Origins of Welsh (and/or Brythonic) plural suffixes

[Jonlang]

I'm looking for some information on the origins of Welsh plural suffixes. I'm interested in when and how they arose. There are about a dozen suffixes for forming plurals and two (I think) for forming singulatives from collectives – something else I am interested in learning more about the development of in Welsh (or the wider Brythonic family). Anyone know of any good reading materials on this?

[alice]

I'm looking for some information on the origins of Welsh plural suffixes. I'm interested in when and how they arose. There are about a dozen suffixes for forming plurals and two (I think) for forming singulatives from collectives – something else I am interested in learning more about the development of in Welsh (or the wider Brythonic family). Anyone know of any good reading materials on this?

You might want to look for J. Morris Jones's "A Welsh Grammar, historical and comparitive". According to section 119, the plural suffixes are the regular descendants of the consonant-stem noun suffixes, which were lost in the singular.

Or you could hunt down Dewrad and bribe him with some free phonemes.

[Linguoboy]

I'm looking for some information on the origins of Welsh plural suffixes. I'm interested in when and how they arose. There are about a dozen suffixes for forming plurals and two (I think) for forming singulatives from collectives – something else I am interested in learning more about the development of in Welsh (or the wider Brythonic family). Anyone know of any good reading materials on this?

You might want to look for J. Morris Jones's "A Welsh Grammar, historical and comparitive". According to section 119, the plural suffixes are the regular descendants of the consonant-stem noun suffixes, which were lost in the singular.

Not just consonant stems. For instance, the termination -oedd can be equally well derived from the i-stems.

Historical phonological works like Jackson's Language and History in Early Britain also cover the topic but in a less focussed way. That is, he mentions the origins of -oedd when he discusses how /ð/ arose from earlier intervocalic *j (which itself sometimes developed from the loss of intervocalic *s), but there's no single section where he summarises all of the plural developments. He also doesn't talk about how certain plurals spread by analogy after being associated with particular semantic areas (e.g. -od for animals, -aid for humans), which is one of the most interesting aspects of the process.

So although I haven't read Morris Jones' work, I agree it's probably a good place to start.

Or you could hunt down Dewrad and bribe him with some free phonemes.

Good luck with that!

[Richard W]

You might want to look for J. Morris Jones's "A Welsh Grammar, historical and comparitive". According to section 119, the plural suffixes are the regular descendants of the consonant-stem noun suffixes, which were lost in the singular.

Additionally, some loanwords have actually brought their foreign plurals in.

Read consonant stem liberally; -oedd comes from the i-stem ending, and -au comes from the u-stem ending.

[Travis B.]

/me wonders how one bribes someone with free phonemes.

[Richard W]

Jones's work is available at Wikisource.

[Jonlang]

Thanks all! I will certainly be reading up on the Morris Jones book.

[alice]

/me wonders how one bribes someone with free phonemes.

I got mixed up between "bribe with phonemes" and "persuade with free phonemes". Welsh does that to you.

[dewrad]

Ahem.

Modern Welsh rejoices in between eight and twenty different methods of plural formation, dependant on the speaker, dialect and analysis. Frankly, this is ridiculous and is more a reason that my native language needs to have a little word with itself than any whingeing about mutation or "not having words for yes and no". A synchronic analysis of these various pluralisation strategies invariably ends in folly, so I shall follow my instincts and preferences in offering a diachronic perspective.

Basically, alice and linguoboy have the right of it: they're essentially remnants of the varying stem formations in Proto-Celtic. Sensu latissimo, we reconstruct nine(ish) stem classes in the language ancestral to Brythonic: analogous to the declensions of Latin or Greek. All of these are relatively synchronically straightforward, and I offer an example of a noun in the nominative singular and plural of each in the table below:

stem class	nominative singular	nominative plural	gloss
ā-stems	merkā	merkās	'maiden'
o-stems	mapos	mapī	'son'
i-stems	vlatis	vlatejes	'dominion'
u-stems	katus	katowes	'battle'
n-stems	altrawū	altrawones	'foster-father'
t-stems	lukots	lukotes	'mouse'
nt-stems	karants	karantes	'friend'
s-stems	tegos	tegesā	'house'
r-stems	brātīr	brātres	'brother'

(There are also ī-stems, but I'm doing this from memory and I can't think of any. The existence of ū-stems is not relevant for Brythonic, and is probably a figment of Holger Pedersen's imagination.)

I've chosen these nouns for a reason. Look at their Middle Welsh reflexes:

stem class	nominative singular	nominative plural	gloss
ā-stems	merch	merched	'maiden'
o-stems	mab	meib	'son'
i-stems	gwlad	gwledydd	'dominion'
u-stems	cad	cadau	'battle'
n-stems	athraw	athrawon	'foster-father'
t-stems	llyg	llygod	'mouse'
nt-stems	câr	cerynt	'friend'
s-stems	tŷ	tei	'house'
r-stems	brawd	brodyr	'brother'

So, we can see what happens: with the o-stems we see i-mutation from *-ī. With the other classes, we can see that the final syllable is dropped in both singular and plural - the Middle Welsh plural forms are basically the oblique stems of Proto-Celtic.

