Defeating the Böjningsmönster

A Discussion Paper on the Middle Cornish Verb Inflections

In order to write software to generate Cornish verbforms, it is necessary to decide on the rules of the game. To some extent any such 'rules' are artificially imposed by grammar writers — speakers work more by analogy from one pattern to another. Often two (or more) models are in competition in some particular situation, so there will always be grey areas in actual usage. What's more, the various patterns will usually lose or gain ground over time, so the pattern is dynamic. A good example is vowel affection in verb stems. This is largely retained in Literary Welsh and largely abandoned in spoken Welsh. In Breton it is absent from the Modern standard language, but present in Middle Breton, but far less reliably than in Middle Cornish. Middle (and so Revived) Cornish has retained vowel affection for the most part, probably because the regularly affected past participle is heavily used in Cornish, and so provides a reminder of which verbs change their stem vowels.

Standard languages are alway a little artificial as they endeavour to impose a relatively simple model on the variation found in actual usage. This is at least helpful to learners, who at a certain stage need to know that X is the 'correct' form. In attempting to write rules for Revived Cornish (i.e. Standard Middle Cornish c.1500) certain problems arise and have led to mistakes and inconsistencies in and between the standard works. These are :

1. Variation in actual MC usage;
2. Variation and obscurity in the written record (e.g. writing /I/ as <e> etc.);
3. Copying errors in the texts;
4. Paucity of examples of the lesser used forms;
5. Historical changes in some of the sounds involved;
6. Historical changes in the grammar itself (e.g. leveling).

We must therefore often be guided by analogy (and where possible comparison with W and B). We should I think aim for a middle ground between consistency and simplicity on the one hand, and faithfullness to the texts on the other, but without being over-influenced by the occasional attestation of a 'wrong' form (any more than we are in spelling for example). We have to accept that there will be a few 'special cases', especially in commonly used words, but be wary of multiplying them excessively without good cause.


The core of the MC verb consists of three stem-forms, 'Simple', 'Prefect' and 'Subjunctive', each of which can have two sets of endings, 'Primary' and 'Secondary' added. These then give rise to six 'tenses' (really a combination of tense, mood and aspect) as follows :

Simple + Primary => Present/Future;
Simple + Secondary => Imperfect;

Perfect + Primary => Preterite;
Perfect + Secondary => Pluperfect;

Subjunctive + Primary => Pres./Fut. Subjunctive;
Subjunctive + Secondary => Imperfective Subjunctive;

And perhaps in theory at least :

Subjunctive + Perfect + Secondary => Conditional;

Although the last except perhaps in the impersonal 'person' (0) is always formally identical with the Pluperfect.

The Simple stem is the stem of the verb without any addition, e.g. kan- 'sing';
The Perfect stem adds an -s- infix, e.g. kans- 'sang';
the Subjuctive stem adds an -h- infix which will usually combine with the final consonant of the stem to give a geminate, which in the case of b, d, g, (and j?) will be unvocied according to the laws of Brittonic phonology, to pp, tt, kk (and cch?). The detailed rules including those relating to clusters have been set out in the Agan Yeth paper, (although these changes are not always shown consistently in the texts).
Hence in this example kann- 'would sing'.

The combination of e.g. kan + h + s => kanns- => kans- means that the Conditional stem kan(n)s- 'were to sing' is identical to the Perfect.

Apart from a very few anomalous verbs, all the stem changes seem to be accounted for by phonological rules which are not special to verbs.

The Secondary endings are the same in all cases, as are the Primary endings in the plural, these are :

Primary Secondary
1s varies -en
2s varies -es
3s varies -e (» -a c.1475 when unstressed)
1p -'yn -en
2p -'owgh -ewgh
3p -+ons -ens
0 varies -ez (-'yz?)

Endings marked (') cause vowel affection of mutable stems. (+) means that affection takes place in the pres/fut. in sympathy with the 3s only.

The plural Secondary endings are the same as the corresponding Primary ones except that the vowel is changed to the characteristic secondary e. There is no distinction of number in the 1st person Secondary endings, -en answers to both 'I' and 'we'.

