Saturday, 27 September 2008
The Old Irish Verbal System
Also known as 'the most demanding intellectual challenge I have experienced in eight years at Oxford University.'
I thought I'd post this as a matter of possible interest to any Classicist readers (or any readers at all...) who might like to survey from a decent height a section of one of the more horrifically difficult languages of Europe.
In short, the Old Irish verbal system nearly did for me. I teach this language to postgraduates, and I simply can't hold it in my head all in one go, without constantly revising it. Now this is of course in part due to my appalling laziness, but some blame has to be ascribed to the notorious, deranged complexity of Old Irish Grammar.
You could make a case that Old Irish is a 'Classical' language, like Latin or Greek or Sanskrit. Latin? After a while, ordinary Latin ends up being more or less transparent to the reader. (Ordinary Latin, not Tacitus or Propertius.) You can read it quite cheerfully. And Greek I think always remains a bit trickier (the midgy drifts of particles, the propensity to dialectical forms, the specialised jargons). And Sanskrit is like an exotic holiday for Classicists: a new script, a complicated phonology, the system of sandhi-variations which obscure the endings, and a general rather bewildering mixture of stylization and lushness. Like the above trio, Old Irish is Indo-European, has a heroic literature, and grammatical features such as inflected nouns and adjectives, plus a complex conjugated verbal system.
But describing the Old Irish verbal system as 'complex' is like referring to the Arctic as 'somewhat chilly'. What follows is something of an exasperated gazeteer by a frustrated Celticist.
* * *
A wise old teacher of mine, now sadly dead, once told me that Greek grammar was like Latin grammar but with one extra of everything (with the exception of cases, where Greek has one less than Latin.) So Greek has an extra mood, the 'optative', and an extra tense, the 'aorist', and an extra voice, the 'middle'. Old Irish takes this principle to an entirely new and dismaying level.
First things first. Old Irish was the version of Irish written between 650AD and about 900AD, by the educated and aristocratic elites, who probably did actually speak it too. Ordinary people were likely already speaking a simplified variety of the language well on its way to becoming Middle Irish.
It has the usual complement of persons: a 1st, 2nd, and 3rd singular and plural. It has the usual number of tenses: a present, a future, a secondary future, an imperfect, a preterite, and a perfect (sometimes). There is no pluperfect; Welsh on the other hand developed one under the influence of Latin. There's a subjunctive, but no optative. There are imperatives. There is no infinitive; rather there is a verbal noun which is formed quite randomly from the verb, and which can belong to any of around 10 noun declensions. Often it looks nothing like its parent verb. (Would you guess that céim, 'stepping' comes from cingid?)
Now the trouble starts. Verbs can be Strong or Weak, and both stong and weak verbs can be simple or compounded. (Irish, like other Indo-European languages, is fond of putting prepositional preverbs onto verbal stems to make new verbs, like INTERficio or PROFfero from *pro-ob-fero, or indeed like English UNDERgoing, OVERgoing, OUTgoing, INgoing and so on. Old Irish can stick up to four of these onto one verbal stem, though one or two are more usual.) Bear this in mind.
The first major blanching-moment for the hearty Classicist comes when he or she says 'Ho there, my good language, let me see your present indicative paradigm!' Every verb turns out to have not one but two paradigms for most tenses. You use one - the 'conjunct' paradigm - when the verb is preceeded by a conjunct particle, which are pesky little words like the word for 'not', ní, or the question particle, in, amongst many others. You use the other paradigm, the so-called 'absolute', when the verb is NOT so preceeded. So the present indicative of 'glenaid', 'sticks fast', looks like this:
The colon before the conjunct endings indicates that a conjunct particle preceeds. In other words, the phrase for 'he sticks fast' is really somewhat different to 'he does not stick fast'. It's not just a matter of sticking a non or an ou in. Oh - did I mention that the pronunciation of Old Irish is infernally complicated? That each consonant has four values, depending on where it is in a word and whether or not it is 'palatal' (i.e. followed by a y- glide like the c in cute)? So even though the 1st persons singlular absolute and conjunct look identical on the page, the former is pronounced GLYENum, and the latter GLYENuv. [The capital letters indicate the stressed syllable, by the way.]The first column is in fact pronounced GLYENum, GLYENee, GLYENuδ, GLYENvee, GLYENthuh, GLYENud. The delta, δ, in the 2nd singular is a soft th- as in 'those'.