However, the first of these is not like the others: the expected reflex of *merkās would be merch, the same as the singular. Instead, a form of the t-stem ending has been imported. Modern Welsh essentially ignores the diachronic origin of a noun and runs rampant with analogy. For example, the plural of cath 'cat' is cathod, but given the fact that it was originally a u-stem we would expect cathau (which is indeed attested in Middle Welsh). However, it has taken the ending -od (probably) from milod 'animals'. In general, the most common plural suffix in Welsh is -au (same in Corish with -ow and Breton -où), but u-stems were some of the least common nouns in Proto-Celtic.

Also, llyg and llygod are interesting here: in Modern Welsh the singular is not llyg but rather llygoden: it's a singulative. These are a Brythonic thing not found in Goidelic (afaik): they basically derive from Proto-Celtic diminutives.

[dewrad]

Or you could hunt down Dewrad and bribe him with some free phonemes.

Good luck with that!

It's all about the 'Gram these days my dude

[Linguoboy]

Also, llyg and llygod are interesting here: in Modern Welsh the singular is not llyg but rather llygoden: it's a singulative. These are a Brythonic thing not found in Goidelic (afaik): they basically derive from Proto-Celtic diminutives.

There are a handful of Goidelic examples, but it was never a particularly productive suffix. For instance, Irish grán "grain" vs gráinne "a grain" (plural gráinní "grains"). The suffixed forms are best analysed synchronically as derived terms rather than members of the same lexeme, e.g. rón "horsehair" vs ruainne (pl. ruainní) "single hair" [of any origin, e.g. ruainne dá folt "a hair of her head"]; snáth "thread, yarn; [spider]web" vs snáithe (pl. snáitheanna) "thread".

[Kuchigakatai]

Also, llyg and llygod are interesting here: in Modern Welsh the singular is not llyg but rather llygoden: it's a singulative.

So what is llyg today? An archaic and rarely-used singular?

I love the whackiness of Celtic declension and mutation from half-assedly retained old distinctions so much.

[Linguoboy]

So what is llyg today? An archaic and rarely-used singular?

A shrew. (It has a distinct plural form, llygon.)

[Kuchigakatai]

A shrew. (It has a distinct plural form, llygon.)

Oh my. Well, I suppose that makes it like those Latin neuter nouns that ended up as two separate words in Romance, one from the singular ending in -o (or -u) and one from the plural in -a, but this Welsh word is more amusing because it ends up with the old n-stem plural, not even -au. Thanks.

[Jonlang]

Ahem.

Modern Welsh rejoices in between eight and twenty different methods of plural formation, dependant on the speaker, dialect and analysis. Frankly, this is ridiculous and is more a reason that my native language needs to have a little word with itself than any whingeing about mutation or "not having words for yes and no". A synchronic analysis of these various pluralisation strategies invariably ends in folly, so I shall follow my instincts and preferences in offering a diachronic perspective.

[...]

Also, llyg and llygod are interesting here: in Modern Welsh the singular is not llyg but rather llygoden: it's a singulative. These are a Brythonic thing not found in Goidelic (afaik): they basically derive from Proto-Celtic diminutives.

Thanks bro! Extremely helpful! Do you know of anywhere where I can read more about it?

[Frislander]

I'm just for interest's sake gonna throw in the Scottish Gaelic reflexes of as many of those terms as I can account for to try and provide a comparison.

stem class	nominative singular	nominative plural	gloss
ā-stems	beann	beanntan	'mountain'
o-stems	mac	mic	'son'
i-stems	flath	flaith/flathan	'prince'
u-stems	cath	cathan/cathannan	'battle'
n-stems	brà	bràthan	'quernstone'
t-stems	luch	luchan/luchainn	'mouse'
nt-stems	caraid	càirdean	'friend'
s-stems	taigh	taighean	'house'
r-stems	bràthair	bràithrean	'brother'

I couldn't find any reflexes of *merkā. Words deriving from the root *altra- can be found in words like dalta "foster child", co-alta "foster sibling" and altramadh "fostering" but the word in question doesn't appear to have a reflex.

Two things are important to note about the system. Firstly, the n-stem endings have been in large part generalised to other stem types, with most of the variation coming from whether the addition of an -n reveals any stem-final consonants that are otherwise lost (e.g. in "mountain" above). Sometimes these stem extensions find themselves analogised as whole plural endings for nouns in a similar semantic field, e.g. modes of transports form their plurals with an -ichean, e.g. from plèana "plane", bàta "boat" and càr "car" we get plèanaichean, bàtaichean and càraichean respectively. The other main plural formation is slenderisation, which has also been subject to some meaning-based analogy, e.g. sagart "priest" > sagairt, ultimately from Latin sacerdos but not using the original plural sacerdotes, probably by analogy with forms such as fear "man" > fir. This occasionally can also reveal lost stem-final consonants, as in caora "sheep" > caoraich, from *kaɸrāxs, or ironically the historic n-stem cù "dog" > coin. So the complexity in the system, unlike the pretty much direct continuation we see in Brythonic, in large part splits and levels the Proto-Celtic variation with new complexity from different sources, as evidenced by the variation we see in some nouns (e.g. sròn "nose" seems to vary between srònan, sròinean and sròintean depending on who you ask).

