tei- could represent the old stem *teges-. Alleged derivations of
characterized plural form are always doubtful, but some sure examples exist.
70
effros (a plant’s name), c. 1400+, is according to GPC a transformation of ewffras (< Euphrasia,
the L name of the plant). euros ‘sunflowers’, 19th c. +, could be a shortening of *eur-flod-os vel
sim. ‘gold(en) flowers’ or represents eur-ros ‘gold-roses’ i. e. ‘golden roses’ (so GPC).
impressed the Britons as every other people, cf. astell ‘lath, plank, board’
(product of the Roman plane), cafell ‘sanctuary, temple, cell’, canol ‘canal’, pibell
‘pipe, duct’, tafell ‘slice, slab’ (could also be originally a kitchen term only), [281]
ysgafell ‘ledge’, ystafell ‘room, chamber’.
A prominent place within the lexicon of words borrowed from Latin is
occupied by kitchen terms. The Romans obviously introduced a remarkable
progress in culinary culture (unfortunately, later barbarian invasions have
destroyed most of that!). Basic terms like cegin ‘kitchen’ [Addendum: coeth 'pure, fine,
excellent' < coctus 'cooked' i.e. prepared in the Roman way'], and melin ‘mill’ are Latin,
and also different utensils like baeol ‘pot, pitcher, bucket’, callor ‘cauldron, kettle’,
cawell ‘basket’, cyllell ‘knife’ (perhaps also cyllellawr ‘set of knives’), hestor
(approximately) ‘bushel’, padell ‘pan, bowl’; and dainties like selsig ‘black pudding’,
porchell ‘porkling’, pysg(-od) ‘fish(es)’, and ysbawd ‘shoulder( meat)’.
Typical for army slang are finally the expressions ceudod < cavitatem for
‘mind, heart, bosom’, and priod < private for ‘girlfriend, intended’ (only later ‘wife,
husband’).

[Zusätze:

zu2:
Harald Haarmann, Der lateinische Lehnwortschatz im Kymrischen, phil.diss. Bonn
1970, was only published outside the common book market (Romanistische
Versuche und Vorarbeiten, 36, Romanisches Seminar der Universität Bonn). It
therefore was, unfortunately, unknown to me when I wrote this article in 1988.
He has a chapter on "Lateinische Elemente in der kymrischen Wortbildung" (pp.
159-165). He certainly is wrong, e.g., in stating that W -us comes from VL *-ōsu
only (cf. above, para 4.5.). Haarmann's "inselkeltisch" for "British Celtic", with the
exclusion of Goidelic, is an error. His analysis of the W lexicon of L origin on the
lines of Dornseiff's "Wortschatz nach Sachgruppen" remains, nevertheless, a
very useful study. Any further study of the subject treated here should take
Haarmann's thesis into consideration.

zu 5.1: Other good examples are

canhwynol / cynhwynol 'innate, natural, congentital' < *continuālis, according to
Russell CWF 128.
OW decaul 'decade' < decim-ālis, attested in the gloss or deccolion gl. decadibus
(Mart. Cap.).
OW feciaul gl. fascia '(face) shroud' < faciālis.]
zu 6.1: Latin -issa is itself a loan from Greek -ισσα. For this, and the problem of
loan suffixes in general, see Michèle Fruyt, L'emprunt suffixal du latin au grec.
BSL 82/1. 1987 [recte 1989], 227-55.]