So. Joy! Joy! The territory to be covered for every verb has just instantly doubled with respect to Latin or Greek. And the pronunciation bears sod-all relationship to the spelling, all the persons sound similar, and they're bloody hard to remember. A nice crisp, rhythmic 'luo, lueis, luei, luomen, luete, luousi' it ain't.
'Glenaid' was a simple verb. (Incidentally, the erratic nature of Irish verbal nouns means they are not suitable for referring to a given verb in the abstract, like you use the Latin present infinitive active, e.g. 'the verb amare'. So we use the 3rd singular absolute present indicative instead.)
Our next hurdle comes when we arrive at compound verbs, which are very common. Essentially a prepositional preverb added to a verbal stem (like Latin ob-, or inter-, or re-, or de-) acts like a conjunct particle, so compound verbs always require the conjunct set of endings. Here's an example. The simple verb beirid, 'bear' has a present indicative paradigm as follows:
ABSOLUTE * CONJUNCT
But its compound relative, do:beir, 'bring, give', has a paradigm like this:
Note that it uses the conjunct set of inflections.
Now Old Irish has a very strong stress accent; in most words it falls on the first syllable, but in compound verbs it falls on the syllable after the first preverb. So BEIRid, /BYERiδ/, but do:BEIR, /do-BYER/. La la la. But then what happens when you want to prefix, say, the negative particle onto a compound verb, to form a simple phrase like 'she does not give'? The answer is that the stress shifts onto the first element after the negative particle. Unfortunately this causes a whole series of hideous phonological side-effects which often totally alter the appearance of a verb.
do:BEIR 'she/he/it gives'
ní:TABair 'she/he/it does not give'
The d- of the do:- hardens to a t-, the vowel alters, and, though this does not appear orthgraphically, the B in 'TABair' is pronounced as a -v-. Working out that a word pronounced 'TAVur' was the direct cognate of Greek phero and English bear took 19th century Indo-Europeanists some time, and who can blame them. It's worth pointing out here that this level of explanation is required to get us to the 'Caecilius est in horto' stage of the language. And we've not even mentioned nouns yet...
At this stage it starts to feel like you've had a heavy night. But it gets so much worse...
Compound verbs like 'do:beir' are called deuterotonic forms, because the stress goes on the second syllable. Forms like 'ní:tabair' are called prototonic because the stress has shifted to the first syllable of the actual verb. So (wait for it) in compound verbs, prototonic forms are to deuterotonic forms what conjunct forms are to absolute forms in simple verbs.
Jesus. I remember doing this for the first time, a degree in Classics supposedly under my belt, and thinking: 'Ok.....' Five minutes later, my forehead was so furrowed I could have held a pencil vertically between my eyebrows.
The problem with the prototonic forms of compound verbs is that the sound changes can be very radical, squidging the consonants together. In many, many verbs this utterly obscures the relationship between deuterotonic and prototonic forms of the same verb.
Here's a random sample - some subjunctives:
do:gnéo, 'I may do', but ní:dén, 'I may not do'
ad:cether, 'he may see', but ní:accatar, 'he may not see'
do:coístis, 'they might have gone' but, in:dechsaitis, 'might they have gone?'
I should point out that these are all more or less normal. The rules governing the sound changes (and the above examples are far from being the most extreme, as the more preverbs a verb has the less closely the deuterotonic and prototonic forms will resemble one another) are so complex that each compound verb essentially has to be learned by rote. Here are some deliciously counter-intuitive (yet historically justified) forms:
im:soí, 'he turns around', but ní:impai, 'he does not turn around'
do:fuiben, 'he destroys', but ní:tuiben, 'he does not destroy'
And of course this insane system makes it impossible to look anything up in a dictionary. If you find a form ní:cumgubat how the hell are you supposed to know that it comes from con-icc? Or that ní:ralae comes form fo:ceird?!