Cornish regular verbs might in the first instance be classified into two main groups, A-verbs having the endings given above, and I-verbs. The latter use a different set of (apparently i-affected) Secondary endings, but ONLY FOR THE IMPERFECT (i.e. with the SIMPLE stem) NEVER in the pluperfect/conditional or imperfect subjunctive :

1s -'yn
2s -'ys
3s -'i
1p -'yn
2p -'ewgh
3p -'ens
0 -'yz

One might have expected 3s y which since it was almost always unstressed would have developed into -e by the time of the earliest texts, (and later to -a) and so been indistinguishable from the A-verbs. Since it is attested as <y> it must have been reinterpreted as i (/i/ not /I/) to maintain the distinction between the two sets of endings, perhaps to be consistent with the i-affected stem vowel. Note that all these endings pass vowel affection forward to the stem vowel, even the 2/3p endings which still have -e-.

We now come to the Primary endings in the singular, which are the most used and which are different for each stem/tense (suppletion) :

Simple Stem Perfect Stem Subjunctive Stem
Pres/Fut. Preterite Pres/Fut. Subj.
1s -av -'y[s] -''iv
2s -'ydh -'ys -''i
3s +(e) -a[s] -o

In the Preterite the [s]'s are the -s of the perfect stem. That is in the third person singular, e.g. kans- does not add an ending as such but inserts the vowel -a- to give kanas 'he/she/it sang'. The first person is formed in the same way but has vowel affection (this is also seen in some 'irregular' perfects, e.g. eth 'he went' » yth 'I went') which is passed forward into the stem in the usual way, so kenys 'I sang'. Because this ending is the result of i-affection I have written it with a y (cf. Welsh -ais, MW -eis, the distinctive product of final i-affection) notwithstanding Breton -is. In contrast the 1s subjunctive, MB -iff, W -wyf has been written as -iv [for now, this needs looking into].

All the endings marked (') cause i-affection of the stem vowel. The subjunctive endings marked ('') usually give rise to enhanced affection of a, o to y (in mutable stems).

Certain verbs differ from the above scheme in one or more of the following ways :

1. The 3s pres/fut is usually the bare stem, e.g. kan 'sings'. It may however undergo i-affection (a,o > e; e > y) or enhanced affection (a, o > y), e.g. koll- 'lose' » kyll 'loses'.

2. In the 3s pres/fut half-a-dozen or less stems add a final -e, (post 1500 and Revived Cornish -a), e.g. hwile/a 'seeks', gorte/a 'waits'. [Note : Are these always verbs with vbl.n. in -az or -os?]

3. Quite a few verbs form their 3s preterite in -'is (almost like the 1s), and this ending unlike -as affects the stem vowel.

4. In the pres/fut. subjunctive some verbs with mutable stems in a, o have only normal affection to e rather than enhanced affection to y. Whichever type occurs in the 1/2s is copied in 1/2p, e.g. for kar-: kyrriv, kyrri, karro, kyrryn, kyrrowgh, karrons; but for prov- (at least according to Smith's CS) preffiv, preffi, proffo, preffyn, preffowgh, proffons.

Note : There seems to have been a tendency in Cornish for /E/, whatever its origin, to be raised to /I/ when stressed before a consonant cluster beginning with a resonant (liquid, nasal, semivowel?). This change may have been sporadic or regular subject to rules we've still to map out. This explains kyns (not *kens), syns 'saints' (not *sens, « sans), and perhaps verbforms like sylwyz, gylwir 'is called' etc. If so this may be how 'enhanced affection' arose. The extent to which it was generalised and/or grammaticalised need looking into.

The impersonal endings seem to have been as follows :

pres/fut preterite pres/fut. subj.
-'ir -az/-'iz -er

These seem fairly certain and have clear analogues in MW and MB. In MW there are a number of different 3s pret. endings including -as and -is along with others not in Cornish. In each case the pret. impersonal ending has the vowel of the 3s but replaces the final -s with a /-d/ (generally written <-t>)[WG (s)175(6)iv(1) p.327]. This is equivalent to Cornish -z. I would suggest therefore that the vowel in this ending follows that of the 3s pret. and the final <s> be written <z> in the impersonal if <z>'s are being used, to distinguish these two forms. (The difference between unstressed final /s/ and /z/ was sufficiently neutralised in MC for them to be rhymed together, even in the earliest texts. However this distinction seems to have remained for the most part in stressed syllables. Maybe these sounds could still be distinguished in unstressed finals in careful or archaic speech.)

