|The Verbs||Present Stem||Perfect Stem|
|sum `be'||su- 1||fu-|
|volo: `be willing'||volu-||volu-|
|fero: `bear, carry'||feri-||tul-|
1. It has no passive.
2. In the Present Indicative,
a. u is deleted before nonnasal suffixes (as stated in footnote 1)
b. sum takes the -m (rather than the o:) suffixes.
c. a sandhi rule is required for 2S: ss# > s#
3. In the Present Subjunctive,
a. VOWEL SWITCH changes u to i:
b. in 3P, i: does not change to iu before -nt (as it does in the Indicative) but simply shortens to i
4. In the Imperfect and the Future,
a. u is deleted before the ba:, re:, and bi increments (all of which begin with nonnasals)
b. additional sandhi rules are needed:
1) sb > r (esba: > era:, esbi > eri)
2) sr > ss (esre: > esse:)
5. In the Future, sum behaves like a:-stem's and e:-stem's by adding -bi and taking the -o: suffixes 2
1. It has no passive.
2. In the Present Indicative,
a. short u (like short i) drops before the -o: 1S suffix
b. as with sum above, u is dropped before the nonnasal suffixes
c. in 2S, the expected volus is instead vi:s
d. there is variation between volt and vult in 3S and 2P
3. In the Subjunctive, the present stem volu- becomes velu-
a. In the Present Subjunctive,
1) VOWEL SWITCH changes u to i: (as with sum above)
2) this i: is not changed to iu before -nt (as with sum above)
b. In the Imperfect Subjunctive,
1) u is dropped before the re: increment
2) the re: increment undergoes a sandhi rule: lr > ll (velre: > velle:)
4. In the Imperfect Indicative, u is not deleted before ba: (as would be expected from the general statements made above [as in 2.b.] about its deletion before nonnasals) but is instead changed to e: just as the i of i-stems is
5. In the Future,
a. volo: behaves like i-stems in that the stem vowel u changes to e: and is followed by the -m suffixes 3
1. The i of this i-stem deletes before r, s, and t; thus, rir > rr, ris > rs, and rit > rt, in the following:
a. Present Indicative Active: 2S, 3S, 2P
b. Present Indicative Passive: 2S, 3S
c. Imperfect Subjunctive, both Active and Passive (ferire: > ferre:)
1. The VOWEL SWITCH rule (which says nothing about short u) can be appended to have it switch to i:. This will remove Irregularity 3.a. under sum and Irregularity 3.a.1) under volo:
2. The rule for the -nt suffix which changes a preceding i to u and i: to iu should be restricted to the Indicative. Otherwise, it would contravene the VOWEL SWITCH rule as appended above. This will remove Irregularity 3.b. under sum and Irregularity 3.a.2) under volo:.
3. The dropping of u from u-stems can be added to the effects of the nonnasal suffixes and one increment: -s (2S), -t (3S), -tis (2P), re: (Imperf. Subj.). This will remove Irregularities 2.a. and 4.a. under sum, and Irregularities 2.a., 2.b., and 3.b.1) under volo:.
4. Some of the sandhi rules operating with these verbs may apply more widely and, if so, could be included in a more general set.
The Perfect has no irregularities for any of these three verbs. The Subjunctive has fewer irregularities than the Indicative in the Present; in the Imperfect it is hard to tell, since it depends on how regular or irregular the various sandhi rules we propose turn out to be. The Imperfect shows fewer irregularities in comparison with the Present in both Indicative and Subjunctive. The Passive of fero: has fewer than the Active. Assuming that the Perfect, Passive, Subjunctive, and Imperfect all are more marked than their opposites, we can say that the latter, the less marked, show greater variation in form. This has been termed the Principle of Deviance in markedness theory.
Alternatively we could say: The paradigm that has a zero increment 4 and which we would therefore expect to be least marked semantically as well formally (according to the Principle of Zero Expression) is also the paradigm that shows greatest irregularity in these verbs. This exemplifies the Principle of Deviance.
1. The irregularities of fero: seem at first glance to be more phonological than those of the other two verbs. However, the places in which the i of its i-stem drops are exactly the same as those in which the u of volo: drops. Yet this latter more morphological approach is ruled out by the need to account for the regular behavior of our short-i exemplar tegi-. For example, with respect to tegi- we want to have the i stem-vowel remain before the -s, -t, and -tis suffixes. We cannot at the same time say that an i stem vowel drops before these suffixes to account for the i of feri-. Nor can we blame it on the preceding r; there are other i-stem verbs with a preceding r which do not drop the i in these places: for example, verro: (present stem verri-) 'sweep', sero: (seri-) 'sow', tero: (teri-) 'rub'. We must conclude that the irregularities of fero: can be attributed neither to the suffixes nor to the r of the stem. They must be attributed to fero: as lexical item. This is an example of what some have called "morpholexical" variation.
2. Although the determination that volo: is an u-stem seems fairly straightforward, the case for sum being one too is less obvious. Yet the advantages of setting up both as u-stems are several. By doing so we are able to bring out parallelisms in their dropping of the u as well as in their vowel switch behavior.
1. The u of this form is deleted before suffixes that do not begin with a nasal. When the u is deleted, thereby creating a consonant cluster, a prothetic e precedes the cluster. (This is equivalent to saying that the present stem is su- before nasal suffixes (and before VOWEL SWITCH), es- elsewhere.)
2. This is not necessarily an irregularity, since we have not encountered u-stems before.
3. Since volo: as an u-stem behaves differently than does sum in the Future, both cannot be regular; at most, only one can be.
4. That is, the increment that stands between Latin verb stems and one of the four suffix sets.