Even worse, this sleep of philological reason brings forth monsters. Latin has 'semi-deponent', 'quasi-passive' and 'defective' verbs - the blind and the halt of its verbal system. But only Old Irish has a class of (wonderfully named) contaminated verbs, which occur when the writers of the language have got so confused by this constant, deranged shifting of verbal outlines that they get two or more verbs hopelessly muddled. Thus they create a hideous class of unruly hapax legomena. (For example, fo:tlen 'removes secretly' and as:luí 'get away, escape' breed to produce the bizarre fothlai, 'wean', i.e. to take a calf from its mother.) It's as though you had to parse an imaginary Latin verb form **pepetivit, which appeared to mean 'he sought out in order to destroy'. You'd have to say 'Oh yes, this is the 3rd singular perfect indicative active from peto 'I seek' but, ah, er... also clearly kind of from pello, 'I strike', as well...' Quite, quite mad.
Anyway. This whole absolute:conjunct, deuterotonic:prototonic malarky holds for almost all the tenses, not just the present, and is normal in the subjunctive too. And it's not as if there is one nice conjugation for all verbs. Weak verbs and strong verbs conjugate differently, and there are subclasses to both. There's a further class of 'hiatus' verbs. Irritatingly, scholars have been unable to agree on precisely how to divide up the classes of the verbs, so that we currently have three different systems with three different classificatory terminologies all operating side-by-side. But however you array them, all these sub-types have deponent verbs in them too. These are not really terribly like Latin deponents, as that would be too bloody simple. Also, several verbs can get confused as to which class they belong to, and adopt endings which are not strictly proper to them. And did I mention that Early Irish scribes were very bad at spelling, so that the form one actually encounters may be spelled somewhat differently to the 'correct' form in the Grammar books?
* * *
It goes on. The next hideous obstacle are the object pronouns. Like Latin and Greek, a pronomial subject nestles quite happily in the verb ending. do:beir means 'he/she/it gives' - there is no need for a separate pronoun to express the subject. Trouble is, OBJECT pronouns simply don't exist as independent entities in Old Irish. They must be infixed within the verbal complex. These are the notorious 'infixed pronouns', a series of single letters or syllables that identify a pronomial object, and which slot into the verb. The principle is easiest to see in a nice compound verb. The infixed pronoun fits in between the preverb and the verbal stem:
do:beir 'he/she/it gives'
dom:beir 'he/she/it gives ME'
dot:beir 'he/she/it gives you [sg.]'
da:mbeir 'he/she/it gives him'
dos:mbeir 'he/she/it gives her'
The intrusive m- before the b in both the 3rd singular forms are due to the famous consonant mutations of the Celtic languages, which have no parallel in either Latin or Greek. (Although interestingly, the minuscule and bonkers Romance language of Sardinia, Sard, has independently developed something slightly similar.)
If the preverb happens to the negative particle, the system holds. Here goes with a simple verb:
ním:beir 'he does not carry me'
nít:beir 'he does not carry you'
ní:mbeir 'he does not carry him'
nís:mbeir 'he does not carry her'
And here with a compound verb:
ním:thabair 'he does not give me' NOT **ní dom:beir
nít:thabair 'he does not give you'
ní:tabair 'he does not give him'
nís:tabair 'he does not give her'
[Again, the change of t-->th in the 1st and 2nd persons singular is a result of consonant mutation.]
But what if you want to use a pronomial object with a simple verb, like beir, 'he carries'? How do you say 'he carries me'? There's nothing to infix the pronoun onto!
Well, when all else fails, just make it up. The early Irish came up with a completely meaningless dummy preverb to attach to uncompounded verbs under these circumstances. This preverb, in a spectacular outbreak of unhelpfulness to unborn English-speakers, takes the form no. The conjunct endings are of course (of course! [wan smile]) obligatory:
nom:beir 'he carries me'
na:mbeir Oh, you get the idea...
Of course, they couldn't leave it that simple. Not only is there an archaic system of suffixed object pronouns which were used in preference to the no system in Early Old Irish, there are in fact also three different classes of infixed object pronoun, known as A, B and C. Which one you use depends on the syntax of the clause in question, and on whether the preverb ends in a vowel or not.
* * *
We're nearing the end of this litany of tears and wailing.
At this point I should say something about the verbs 'to be'. Yes. Verbs, plural. Old Irish (and indeed, Modern Irish and Scottish Gaelic) has two verbs 'to be'. In Old Irish, each has a complete paradigm, including absolute and conjunct forms. One, the so-called 'copula' is used to link a predicate to another noun or adjective: 'Old Irish is vile', 'Blogs are a liability'. The other, the so-called 'substantive verb', is used to link a predicate to an adverb or a prepositional phrase: 'He is over there', 'the cat is under the table'. I am told that the distinction resembles that between ser and estar in Spanish.