[Linguoboy]

I'm just for interest's sake gonna throw in the Scottish Gaelic reflexes of as many of those terms as I can account for to try and provide a comparison.

Oh, WTH, I'll do Irish, since it's undergone less levelling in this regard than Scottish Gaelic.

stem class	nominative singular	nominative plural	gloss
ā-stems	beann	beanna	'horn'
o-stems	mac	mic	'son'
i-stems	flaith	flatha	'prince'
u-stems	cath	cathanna	'battle'
n-stems	altra	altraí	'foster father'
t-stems	luch	lucha	'mouse'
nt-stems	cara	cairde	'friend'
s-stems	teach	tithe	'house'
r-stems	brathair	bráithre	'brother'

No examples here of the "strong" velar ending -acha, which has been extended by analogy at least as much as -anna, though it can be found on deartháir, a derivative of bráthair (via pre-reform dearbhráthair; the term bráthair was used so much in reference to clergy that the prefix dearbh "certain" was added to form a retronym designating an actual sibling), which is pluralised deartháireacha (Connemara deartháracha with deslenderisation).

The ending -aí on altraí derives from the d-stems and represents pre-reform -aidhe. (An etymological examples would be Old Irish arae "charioteer", nom.pl. araid; pre-reform Mod. Ir. ara, pl. araidhe.) -the is a t-stem accusative ending which was generalised to s-stems like teach (whose OIr plural tige should have yielded pre-reform tighe, as it did in the homophonous genitive singular), as well as to velar stems and others.

Even more so than in Welsh (since the semantic associations of the various plural endings seem to be weaker), plural variation is common in Irish. For example, take the Old Irish feminine n-stem aub "river", N/A plural aibne. The etymological plural aibhne is found in Munster, but so are the double plurals aibhnte and aibhní; the standard (originally Connemara) plural form is aibhneacha. For óráid "speech" (from Latin ōrātiō, but borrowed as an i-stem), Dillon and Ó Cróinín give the plural variants óráidí, óráideanna, óráideacha, and óráidíocha "all [of which would] be understood, and none [of which] would be shocking to a native speaker".

[evmdbm]

Oh, WTH, I'll do Irish, since it's undergone less levelling in this regard than Scottish Gaelic.
stem class nominative singular nominative plural gloss
ā-stems beann beanna 'horn'
o-stems mac mic 'son'
i-stems flaith flatha 'prince'
u-stems cath cathanna 'battle'
n-stems altra altraí 'foster father'
t-stems luch lucha 'mouse'
nt-stems cara cairde 'friend'
s-stems teach tithe 'house'
r-stems brathair bráithre 'brother'

Errm.... direct me to something to read if you like, but I see no "o" in mac, or "u" in cath or "n" in altra. I could go on, but how do I know what the stem is... without knowing Proto-Celtic because the stems seem only to correspond to their names there. Although even there I see no "n" in altrawu...

[Linguoboy]

Errm.... direct me to something to read if you like, but I see no "o" in mac, or "u" in cath or "n" in altra. I could go on, but how do I know what the stem is... without knowing Proto-Celtic because the stems seem only to correspond to their names there. Although even there I see no "n" in altrawu...

There's no n in the nominative/vocative singular, but there is in every other form. (Viz. Latin n-stems like ōrātiō.)

And you would know this the same way you would know the ablaut class of a strong verb in Germanic: by consulting a lexicographical work. These are just labels of convenience; one could just as well as name them after their model paradigms and call macc "a fer-class noun" or altra "talam-class". But that's just as arbitrary, more opaque, and less useful to anyone doing comparative IE work. Alternatively, you would memorise however many principle parts you need to identify the paradigm. I think for most Old Irish nouns it could be done with just the nominative and genitive singular (e.g. macc/micc, altra/altran), but I haven't tried this myself so I can't be sure.

This is far more information than you need for Modern Irish, btw, where the dozen or so classes of Old Irish and Proto-Celtic have collapsed to four plus a grab-bag of irregulars. For these, you need to memorise both the genitive singular and the nominative plural, if different. (Or just the plural; Ó Siadhail argues that the genitive is a fossilised relic in contemporary spoken Irish, but I would argue it's common enough in compounds and proper names that it's still worth learning.)

[MacAnDàil]

This is all very interesting, but isn't it easier to predict the singular using the plural than the reverse? Like with French adjectives?