imperfect pluperfect imper. subj,
-ez/-'yz -'syz -ez

There is much confusion between the various 'authorities' which probably reflects confusion in the texts over these little-used forms. Cognates show that all end in /z/. The original form seems to have been -ez in line with the -e- elsewhere in the imperfect endings, so that -yz might be expected only in the imperfect of I-verbs which have -y- elsewhere in this tense. However -yz is also the past participle ending and confusion between the two is probably why this ending largely dropped out of use. For example "Y hanow a ylwyz Yowan" 'his name was called John' would be reinterpreted with a past ptcpl and the 'missing' auxilary inserted, "Y hanow a *veu* gylwyz Yowan." That may have favoured the -ez imperfect ending over -yz even in I-verbs where it belonged. On the other hand the -ez ending risked confusion with the 2s imperfect -es in A-verbs, once the /s ~ z/ distinction began to be lost in unstressed syllables.

In the pluperfect/conditional however the -s- added to stem prevented any confusion with the past participle, so generalising -s-yz would be advantageous in A-verbs, (to distinguish it from 2s -s-es) and indeed the resemblence to the past. ptcpl. may well have helped to fix the -y- form here.

Unless therefore there is overwhealming contrary textual evidence, I would suggest the above scheme as a rationalisation for the standard language. Imperfect: -ez in A-verbs and -yz in I-verbs, that is the vowel follows that of the other imperfect endings. Pluperfect: -syz everywhere (analogy with past participle). Imperfect Subjunctive: -ez everywhere (probably the original ending).

The endings in y and i marked (') above will give rise to i-affection of the stem vowel(s).

We can now categorise individual regular verbs according to a number of categories. Until I've analysed Tony Hak's database I can't say if all these are independent variables, or if any can be predicted by the phonemic shape of the root. The categories are :

1. The form of the verbal noun. The different endings are often good (but not infallible) predictors of how the verb will conjugate.

2. The stem as seen in the 1s pres/fut. This shows us the basic stem vowel which might have been affected in the verbal noun. Or the verbal noun might in some cases be irregular.

3. Whether the stem is subject to i-affection or to enhanced i-affection. This can be seen by looking at the 2s pres/fut. for example.

4. The form of the 3s pres/fut. If affected the 3p pres/fut. will also change. However if the 3s has enhanced affection the 3p will only show normal affection, unless enhanced affection is general for that verb.

5. Whether an A-verb or an I-verb, as shown by e.g. the 1s imperfect, -en or -'yn. In a I-verb the whole of the imperfect shows i-affection, including 2p and 3p regardless of their endings.

6. The form of the 3s preterite, -as or -'is.

7. Whether the 1/2 s/p of the pres/fut. subjunctive show affection, enhanced affection, or none.

Unless we can find further redundancy in these rules, we would need to give SEVEN 'principle parts' in a dictionary entry in order to fully characterise even a *regular* verb! Here are a few examples adapted from CS with the significant (unpredictable from previous columns) features capitalised.

vbl.n. Pres/Fut. Imperf. Pret. Pres/Fut.Subj.
1s 2s 3s 1s/p 3s 1s
kara karav kerydh kar karen karas kyrriv
gwelez gwelav gwelydh gwel gwelYn gwelas gwylliv
redya redyav redydh red redyen redyas rEttiv
fia fiav fiydh fi fien fias fiiv
leverel lavarav leverydh lEvEr leverYn leverIs lEvErriv
gelwel galwav gelwydh gElow gelwYn gelwIs gylwiv
previ provav prevydh prEv prevYn provas prEffiv
glanhe glanh[a]av glanhydh glanha glanhYn glanhas glanhahiv
dyski dyskav dyskydh dysk dyskYn dyskas dyskiv
klywez klywav klywydh klyw klywen klywas klywFFiv*
kavoez kavav kevydh kYv kevYn kavas kyffiv
dybri debrav dybrydh deber dybrYn dybrIS dyppriv
pedri podrav pedrydh pEder pedrYn podras pEttriv

*{pt. of boz, also klywo}

I suspect that the 'enhanced affection' in gylwiv is caused by the -lw- group and so if the texts are examined would probably be found all through this verb except for gelow and the imperative galow where the added vowel breaks up the group.

One might expect that I-verbs would always have an affected 3s pres/fut. but then why previ, provav, prev and pedri, podrav, peder, BUT dybri, debrav, deber (NOT *dyber??). Maybe there's a phonological rule relating to the added vowel, some kind of harmony?

Hag hemm yw lowr rag a'n eur-ma my a dyb!

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License