Naturally, they look quite different, at least in the present tense; but outside the present tense, they look bewilderingly similar, usually taking the form b+vowel+something else. Ba, bi, bie, bad, bud, bo.... it all blends into one.
A particular swollen chancre is the conjunct form of the substantive verb. Not only does the absolute form come ultimately from a verb that means 'to stand', to form the conjunct the early Irish decided to press an archaic form of 'to see' into service. The form that came to hand was some kind of imperative, fil, 'see!'. So to say 'Boris is not under the table' you have to say: 'See Boris not under the table!'. As that English example indicates, the logical subject actually becomes an object, and goes in the accusative. And when the logical subject is a pronoun, it naturally becomes an object pronoun. ('See me not under the table!'). And that means it has to be infixed:
Ním:fil under the table.
I am not under the table.
Such joy. In theory it's as though in French you had to say 'me voici (from 'voi ici', 'see here') au dessus de la table' when you meant **'Je suis au dessus de la table', but it feels so much more complicated.
* * *
I've said nothing so far about the very peculiar syntax of the Irish verb, such as the Verb-Subject-Object order which is pretty compulsory. I'll pass over it in silence, and note instead the ludicrous system for doing passive constructions.
Unlike Latin or Greek, Old Irish does not have a full complement of passive forms for every person for every tense. There are only specialised passive forms for the 3rd persons, both singular and plural. Natch, these have absolute and conjunct or prototonic and deuterotonic forms too:
marbathair, 'he is killed'
ní:marbthar, 'he is not killed'
marbaitir, 'they are killed'
ní:marbatar, 'they are not killed'
If, as well you might, you want to say 'I am killed', or indeed use any other person other than the 3rd, you take the 3rd person form and infix the subject as an object pronoun, after the preverb, or if there isn't one naturally, after the dummy preverb no:
nom:marbthar, 'I am killed'
ním:marbthar, 'I am not killed'
* * *
There are further things I could bring in here. There are numerous badly irregular verbs. Each tense has (bizarrely) three specialised relative forms, a 3rd singular one, a 1st plural one, and a passive singular one. (So 'the man who sells linen' needs no word for 'who'; there is a special form of the verb which in and of itself means 'who sells'.) Despite having this absurdly complicated system of tenses and forms, the early Irish still found it necessary to create progressive tenses as well, using the substantive verb as an auxilliary. There's also a funny little 'augment', ro, one function of which is to turn a preterite into a perfect, and which can support an infixed pronoun. The subject intrinsic within a finite verb-form can also be shored-up or buttressed by a number of 'emphasising particles' which normally look like -si or -so or -se- or -som. These get tacked onto the end of a verb, which may already have mutimple preverbs and an infix: for example, don:gniat-som, 'they make us'. Kindly editors separate the various preverbs and infixes and particles and suffixes with colons, dots and hyphens: the scribes just wrote them as single words, like dongniatsom, and left readers to fend for themselves.
Sigh. By around 900AD, dissention was growing in the ranks at having to write and speak this horrific language. Things started to collapse. The linguistic centre could not hold. Infixed pronouns began to be jettisoned, in favour of independent object pronouns like English or Latin. The absolute:conjunct distinction was gradually, painfully abandoned, with many strong verbs being remodelled as weak verbs on the basis of their verbal noun. Compound verbs became locked in their prototonic forms. Some of the wilder tense formants were discontinued. The entire island breathed an enormous, heartfelt sigh of relief.
This explains why Old Irish is a favourite hunting ground for Indo-Europeanists, and is easily amongst the very hardest of the older languages of Europe, and yet its direct descendent, Modern Irish, is about as hard as French. The most recent introduction to the old language, David Stifter's Sengoidelc, is a hard-core philological tome, and yet introduces cheerful cartoons of sheep dressed in Gaelic costume specifically to cheer the reader up.
Of course, the Middle Irish period which followed has been described as one of linguistic bedlam, with perfectly good Old Irish forms and near-Modern Irish forms all squabbling within the same text - a phenomenon which brings its own specialised difficulties. Clearly even the scribes usually didn't have a clue what was going on. But that is most definitely another tale.